Category: hydroponic agriculture

Posted on

It is produced primarily under aerobic conditions but also can be generated in the absence of oxygen

Global models that have incorporated both biophysical and socio-economic parameters have predicted that negative impacts on food production from climate change will largely be felt in the developing world, but positive impacts will be felt in the developed world . These studies conclude that the magnitude of this disparity will be determined by which future IPCC’s emissions scenario is adopted and the degree to which crops will respond to CO2 fertilization. Low latitude regions of the world may not benefit from CO2 fertilization, because the benefits are overshadowed by the predicted detrimental effects of increased temperature and/or precipitation changes . As a result, regions such as Africa or parts of Asia are predicted under the GHG intensive A1fi scenario to experience yield reductions up to 30% of 1990 levels by 2080 . The population at risk of hunger in Mali, for example, is predicted to increase from the current 34% to 44%, due to land degradation and then up to 72% due to the additional impacts of climate change by 2030 . These regions are at particular risk because their lack of infrastructure and technology impedes their producers’ ability to adapt to adverse and/or altered climate conditions. In contrast, the stress caused by A1fi climatic conditions is expected to be offset for some crops such as cereals , by the effects of CO2 fertilization, resulting in small increases in yield in Australia, North America, and South East South America.Assessments of agricultural production in the United States have used an Integrated Assessment approach, which includes complex interactions of temperature and precipitation changes with increased climate variation, changes in pesticide use, environmental effects caused by agriculture , changing global markets, societal responses, and technological adaptation, to model agricultural response to climate change . Consistent among these studies are the conclusions that there will be a dramatic difference in regional impacts,blueberry packaging box but agricultural production in the United States overall will increase, commodity prices will fall and irrigation use will go down due to increased precipitation and potentially higher water use efficiency that results from CO2 fertilization .

Climate change is therefore expected to be economically positive for U.S. consumers and negative for producers, but will entail increased pesticide use and result in increased environmental degradation . Regional-level forecasts could be quite different in California than nationally, due to its limited water resources and its focus on specialty crops.Water supply is central to the success of Californian agriculture. In addition to changes in precipitation, water availability will likely be influenced by rising temperatures, and consequential increases in water demand from other sectors . Increased temperatures will affect the amount of water collected and stored in the Sierra snow pack. By the end of the century, the Sierra snow pack is predicted to be 30% to 70% lower than the current winter total, due to an increase in rainfall vs. snowfall, and earlier melting of the snow pack . This will be most prominent in the southern Sierra Nevada, and at elevations below 3,000 m where 80% of California’s snow pack storage currently occurs. The changing availability of water both within California and to California agriculture, may lead to heavy reliance on groundwater resources, which are currently over drafted in many agricultural areas . Approximately 42% of current ground and surface water is used for agricultural purposes . Demand for water resources will be further exacerbated by an increase in the population of California in the coming century, which is projected to be > 46 million people by 2030, and may reach 90 million by 2100 . As will be discussed below, gradual shifts in climate over the next hundred years will necessitate adaptations that may not necessarily require direct government intervention, and could be driven, largely by market forces, changing management practices, and technological advances . California agricultural producers have had a history of adapting to new locations, development of water resources, and changes in markets. New adaptations will be made easier and more efficient by the availability of predictive information to producers, and an appropriate policy environment. Some sectors also lend themselves to more rapid change than others. For example, perennial tree crops and vines, of which many are unique to California in the U.S. context, may be particularly vulnerable to problems.

The adaptation to rapid change or extreme climatic events, such as floods, droughts, and heat waves are much more difficult to predict. Such extreme events may exceed the adaptive capacity of markets and be much more difficult for producers to cope with . Thus, development of risk and response strategies to various extreme climate change scenarios may gain more attention in the coming years. Beyond responding to changes in climate, California producers will most likely find opportunities to mitigate the release of GHG. Agriculture will play a significant role in a portfolio of national mitigation strategies, for example, as a first step to sequestering carbon . United States agriculture and forestry could remove more than 425 million metric tons of carbon equivalents of combined greenhouse gases , based on modeling of extreme increases in carbon prices. Carbon trading could have substantial impacts for agriculture, such as increased crop value and reduction of environmental externalities. Greenhouse gases include carbon dioxide , nitrous oxide , methane , and high global-warming-potential gases such as sulfur hexafluoride , hydrofluorocarbons , and chlorofluorocarbon . Since these gases absorb the terrestrial radiation leaving the earth’s surface, changes in the atmospheric concentrations of these gases can alter the balance of energy transfer between atmosphere, land, and oceans. All atmospheric GHG concentrations are increasing each year due to anthropogenic activity, which, in turn, leads to climate changes at the local, regional and global scale . The present section focuses in particular on CO2, N2O, and CH4 because they are the three major bio-genic GHGs produced by the agricultural sector in California and across the globe. It summarizes current sources and sinks of GHGs, i.e., total amounts of emissions by each type of gas, the contribution of the agricultural sector to California GHGs emissions, and consider agriculture and forests as potential sinks of GHGs. Potential impacts of climate change on CO2, N2O, and CH4 emissions and possible mitigation strategies for GHGs produced by the agricultural and other sectors are presented. California produced 493 million metric tons of CO2-equivalent GHGs emissions and in 2002 was ranked as the second largest U.S. state emitter after Texas . Most emissions were CO2 produced from the combustion of fossil fuels from industrial and transportation sources. Overall, the contribution of the agricultural sector to GHG emissions as a whole in California is relatively small. Taken together, agriculture and forestry contributed approximately 8% of the state’s total GHG emissions, including GHGs from all agriculturally related activities such as fossil fuel combustion associated with crop production, livestock production,blueberry packaging containers and soil liming . Emissions arising from transportation of agricultural commodities are not included in this estimate. CO2 emissions from non-fossil fuels, including agricultural activities, were 2.3% of the total GHG emissions of California. Of the 2.3% of CO2 from non-fossil fuels, agricultural activities contributed about 38%. Thus, the total contribution of CO2 from agricultural activities to the total GHG emissions was 0.9% in 2002 .

Nitrous oxide and CH4 emissions contributed 6.8% and 6.4%, respectively, to the total GHG emissions in 2002, with approximately 59% and 38%, respectively, originating from agricultural activities. An estimated 18.6 and 0.9 metric tons of CO2-equivalent GHG came from agricultural practices and manure management, respectively, in 2002 . Thus, the agricultural contribution to the state’s 2002 emissions of N2O and CH4 was 4% and 3%, respectively. Methane emissions from California flooded rice fields constituted a total of 0.5 metric tons of CO2-equivalent GHG and constituted less than 2% of total CH4 emission in California in 2002 . Methane emissions from animal production included 7.3 and 6.6 metric tons of CO2-equivalent from enteric fermentation and manure, respectively .Greenhouse gases are produced primarily by soil microorganisms carrying out oxidation-reduction reactions, including nitrification, denitrification, methanogenesis and organic matter decomposition . Because changes in temperature and precipitation alter the activity of soil microorganisms, GHG emissions from agriculture would likely be affected by climate change. This section considers, in general terms, the impacts of climate change on respiration and soil organic matter dynamics, as well as N2O and CH4 emissions. Carbon dioxide is the end product of respiration by soil biota .A potential impact of increased temperature is loss of carbon from the large reservoir of C contained in SOM in agricultural and forest soils. Although forests are currently a sink for CO2 , they might become a source of CO2 with temperature increases from global warming . This issue is critical, because SOM contains roughly two-thirds of the terrestrial C and two to three times as much C as atmospheric CO2 . Many researchers have investigated the effects of temperature on decomposition rates of SOM in mineral soils. Trumbore et al. reported that temperature is a major controller of turnover for a large component of SOM, as long as soil moisture is not a limiting factor. It is hypothesized that the decomposition of soil labile C is sensitive to temperature variation whereas resistant components are less sensitive. However, Fang et al. suggested that the temperature sensitivity for resistant SOM pools does not differ significantly from that of labile pools, and that both pools of SOM will therefore respond similarly to global warming. If this conclusion is correct, more C will be released than predicted by the HadCM3 model that assumes an insensitivity of resistant C pools to temperature. In contrast to observations that decomposition is enhanced by increases in temperature, Giardina and Ryan reported, based on analyses of data from locations across the world, that rates of SOM decomposition in mineral soils were not controlled by temperature limitation to microbial activity and that estimates made from short term studies may overestimate temperature sensitivity. Because moisture content in soils strongly affects the activities of soil microorganisms in direct, and perhaps more importantly in indirect ways , changes in precipitation patterns due to global warming may be one of the main impacts of climate change on SOM decomposition in mineral soils. Predictions of changes in precipitation are problematic, differing among climate models, thus making influences on SOM difficult to predict. Other factors influence SOM decomposition , some of which may be affected by climate change, must also be considered in projections of how SOM will behave. Temperature should not be viewed in isolation from other factors . Unfortunately, the magnitude and relative importance of these factors in governing SOM dynamics have received little attention in the literature.Nitrous oxide is produced primarily during denitrification, an anaerobic microbial process in soils or sediments, in which nitrate is used as an electron acceptor in the absence of oxygen . Though nitrification, the oxidation of ammonium to nitrate, also produces some N2O, this process is thought to be less important than denitrification . Agricultural activities—soil emissions from fertilizer use, residue burning and animal production—are responsible for an estimated 80% of anthropogenic emissions of N2O . Few studies have investigated N2O emissions in agroecosystems in California. In a comparison of organic and conventional managed tomato soils in the Central Valley, N2O emissions were found to be of short duration; followed addition of organic or mineral fertilizer in the organic and conventional systems, respectively; and occurred immediately after irrigation events . There are, however, no published extensive, systematic studies collecting field measurements of N2O over the growing season in different soil types of California to permit identification of relationships between fluxes, management practices, and environmental variables. Other studies outside of California have indicated that emissions of N2O are primarily controlled by soil moisture content, in particular the water-filled pore space , temperature , organic carbon availability , and concentration of mineral nitrogen . The latter factor is often optimal in agricultural soils for N2O fluxes because addition of synthetic N fertilizers and organic manures lead to elevated mineral N concentrations at least temporarily. In addition, California agricultural systems are frequently irrigated, leading to ideal moisture conditions for denitrification and potentially N2O fluxes .

Posted on

The estimated productivity gaps in GLW are an order of magnitude larger than our estimates

The shift out of agriculture and into other more “modern” sectors has long been viewed as central to economic development. This structural transformation was a focus of influential early scholarship with the issue even stretching back to Soviet debates over whether to “squeeze” farmer surplus to hasten industrialization . A more recent macroeconomic empirical literature has revived interest in these issues, often using data from national accounts . This body of work has documented several important patterns that help shed light on the sources of income differences across countries. First, it shows that the share of labor in the agricultural sector correlates strongly with levels of per capita income: most workers in the poorest countries work in agriculture while only a small share do in wealthy countries. Importantly, while income per worker is only moderately larger for non-agricultural workers in wealthy countries relative to poor countries, agricultural workers are many times more productive in rich countries. This creates a double disadvantage for poor countries: agricultural work tends to be far less productive in low-income countries, yet the workforce is concentrated in this sector.Studies that explore the closely related gap between the urban and rural sectors reach similar conclusions. Several recent studies have examined the extent to which these productivity gaps across sectors can reasonably be viewed as causal impacts rather than mainly reflecting worker selection. By a causal impact of sector,wholesale grow bags we mean that a given worker employed in the non-agricultural sector is more productive than the same worker employed in the agricultural sector. In contrast, worker selection would reflect differences driven by the fact that workers of varying ability and skill levels are concentrated in particular sectors.

This paper seeks to disentangle these two competing explanations by estimating sectoral wage gaps using unusually long-run individual-level panel data from two low-income countries, Indonesia and Kenya. If there are causal impacts of sector, the large share of the workforce employed in the agricultural sector in low-income countries could be viewed as a form of input misallocation along the lines of Hsieh and Klenow and Restuccia and Rogerson . The resolution of this econometric identification issue, namely, distinguishing causal effects from selection, is not solely of scholarly interest: the existence of causal sectoral productivity gaps would imply that the movement of population out of rural agricultural jobs and into other sectors could durably raise living standards in low-income countries, narrowing cross-country differences. The existence of large causal sectoral productivity gaps also raises questions about the nature of the frictions that limit individual movement into more productive employment, and the public policies that might promote such moves or hinder them . Gollin, Lagakos, and Waugh and Young are two important recent studies that explore this identification issue. GLW examine labor productivity gaps in nonagricultural employment versus agriculture using a combination of national accounts and repeated cross-sectional data from micro-surveys, and document a roughly three-fold average productivity gap across sectors. In their main contribution, GLW show that accounting for differences in hours worked and average worker schooling attainment across sectors—thus partially addressing worker selection— reduces the average estimated agricultural productivity gap by a third, from roughly 3 to 2. They also find that agricultural productivity gaps and per capita consumption gaps based on household data remain large but tend to be somewhat smaller than those estimated using national labor surveys, possibly in part due to differences in how each source measures economic activity. GLW remain agnostic regarding the causal interpretation of the large agricultural productivity gaps that they estimate. If individual schooling captures the most important dimensions of worker skill and thus largely addresses selection, GLW’s estimates would imply that the causal impact of moving workers from agriculture to the non-agricultural sector in low-income countries would be to roughly double productivity, a large effect.

Of course, to the extent that educational attainment alone fails to capture all aspects of individual human capital, controlling for it would not fully account for selection. Young examines the related question of urban-rural differences in consumption , rather than productivity, and similarly finds large cross-sectional gaps.Using Demographic and Health Surveys that have retrospective information on individual birth district, Young shows that rural-born individuals with more years of schooling than average in their sector are more likely to move to urban areas, while urban-born individuals with less schooling tend to move to rural areas. Young makes sense of this pattern through a model which assumes that there is more demand for skilled labor in urban areas, shows that this could generate two-way flows of the kind he documents, and argues that he can fully explain urban-rural consumption gaps once he accounts for sorting by education.3 The current study directly examines the issue of whether measured productivity gaps are causal or mainly driven by selection using long-term individual-level longitudinal data on worker productivity. Use of this data allows us to account for individual fixed effects, capturing all time invariant dimensions of worker heterogeneity, not just educational attainment . We focus on two country cases – Indonesia and Kenya – that have long-term panel micro data sets with relatively large sample sizes, rich measures of earnings in both the formal and informal sector, and high rates of respondent tracking over time. The datasets, the Indonesia Family Life Survey and Kenya Life Panel Survey , are described in greater detail below.4 For both countries, we start by characterizing the nature of selective migration between non-agricultural versus agricultural economic sectors, and between urban versus rural residence. Like Young , we show that individuals born in rural areas who attain more schooling are significantly more likely to migrate to urban areas and are also more likely to hold non-agricultural employment, while those born in urban areas with less schooling are more likely to move to rural areas and into agriculture.

We exploit the unusual richness of our data, in particular, the existence of measures of cognitive ability , to show that those of higher ability in both Indonesia and Kenya are far more likely to move into urban and non-agricultural sectors, even conditional on educational attainment. This is a strong indication that conditioning on completed schooling is insufficient to fully capture differences in average worker skill levels across sectors. We next estimate sectoral productivity differences, and show that treating the data as a repeated cross-section generates large estimated sectoral productivity gaps, echoing the results in existing work. In our main finding, we show that the inclusion of individual fixed effects reduces estimated sectoral productivity gaps by over 80 percent. This pattern is consistent with the bulk of the measured productivity gaps between sectors being driven by worker selection rather than causal impacts. Specifically, we first reproduce the differences documented by GLW for Indonesia and Kenya, presenting both the unconditional gaps as well as adjusted gaps that account for worker labor hours and education . These are large for both countries,grow bags for gardening with raw gaps of around 130 log points, implying roughly a doubling of productivity in the non-agricultural sector. When we treat our data as a series of repeated cross-sections, the gaps remain large, at 60 to 80 log points. These are somewhat smaller than GLW’s main estimates, though recall that GLW’s estimates using household survey data also tend to be smaller. Conditioning on individual demographic characteristics as well as hours worked and educational attainment narrows the gap, but it remains large at between 30 and 60 log points. Finally, including individual fixed effects reduces the agricultural productivity gap in wages to 4.7 log points in Indonesia and to 13.4 log points in Kenya, and neither effect is statistically significant. Analogous estimates show that productivity gaps between urban and rural areas are also reduced substantially, to zero in Indonesia and 13.2 log points in Kenya. We obtain similar results for the gap in per capita consumption levels across sectors where this is available for Indonesia. This is useful since consumption measures may better capture living standards in less developed economies than earnings measures, given widespread informal economic activity. Furthermore, we show that the productivity gap is not simply a short-run effect by demonstrating that gaps do not emerge even up to five years after an individual moves to an urban area. We also find that productivity gaps are no larger even when considering only moves to the largest cities in Indonesia and Kenya .

Our methodological approach is related to Hendricks and Schoellman , who use panel data on the earnings of international migrants to the United States, including on their home country earnings. Mirroring our main results, the inclusion of individual fixed effects in their case greatly reduces the return to international migration . Similarly, McKenzie et al. show that cross-sectional estimates of the returns to international immigration exceed those using individual panel data or those derived from a randomized lottery. Bryan et al. estimate positive gains in consumption in the sending households of individuals randomly induced to migrate within Bangladesh, although no significant gains in total earnings. Bazzi et al. argue that cross-sectional estimates of productivity differences across rural areas within Indonesia are likely to overstate estimates derived from panel data using movers. Other related studies on the nature of selective migration include Chiquiar and Hanson , Yang , Beegle et al. , Kleemans , and Rubalcava et al , among others. A limitation of the current study is that we focus on two countries, in contrast to the scores of countries in GLW and Young . This is due to the relative scarcity of long-run individual panel data sets in low-income countries that contain the rich measures necessary for our analysis. That said, the finding of broadly similar patterns in both countries, each with large populations in two different world regions, suggests some generalizability. Another important issue relates to the local nature of our estimates, namely, the fact that the fixed effects estimates are derived from movers, those with productivity observations in both the non-agricultural and agricultural sectors. It is possible that productivity gains could be different among non-movers, an issue we discuss in Section 2 below. There we argue that, to the extent that typical Roy model conditions hold and those with the largest net benefits are more likely to move, selection will most likely produce an upward bias, leading our estimates to be upper bounds on the true causal impact of moving between sectors. However, absent additional knowledge about the correlation between individual preferences, credit constraints, and unobserved productivity shocks, it is in principle possible that selection could bias our estimates downward instead. Similarly, it is possible that very long-run and even inter-generational “exposure” to a sector could persistently change individual productivity due to skill acquisition, and this opens up the possibility that selection and causal impacts are both important. We return to these important issues of interpretation in the conclusion, including ways to reconcile our estimates with existing empirical findings. The paper is organized as follows. Section 2 presents a conceptual framework for estimating sectoral productivity gaps, and relates it to the core econometric issue of disentangling causal impacts from worker selection. Section 3 describes the two datasets ; characterizes the distinctions between the non-agricultural and agricultural sectors, and urban vs. rural areas; and presents evidence on individual selection between sectors. Section 4 contains the main empirical results on productivity gaps, as well as the dispersion of labor productivity across individuals by sector, consumption gaps, dynamic effects up to five years after migration, and effects in big cities versus other urban areas. The final section presents alternative interpretations of the results, and concludes. We present a development accounting framework to disentangle explanations for the aggregate productivity gap across sectors. We consider both observable and unobservable components of human capital, and whether intrinsic worker preferences for sector may bias direct measurement of the productivity gap. A standard model suggests that worker selection is most likely to bias sectoral productivity gaps upward when estimated among those moving into non-agriculture but lead to a downward bias when estimated among those moving into agriculture.

Posted on

Exposures to pesticides in the third trimester did not increase risk for preterm birth

Each PUR record includes the name of the pesticide’s active ingredient, the poundage applied, the crop type, and the location and the date of application. The California Department of Water Resources performs countywide, large-scale surveys of land use and crop cover every 7–10 years. Land use maps increase spatial resolution because they provide more detailed land use geography that allows us to refine the pesticide applications . We then combined PUR records, land use maps, and geocoded birth addresses to produce estimates of pesticide exposure during pregnancy. Monthly exposure estimates were calculated by adding the poundage of pesticide applied in a 2-kilometer buffer surrounding each address and weighting the total poundage by the proportion of acreage treated within the buffer. Previous pesticide studies relied on different buffer sizes from 500m , half a mile , 1000m , 1250m , 1600m , 5000m , to up to 8000m distances , depending on the pesticide of interest, landscape, and weather conditions. In light of previous research, the buffer of 2-km we chose, will provide a reasonable distance for assessing pesticide applications around residential addresses. For each calendar month, our integrated GIS-system returned continuous measures for each specific chemical applied within 2-km of individuals’ residences. We defined the first, second, and third trimesters as 0-12 weeks, 13-25 weeks, and ≥26 weeks of pregnancy, respectively. For preterm birth, the length of gestation and hence exposure period are shorter than term birth by design; to account for that, we assessed the third trimester exposures using 27- 32 weeks of gestation only since more than 88% of all preterm births had a gestational length longer than 32 weeks. For each pesticide, daily poundage for each gestational day of pregnancy was calculated based on monthly values,seedling starter pot and then averaged across all days in each trimester. We then categorized prenatal exposure as ever/never exposed to a specific chemical in each trimester.

We selected 17 individual chemicals previously observed to have reproductive toxicity . Additionally, we also considered all pesticides from three widely used chemical classes that have been linked to reproductive toxicity based on the Pesticide Action Network pesticide database , i.e. 24 n-methyl carbamate/dithiocarbamates, 50 organophosphates, and 29 pyrethroid pesticides to which one or more study subjects were exposed according to our 2km buffer criterion . For each class, we used the sum of the total number of individual chemicals that each subject was ever exposed to in each time period of interest. We divided subjects into high , low , and no exposure to the respective pesticide, and compared high and low with the no exposure group as the reference. Since information about the specific location of non-agricultural pesticide applications are not provided by the PUR and because some individuals in urban areas are highly exposed to traffic-related air pollution or hazardous air toxics that are known risk factors for adverse birth outcomes , we restricted our analyses to individuals born in agricultural regions, defining those as residences within 2km buffer of any type of agricultural pesticide application during pregnancy . We conducted unconditional logistic regression analyses adjusting for matching factors and the source of control subjects and estimated odds ratios and 95% confidence intervals . To account for the unbalanced gender ratio and birth year distribution in this combined sample, we included the inverse of the sampling fraction as a stabilized weighting factor to reflect the sex and birth year distribution of all California births. Statistical analyses were performed using SAS software, Version 9.4 . We additionally adjusted for covariates as potential confounders and effect measure modifiers based on the literature : including maternal age at delivery , maternal race/ethnicity , maternal birthplace , maternal education , parity , payment source for prenatal care as a proxy for family income , prenatal care in the first trimester , and a previously developed neighborhood-level SES metric . Furthermore, we conducted stratified analyses by maternal race/ethnicity since exposures may be higher among Hispanics, especially recent immigrants, who may live close to agricultural fields and have poor housing conditions ; by infant sex because males are more likely to be born preterm ; as well as by season of conception , estimated from the last menstrual period and length of gestation, because of seasonal variations in pesticide applications .

In sensitivity analyses, we compared effect estimates with and without adjusting for two risk factors for adverse birth outcomes, maternal cigarette smoking during pregnancy and prepregnancy Body Mass Index , calculated as maternal pre-pregnancy weight divided by maternal height  for births in 2007-2010 only, since these variables are only available on the birth certificate from 2007 onward. We also investigated the potential confounding effects from outdoor air pollution that can impact fetal growth during critical periods among the autism controls only due to data availability. We estimated trimester-specific exposures to local, traffic-derived NOx, PM2.5, and CO, including roadways within 1.5 km of subjects’ birth addresses, i.e. inter-quartile range -scaled measure of NOx as a local traffic marker derived from the CAlifornia LINE source dispersion model model . Additionally, we adjusted for co-exposure to at least one of other individual chemicals as a single variable when assessing each individual chemical, and estimated mutually adjusted ORs for the three chemical class exposures during the same exposure window. When evaluating later trimester exposures we adjusted for exposure during prior pregnancy periods, because these effect estimates may be altered by earlier exposures . Since a low geocode quality is likely to introduce spatial exposure misclassification, we excluded those with a geocode quality at the USPS Zip Code Area centroid level or coarser. Lastly, we examined spontaneous vaginal deliveries only, excluding medically indicated preterm deliveries more likely to be due to severe maternal pregnancy complications including pre-eclampsia and gestational diabetes that might or might not be in the causal pathway for pesticide exposures and the outcome. Infants born preterm or born term with low birthweight were more likely to have mothers of younger age, less education, lower neighborhood SES, starting prenatal care after the first trimester, and using Medi-Cal or other government programs instead of private insurance. In addition, infants born preterm were more likely to be a third or later born child, and have mothers with Hispanic or Black race/ethnic background; infants born term but with low birthweight were more likely to be female and a first born child, and born to Black and Asian mothers . First- and second trimester exposures to some pesticides we have selected were associated with a small increase in risk for preterm birth. Specifically, in multivariate adjusted models, first trimester exposures to glyphosate compounds, paraquat dichloride, chlorpyrifos, imidacloprid, permethrin, dimethoate, and methyl bromide, and second trimester exposures to chlorothalonil, glyphosate compounds, paraquat dichloride, simazine, and imidacloprid, yielded adjusted ORs between 1.03 and 1.07 with 95% CIs excluding the null value .

Maternal education changed the OR estimates the most among all covariates.Effect estimates were generally slightly stronger in female infants, except for simazine, which shows stronger effect in males with an OR of 1.06~1.07 . Stratified analysis by season of conception suggested that effect estimates were generally stronger when the peak season of pesticide application concurred with the first or second trimester of pregnancy . When examining chemical classes,round nursery pots first trimester exposures to carbamates , or pyrethroids increased ORs for preterm birth in the high exposure group, compared with the no exposure group, while second trimester exposures to carbamates, organophosphates, or pyrethroids were all associated with small increases in ORs for preterm birth . We generally did not observe elevated ORs for preterm birth among male infants, but observed a stronger 7–11% increase with exposure during the first or second trimester among female infants . Exposure prevalence and effect estimates were generally stronger in infants born to the foreign born or US-born Hispanic mothers than White mothers . Associations between the selected individual pesticides or chemical classes and term low birthweight for each trimester in pregnancy were mostly null. In multivariate adjusted models, we only saw increased ORs for second or third-trimester exposures to myclobutanil ; similarly, exposures to the three chemical classes were not associated with term low birthweight in general, except for marginally elevated odds in infants exposed to 2 or more pyrethroids . Results were similar in our sensitivity analyses, with additional adjustment for maternal prepregnancy BMI and maternal smoking in the years 2007-2010, for NOx as traffic-related air pollution, or restricting to those with a high geocode quality only. For each individual pesticide, adjusting for co-exposure to other pesticides resulted in attenuation of odds by 2-3%; ORs mutually adjusted of three chemical classes or adjusted for prior exposures were mostly similar to or slightly decreased; the mutually adjusted OR for pyrethroids was most stable, suggesting a more robust association with pyrethroids, which were used more in recent years . ORs were generally stronger when we restricted to spontaneous preterm births only for both individual chemicals and chemical classes. In this large California study of women living within 2km distance from agricultural fields on which pesticides were applied, we found that early and mid-pregnancy exposure to selected pesticides known or suspected to be reproductive toxicants and chemicals in the classes of pyrethroids and possibly also carbamates or organophosphates, are associated with a small to moderate size increase in risk of preterm birth between 1998 and 2010.

We found little evidence for pesticides being related to term low birthweight, except for exposures to pyrethroids as a class further corroborating their adverse influence on pregnancy observed for preterm birth and possibly one single pesticide myclobutanil – however, this might have been a chance observation given that we tested 17 individual chemicals. Yet, term low birth weight is a much rarer event than preterm birth and we had less statistical power to estimate small effects accurately. Our positive findings for preterm birth are consistent with biomarker-based studies with measured organophosphates, or pyrethroids and their metabolic breakdown products in maternal blood or urine or umbilical cord blood , though most of the literature assessing environmental exposures to pesticide found inadequate evidence for associations with preterm birth . Less than a handful of studies conducted in the US examined associations for environmental exposures to pesticides from agricultural applications and preterm birth and/or low birthweight and provided month- or trimester-specific estimates . These studies were almost exclusively conducted using California’s unique PUR system, nevertheless they differed in terms of how they assessed exposures and pregnancy outcomes. Our study was in line with an earlier study in the San Joaquin Valley that assessed pesticides labeled with EPA signal word toxicity by summing up their active ingredients applied in the 2.6 km2 section surrounding maternal residences and reported high exposure to pesticides increased risks of preterm birth and low birthweight by 5-9% overall . In contrast, one study reported mostly negative associations between spontaneous preterm deliveries and exposure to 69 chemical groups or 543 specific chemicals in 1998-2011 in the San Joaquin Valley , perhaps because this study focused on late pregnancy instead of early or mid-pregnancy, which is believed to be the critical period for exposures causing preterm birth , and in addition a ‘live-birth selection bias’ could in part explain the negative effect. The other study in northern California reported methyl bromide use within 5 km of mother’s home was also positively associated with gestational age in the first trimester; yet their results were sensitive to buffer size and could potentially be confounded by chloropicrin or diazinon, often used conjunctively with methyl bromide . Maternal, placental, and fetal factors are thought to determine risk of preterm birth and may be affected by prenatal exposure to environmental chemicals . For example, it is known that chlorpyrifos can cross the placenta and enter the fetus, possibly altering the growth and development of the fetus . Mechanisms by which pesticides may affect risks of preterm birth include interference with immune pathways and inflammation , or with metabolic and endocrine regulatory pathways as well as oxidative stress . For example, in-vitro study results suggested that phosmet and chlorpyrifos alter cell viability and induce an inflammatory cytokine profile, indicating that organophosphates may adversely affect trophoblast cells .

Posted on

Climate models are widely used to study the effects of agriculture on climate

With respect to sociodemographic factors, mothers that are exposed to extreme levels of pesticide are more likely to be minorities and have lower education than the sample population as a whole. While we control for these factors, there is potential for the high exposure sample to differ in other unobserved ways that could yield a higher likelihood of adverse birth outcomes. If so, this would result in overestimates of the effects of pesticide exposure on adverse birth outcomes. Additionally, we measure pesticide exposure as all pesticide use on production agriculture in the 2.6 km2 PLS Section encompassing mothers’ addresses. We do so because the diversity of chemicals applied in the San Joaquin Valley is extensive and the cumulative effects of multiple exposures are not well understood. However, some chemicals or combinations of chemicals may not be relevant to reproductive risk. Thus, our coefficients are likely underestimates for individuals exposed to a disproportionately high fraction of chemicals of reproductive concern for their PLS Section, year and birth month. There is some indication that closer proximity to agricultural fields results in increased odds of adverse birth outcomes. For a study of this spatial and temporal breadth it is infeasible to directly measure distance from a sprayed field. However, for the San Joaquin Valley, PLS Sections that have any agriculture generally are agriculturally dominated. Furthermore, the PLS Section is roughly 2276 m on a diagonal. Thus it is highly likely that the vast majority of households in PLS Sections with pesticide use are within 1000 m of a treated agricultural field. If pesticides dissipate much more rapidly,cut flower bucket such that the effect is concentrated within 100 m of pesticide use, our study design would underestimate this relationship due to dilution with individuals living farther away from fields but still within the same PLS Section exposure.

However, for this to be occurring, the population residing on-farm or adjacent to fields must be much smaller than the broader population residing in the San Joaquin Valley for us to observe such small coefficients. Indeed, this makes intuitive sense for California, where farm workers overwhelmingly report living independent of their employers in houses or rental units, particularly if they have a spouse or children. However, our results may under predict adverse birth outcomes in regions where a larger proportion of workers reside in employer-provided housing on or adjacent to fields, where a larger fraction of pesticide are applied aerially, or where permissible chemicals are more environmentally persistent or toxic to humans. We also lack information on residence time at mother’s address and employment. Much of the San Joaquin Valley economy is driven by the agricultural industry. If farm workers were mostly migratory and followed the harvest, our measure of residential pesticide exposure would be inaccurate for this subset of the population. Yet, according to the National Agricultural Workers Surveys for 1996–2011, California farm workers, especially if they have a spouse or children in their household, are settled. Our measure of exposure would also be artificially high if women were applying agricultural pesticides during pregnancy. While ~18% of California farm workers are women, only 1.5% of women reported using pesticides in the past 12 months and 0% of women with a spouse or children had reported doing so. Women could get additional exposure via their spouses, and ~16% of male farm workers reported loading, mixing or applying pesticides in the past year. Finally, the San Joaquin Valley is well known to have substandard environmental quality, frequently exceeding EPA contamination levels for air quality. If such exposures co-vary with pesticide use and vary at small spatial and temporal scales, the coefficient on pesticide exposure could capture additional contamination despite our PLS Township-year and birth month controls. While we cannot be certain we have eliminated all sources of contamination that co-vary with pesticides, including a rich set of ambient air quality and temperature metrics did not change the basic results of this paper. In conclusion, there is a growing literature on the relationship between pesticide exposure and adverse birth outcomes. Yet, evidence of a causal link between infant health and agricultural pesticide exposure remains uncertain due to small samples and lack of maternal or birth characteristics.

Our study is the most comprehensive to date, bringing together the largest data file ever compiled on street-address level birth outcomes and fine scale exposure to agricultural pesticides. We provide robust evidence that there are multiple negative effects of residential agricultural pesticide exposure on adverse birth outcomes, but only for births exposed to very high levels of pesticides during gestation. The documented concentration of impacts in the extreme upper tail of the pesticide exposure distribution may explain why previous studies fail to consistently detect effects of pesticides on birth outcomes. Furthermore, the concentration of impacts in the extreme tail of the pesticide exposure distribution provides policy challenges and public health opportunities to balance these potentially serious but rare outcomes with the societal benefits of continued pesticide use.Although the response of agricultural systems to climate is drawing considerable attention because of the potential for a global food crisis, current understanding of how climate affect agricultural production is highly uncertain since the feed backs between them are not well studied. Agricultural systems are highly vulnerable to climate variability, where the area suitable for agriculture, the length of growing seasons and yield potentials are expected to change under warming scenarios [IPCC, 2007]. In addition, crop growth alters some important physical climate forcings, such as latent heat flux, shortwave radiation, long wave radiation and soil moisture. This two-way interaction is often referred to as a feedback, describing a nonlinear cycle between two systems. Clarifying the importance of these feed backs could improve regional climate simulations in agriculturally intensive areas and enable better prediction of crop production. Variability in atmospheric CO2, temperature and precipitation highly affect agricultural production. The elevated CO2 could increase photosynthetic productivity [Aoki and Yabuki, 1977; Cooper and Brun, 1967; Moss, 1962] and therefore lead to an increase of yield. Amthor [2003] reviewed the previous observations and suggested doubling CO2 could increase the yield by 31% in average. At the same time, double CO2 could lead to 34% reduction of transpiration and double water use efficiency [Kimball and Idso, 1983]. In one study, increase in variability of temperature and precipitation resulted in significant increases in yield variability and crop failures [Mearns et al., 1992]. Warming by 2-4 o C could results in substantial shortening of the growing season, and change of crop calendar, particularly in winter [Butterfield and Morison, 1992]. Furthermore, increasing temperature and precipitation could have different impacts on yields for different crops. For example, a simulation study indicated potato production was increasing while wheat and faba bean was decreasing with increased temperature, and increasing of precipitation had no effect on the yield of potatoes or spring wheat, but could reducing winter wheat yield [Peiris et al., 1996]. Meanwhile, agriculture also affects climate by altering the surface energy, water, and carbon cycle. Cropland plays a very important biogeophysical role in changing climate [Feddema et al., 2005; Foley et al., 2005; Pitman et al., 1999].

Agricultural expansion in business as usual scenario results in significant additional warming over the Amazon and cooling of the upper air column and nearby oceans [Feddema et al., 2005]. Crops alter the small-scale boundary layer structure [Adegoke et al., 2007], such as surface wind and boundary layer height, by increasing canopy height during the growth process. Compared to natural vegetation,flower display buckets cropland has higher albedo that alters the energy budget when converting between forest and cropland [Bonan, 2008; Oleson et al., 2004]. Cropland also alters the water cycle. Both field observations and modeling have shown that conversion of forest to cropland can reduce evaportranspiration and precipitation at the regional scale [Sampaio et al., 2007]. Moreover, agriculture and associated management practices were found to affect the carbon cycle [Lal, 2004]. Global simulation indicates a 24% reduction in global vegetation carbon due to agriculture [Bondeau et al., 2007a]. Growing biofuel crops at previously natural vegetation land could increase greenhouse gas emissions by 50% [Searchinger et al., 2008]. Both observations and numerical modeling are used to study climate effects on agriculture. Laboratory studies using growth chambers and greenhouses showed elevated CO2 could increase net photosynthesis [Aoki and Yabuki, 1977; Cooper and Brun, 1967; Moss, 1962]. These stuides had a short period measurements and the high CO2 concentrations were not realistic. Free air CO2 enrichment experiments [Ainsworth and Long, 2005; Ainsworth et al., 2002; Long et al., 2006] using long term observation confirmed some chamber experiment results that trees were more responsive than herbaceous species to elevated CO2, but crop grain yields increased far less than in previous enclosed studies. Regression models [Rosenberg, 1982] also have been employed to study how climate affects crop yield and this method is still widely used today [Diffenbaugh et al., 2012; Lobell et al., 2008b]. Finally, crop growth models enable yield prediction and hazard prevention.Climate models were first developed for numerical weather prediction in the 1950s, and had a very coarse resolution only contained atmosphere circulation. In 1960-1970s, the climate model included both ocean and atmosphere circulations. In 1980-2000s, the development of regional climate model and sub-grid physical process model not only aim to improve the forecasting skill but also to study the climate change. In climate model, the land surface model provides sensible, latent, and momentum flux for atmosphere model to solve the atmospheric equations. The potential climate sensitivity to land use change is determined by the difference between two simulations that differ only in land use. A key determinant in accuracy of such research is how well the land surface model simulates the surface energy fluxes . The development of land surface model is getting more and more comprehensive to reflect the reality [Bonan, 2008]. Early land surface models represented the physical processes using simple parameterizations. For example, the soil hydrology was represented as a bucket, which could hold some maximum amount of water filled by precipitation, with the excess water becoming runoff.

Currently, most land surface models include all the major parameterizations, such as vegetation photosynthesis and conductance, snow accumulation and melting, radiation transfer, and turbulence processes above and within the canopy, etc. Moreover, some advanced land surface models include the carbon cycle and dynamic vegetation growth. Coupling a land surface model that incorporates dynamic crop growth into a climate model enables simulation of the two-way interactions between climate and crop growth. Recent work incorporating crop growth models into climate models has revealed that dynamic crop growth strongly influences regional climate patterns by altering land surface water and energy exchange. Most of these studies have not rigorously evaluated results against observations of climate and crop variables. Further, interactions between crop growth and irrigation effects on climate are not well examined. The aim of the work is to improve a regional climate model by incorporating a land surface model that simulates dynamic crop growth. Particularly, my work focuses on the improvement and evaluation of the Weather Research and Forecasting Model with updated Community Land Model , a dynamic crop growth model, and an irrigation scheme. As the next-generation mesoscale numerical model, the standard version of WRF includes relatively simple land surface schemes, which potentially constrain model applications for studying the land surface and ecosystem-atmosphere feed backs. By adding the CLM into WRF, I expected an improvement in surface energy flux simulations. Therefore, I first validated the performance for the surface energy fluxes for four vegetation types across the continental of United States in the first chapter [Lu and Kueppers, 2012]. Since one problem in this model was related to the low crop LAI bias and lack of irrigation, I further incorporated the dynamic crop growth model and irrigation into a new version . I evaluated the crop growth and climate variables in the new version and the influence of dynamic crop growth on irrigation effects was quantified. In the third chapter, I used the coupled model to study irrigation effects on heat wave frequency, duration, and intensity.

Posted on

The cost of operation can sometimes be offset through the production of commercial crops

Further measures in 1988, including drainage of evaporation ponds as well as covering of the deeper ponds with fill soil, led to the elimination of aquatic habitat at the Kesterson Reservoir, thus preventing any additional waterfowl from being exposed to selenium at the location . Despite the closure of the Kesterson Reservoir, problems of excessive selenium concentrations persisted in the greater surrounding Grassland wetland area , the largest freshwater wetland ecosystem in California . Farmers of the Grassland drainage area had historically discharged their surface and subsurface runoff through the natural channels of the Grassland wetlands to the San Joaquin River . As a result of increased scrutiny following the events at Kesterson , 33 km2 of the Grasslands were added to California’s Clean Water Act section 303 list of impaired waters due to excessive selenium concentrations in 1988. The wetland’s two major flow channels Salt and Mud Slough followed in 1990 . In 1996, the Grassland Bypass Project was created to amend this situation . The GBP consists of a series of measures to reduce selenium loads in the Grassland marshes and the San Joaquin River, including the reopening of a stretch of the San Luis Drain bypassing the wetlands . The GBP is analyzed below in the section entitled “Selenium load reduction coupled to conveyance into the San Joaquin River”.Two fundamental approaches have been used to manage seleniferous runoff in the San Joaquin Valley: local disposal and conveyance out of the San Joaquin Valley. Geographically,procona buckets the separation between these two approaches aligns with the drainage areas defined under the San Joaquin Valley Drainage Program’s grand management plan . The southern subareas Kern, Tulare and, after the San Luis drain closure, also the Westlands subarea dispose of runoff locally, while the Northern and Grassland subareas convey drainage to the San Joaquin River .

The two approaches share common elements . In either case, methods are employed to decrease the disposal load by first decreasing drainage production and then decreasing the volume and selenium concentrations of the drainage itself . Fundamentally, only the final disposal step differs, with drainage either being evaporated and the salts disposed, or channeled to be diluted in a larger water body .The debate on how to sustainably manage selenium loads and drainage needs in the Central Valley is far from over. Mass balance analysis by the USGS reveals that the drainage needs of the Westlands subarea, the greatest near surface selenium reservoir among the subareas, cannot be met without the retirement of at least one third of the 1,200 km2 of agricultural lands and the use of treatment methods for selenium and salt removal that are as of yet unproven . For a detailed discussion of proposed management scenarios for this subarea the reader is referred to the Final Environmental Impact Statement of the San Luis Drainage Feature Re-evaluation and the technical analysis of proposed plans by Presser and Schwartzman . The current plans under discussion for the Westlands in terms of selenium removal include reverse osmosis and reductive precipitation in microbial bioreactors that have so far only been tested at the pilot scale . It is uncertain whether the proposed bioreactors will be effective with the high salinity inputs expected to result from reverse osmosis . Given such uncertainty I focus on the proven management methods by which seleniferous drainage is actively being managed in the San Joaquin Valley .Locally, seleniferous runoff can be treated using technologies that physically, chemically or biologically remove selenium from water, reused through irrigation of designated land planted with salt tolerant crops, or disposed of in evaporation facilities. A number of removal technologies have been proposed post-Kesterson , however the physical and chemical methods are too costly and the biological methods fail to reduce selenium concentrations in treated drainage to below 2 µg/L at relevant scales . Thus, for lack of better alternatives, drainage reuse and evaporation ponds are so far the local remediation and disposal methods of choice. The reuse of seleniferous drainage as irrigation water on designated reuse plots thus reducing the water volume requiring final disposal, has seen a marked expansion over the last decade, with areas of reuse in the Grassland subarea alone increasing from 7.4 km2 to more than 20 km2 from 1996 to 2009 .

A number of innovative approaches often involving a sequence of increasingly salt tolerant crops in either time or space have been developed and tested in the San Joaquin Valley . Of the 2,600 kg selenium produced in the Grassland subarea in 2009, about half were disposed through drainage reuse . The exact fate of this selenium is as of yet unclear. In particular, there are concerns about the long term sustainability of drainage reuse, after increasing concentrations of dissolved selenium were observed at all monitored soil depths in the only study monitoring soil selenium for an extended time period on a reuse plot in the San Joaquin Valley . Additionally, there are concerns that endangered wildlife such as the San Joaquin kit fox , the kangaroo rat , and the blunt-nosed leopard lizard may be adversely affected if their ranges overlap with reuse areas . Evaporation ponds, which are shallow basins used for the evaporative disposal of drainage water, are deployed primarily in the Tulare and Kern subareas . Since there are no drainage channels or rivers to convey drainage out of these two subareas, evaporation is the only option for final disposal . The design of evaporation ponds has been optimized post-Kesterson to reduce wildlife use and thus the risk of exposure to elevated selenium concentrations in pond water . Specifically, steep levee slopes, elimination of windbreaks, and a minimum water depth of 0.6 m are used to deter waterfowl. Enhanced solar evaporator designs that use sprinklers to ideally eliminate standing water altogether have been tested at the pilot scale, but have not seen wide-spread deployment to date . Evaporation ponds are currently exempt from water quality guidelines that apply to natural waters, however management needs to include active and continuous measures to limit wildlife use through hazing and removal of pond vegetation. Additionally, the provision of nearby alternative habitat for waterfowl is recommended when selenium concentrations exceed 2 µg/L . While these management procedures are necessary during the operation of evaporation ponds to avoid ecological damage, they also greatly increase the cost of operation. In fact, the owners of many of the privately operated evaporation ponds in the San Joaquin Valley have decided to cease operation as a result of imposed requirements .

Another drawback of this method of local disposal is that the evaporate needs to be stored at dedicated disposal sites. This is particularly problematic if evaporation is chosen to continuously dispose drainage from large source areas. For example, it has been estimated that if the drainage program in the Westlands district is expanded as planned, up to 400,000 tons of salt may need to be disposed yearly,procona florida containers which would require dedicated dumpsites covering an area of around 1 km2 every 50 years .The Northern and the Grassland subareas channel a large portion of their runoff to the San Joaquin River. Among the two, only the Grassland subarea has been marked by problematic selenium loads . Due to the Grassland Bypass Project , it is also the subarea in which management of seleniferous runoff has made the greatest progress over the last 20 years . The GBP was created in 1996 through an agreement between the U.S. Bureau of Reclamation and the regional drainage entity of the Grassland Area Farmers under the legal umbrella of the San Luis and Delta-Mendota Water Authority. It consists of three central measures. First, a 45 km stretch of the San Luis drain was opened to convey subsurface drainage from the GAF to Mud Slough, thereby bypassing most of the wetlands . Second, limits were imposed on the total allowable selenium discharge by the GAF and these limits were set to decrease over time . Finally, ongoing monitoring of water quality and quantity was initiated across the project area to enforce limits and gage impact . The original 5 year project was extended by 8 years in 2001 and then by 10 years in 2009 . Whereas the use agreement specified load dependent incentive fees for exceedance of the specified selenium load limits, the true innovation consisted in enabling the GAF to develop an internal selenium load trading program . This trading program represents the first ever cap-and-trade style program for any pollutant allowing trades directly between non-point sources. It is also the first instance in which “total maximum daily load” requirements under the Clean Water Act have been successfully enforced against a non-point source . Since 2001 the drainage management efforts of the GAF have shifted away from the load trading program and towards centralized management for the region , however the overall strategy has remained consistent . Funds obtained as part of the GBP are used to support programs and actions aimed at reducing selenium loads , such as the local drainage reuse measures described above. The direct aid provided by the GBP as well as the economic incentives for load reduction and conservation initially created by its load trading program have led to an overall reduction in selenium loads .

In fact, the total annual loads of selenium discharged through the San Luis Drain have decreased continuously over the course of the project, from 3,110 kg in 1997 to 560 kg in 2009. This has been reflected in a reduction of selenium loads in the San Joaquin River from a pre-project annual average of 3,690 kg to 690 kg in 2009 . Weekly selenium concentration monitoring data collected below the confluence with the Merced between 1996 and 2012, reveal seasonality in selenium loads , but reaffirm the overall load reductions reported by the GBP . However, the reduction is primarily due to a reduction in total discharge from 6.1×107 to 1.6×107 m3 in 2009. Average selenium concentrations of discharge decreased from 67 to 33 µg/L in the same time period and were still significantly above the water quality criterion of 5 µg/L. Much of the reduction in selenium loads can be credited directly to the impact of the projects and measures implemented as part of the GBP. For example, through the San Joaquin River Water Quality Improvement Project, more than 20 km2 of land have been purchased and planted with salt tolerant crops providing the capacity to reuse up to 1.9×107 m3 of drainage water per year . A GBP unrelated factor in the successful load reduction was the prevalence of drought during the early 1990s which incentivized investments in efficient irrigation technology and other conservation measures on the side of the GAF . Therefore, considerations of best management practices and technology aside, the creation of quantitative economic incentives for selenium load reduction should be a priority of any seleniferous drainage management program. Unfortunately, comprehensive incentive programs such as the GAF load trading program remain the exception rather than the rule, for selenium as they do for many other pollutants.The primary motivation of the GBP was the protection of the sensitive wetland habitat in the Grassland area. Through circumvention, the GBP has effectively removed all seleniferous drainage water from 145 km of channels that supply water to more than 650 km2 of sensitive wetlands and wildlife areas . This led to a rapid decrease in selenium concentrations at the monitoring stations in this area after the GBP’s implementation in 1996. In Salt Slough for example , concentrations have dropped from 16 µg/L in water year 1996 to below 2 µg/L for most of the project duration . In addition, selenium concentrations monitored in the tissue of various fish species collected at the slough , minnows , and others) have dropped from toxic levels between 1992 and 1995 to below levels of concern after 1998 . Accordingly, the slough was removed from California’s 303 list of impaired waters in 2008 .

Posted on

Diflubenzuron is an insect growth regulator and is not listed in the draft regulations

As an example, Ogunjemiyo et al studied a more simplified crop setting, consisting of a single, highly evapotranspiring crop in an area with simplified meteorological conditions . Similar uniformity occurs throughout much of the Midwest, where mono-cultures of corn and soybean would permit the study of water vapor patterns while reducing the impact of variation in crop type. The Central Valley, with its large variety of crops and management practices, resulting in non-uniform distributions of aerodynamic roughness, ET rates, and landscape structures throughout the scene, was perhaps not the ideal location to test this approach. Unfortunately, the type of data we used, including AVIRIS– derived water vapor, and LST from MASTER, is not widely available outside of data sets acquired for the HyspIRI-Airborne Campaign. We would suggest a targeted campaign, acquiring combined AVIRIS-MASTER for agricultural studies, over more simplified and better instrumented sites would be of great benefit. Beyond advancing our ability to capture patterns of field-level ET with water vapor imagery, this imagery may prove valuable for regional analyses of water transport. There are many challenges associated with linking water vapor to crops at the field-level as outlined above, but the idea behind this work will likely hold at a smaller scale. Lo and Famiglietti found that in the Western United States the irrigation from California has been shown to increase the summer stream flow of the Colorado River by 30 percent. Water vapor imagery, if acquired more consistently and over larger areas, offers an additional tool that could be used to capture finer scale water vapor transport, to complement models and coarser scale observations from sensors such as MODIS. These large movements of water vapor have implications for climate change and land use, and call upon the need to increase monitoring of water vapor patterns in areas with large irrigation inputs. Therefore, 10 liter pot a study that examines the ability of water vapor imagery to assist in regional water transport assessments could be of high value.

CDPR’s draft use regulations, designed to address pesticide runoff and drift, could have a potentially significant economic impact on California agriculture, as well as to the supporting industries and communities, due in part to the large number of active ingredients listed in the draft regulations. Assessing the potential effects of the regulations is complicated by a number of factors. First, pest management programs for many crops, such as alfalfa, walnuts, strawberries, lettuce, rice and several others, include at least one of the CDPR’s targeted 68 active ingredients and efficacious alternatives are not always available. Often, the most common alternatives to an individual active ingredient are also subject to the draft regulations. Second, mandated buffer zones that define the minimum distance that must be left between a sprayed area and a “sensitive aquatic site,” as a function of the application method, are an important component of the regulations. Third, the amount of field acreage affected by buffers depends on the distribution of crop acreage relative to the location of a sensitive aquatic site. Fourth, the draft regulations propose to use the definition of sensitive aquatic site as “any irrigation or drainage ditch, canal, or other body of water in which the presence of pesticides could cause adverse impacts on human health or aquatic organisms.” This article focuses on the potential economic impacts of the draft regulations for rice production in Colusa and Butte counties due to the listing of two selected active ingredients. It is drawn from a larger report that considers the economic effects of the draft regulations for 20 county-crop pairs. The analyses are performed at the county level because the distribution of crop acreage relative to the location of sensitive aquatic sites is an important determinant of potential economic impacts.According to the National Agricultural Statistics Service , California’s rice crop is the second largest in the United States, accounting for 22% of the value of national rice production. Rice is the tenth most valuable crop grown in California, contributing 2.8% to the total value of production in 2009. In 2009 there were 563,974 acres of rice in California.

The statewide average yield was 4.38 tons per acre, production totalled 2,472,614 tons, and the price of rice was $390 per ton with a total farm gate value of $963,526,000. The top rice-producing counties in California by value are Colusa, Sutter, Butte, Glenn, and Yuba, according to county agricultural commissioners’ data reported by NASS. Rice is the most valuable crop in both Colusa and Butte counties. In 2009 rice accounted for 40.7% and 33.9% of total crop value in Colusa and Butte counties, respectively. Colusa was the top rice-producing county in the state, accounting for 25% of the value of all California rice. In 2009 Colusa County farmers grew 152,400 acres of rice, which yielded an average of 4.5 tons per acre, and produced a total of 685,800 tons of rice; the price was $355 per ton, for a total value of production of $243,459,000. In 2009 Butte County farmers grew a total of 103,416 acres of rice, which yielded an average of 4.7 tons per acre and produced a total of 486,055 tons of rice; the price was $379 per ton, for total value of production of $103,265,000. Together, Butte and Colusa counties accounted for 44% of California rice production in 2009.According to data from Demars and Zhang , a draft report under preparation for CDFA, there was a total of 250,800 acres of rice in Butte and Colusa counties in 2009, divided among 4,947 different fields. For that report they used Geographic Information System technology to combine U.S. Geological Survey National Hydrography Dataset and California Department of Water Resources land-use layer data into a common projected coordinate system. These were then run through a custom script that reported the amount and percent of crop land bordering sensitive aquatic sites. Affected Acreage Demars and Zhang concluded that while the actual acreage that would be in 25-ft. buffers was a small share of total acreage , the number of fields affected was a large share of the total number of fields . Thus, the increased management costs due to the buffers could be substantial. Under a 150-ft. buffer, both the acreage in buffers and the number of fields affected were substantial shares of the total: 19% and 96.5%, respectively. We used Demars and Zhang’s findings, along with base yield information from county agricultural commissioners’ reports, cost information from UC Cooperative Extension cost studies, and estimates of yield reduction from the scientific literature to estimate the changes in gross and net revenues for the two counties in response to the regulations.

The most important active ingredients for rice production that would be prohibited for use in buffers under the draft regulations are propanil , which is used as a cleanup herbicide for weed control, and lambda-cyhalothrin , a pyrethroid insecticide which is used to control rice water weevil. Weeds are the most important pest in rice, reducing yields by 17% in the United States as a whole compared with 8% and 7% losses yield losses due to insects and diseases, respectively. Thus,drainage gutter weed control through a combination of water management, herbicide application, and other methods is crucial for sustaining the productivity of U.S. rice-cropping systems. Propanil is the most widely used herbicide in rice. It is a relatively inexpensive material, to which water grass weeds in rice have not yet developed resistance, unlike other available herbicides, including thiobencarb, cyhalofopbutyl, and bensulfuron-methyl. Thus, growers are able to use it as a cleanup herbicide post-planting, following the application of one or more other active ingredients. Most propanil in Butte and Colusa counties is ground-applied, so there is relatively little scope for growers to reduce the impact of the draft regulations by changing from aerial to ground applications. Because of widespread herbicide resistance among common weed species in rice fields, it is difficult to identify post-planting alternatives to propanil as part of an effective weed management program. There are a few cultural alternatives, including increasing the depth of water in order to “drown” weeds, severely drying the field to desiccate sedges, or using flooding to germinate weeds early and then kill them preplant with a broad-spectrum herbicide such as glyphosate. However, none of these methods are a perfect substitute for propanil. Each compromises the efficiency of the production system and may result in considerable yield loss. In order to compute the effects of the draft regulations on total and net revenues, we specify that propanil is used as a cleanup spray, except in the 25-ft. buffer where no cleanup application is made, and that only half of total field acreage requires a cleanup spray. Based on the scientific literature, we assume that rice yields decline by 40% in the untreated buffer. The per-acre cost of treatment declines by 100% in the buffer because no cleanup spray is applied. Also, the uncontrolled weeds in the non-treated buffer zones will produce large quantities of seeds, thereby fortifying the weed seed bank and ultimately increasing weed populations over time. The rice water weevil is one of the most economically damaging invertebrate pests in California rice. Root pruning by larvae reduces growth, tillering, and yield of affected plants. Buffer zone requirements are particularly problematic for rice water weevil management due to its life cycle and distribution in rice fields. This insect overwinters in grassy areas around rice fields; these areas are usually associated with sensitive aquatic sites such as sloughs and ditch banks. In early spring the rice water weevil moves to flooded rice fields but does not tend to establish very far into the fields. A 25-ft. buffer would encompass most of the area where damage from rice water weevil would be expected to occur. Lamda-cyhalothrin is the major insecticide currently used to control rice water weevil. In 2009 all applications of lambda-cyhalothrin in Butte and Colusa counties were made by air, according to CDPR Pesticide Use Reporting data. This is driven by the timing of post planting applications; rice fields are treated with lambda-cyhalothrin when the rice plants have one to three leaves.

At this stage of development, water movement and soil disturbance caused by the equipment used for ground applications can uproot rice plants, reducing stands. Thus, the timing of the application must be changed in order to change the application method and avoid the 150-ft. buffer requirement. Lamda-cyhalothrin applications made after the three-leaf stage of rice are not effective against the rice water weevil. Recent scientific research findings indicate that pre-flood applications of lambda-cyhalothrin can be effective, although this approach has not been adopted widely by growers.The UC Integrated Pest Management Guidelines for rice water weevil list two alternative chemical controls to lambda-cyhalothrin: -cypermethrin and diflubenzuron . -cypermethrin is a pyrethroid that is also listed in the draft regulations. Hence, it would not be a viable alternative to replace lambda-cyhalothrin if the draft regulations were implemented.However, diflubenzuron is also not available as an alternative buffer treatment because of label restrictions that require a 25-ft. vegetative buffer between ground application areas and bodies of water. Given these limitations, growers concerned with rice water weevils would likely use a preventive, pre-flood ground application of lambda-cyhalothrin if the draft regulations were implemented. In order to evaluate the economic effects of the draft regulations on rice water weevil management costs and associated rice revenues, we compare the current post-flood aerial application method to pre-flood ground application to eligible acreage under the draft regulations. Rice water weevils tend to be economic pests near field edges, and growers do not usually treat entire fields. We proxy this management pattern by assuming that the land within 100 feet of the edge of a field represents, roughly, the land that is treated currently. Under the draft regulations, acreage within 25 feet of a sensitive aquatic site cannot be treated with a ground application. We assume that acreage within this buffer is left untreated, and that lambda-cyhalothrin is groundapplied on the remaining eligible acreage within 100 feet of the field edge. Based on the scientific literature, we assume that the acreage treated with a pre-flood ground application sustains a 15% yield loss and the untreated acreage sustains a 23% yield loss.

Posted on

Hypatia automatically deploys clustering experiments to account for all of these challenges

Local tasks fetch from the file system or co-located database, remote tasks fetch their data over the network between EC and PC. Hypatia estimates the input/output dataset transfer time to the remote in two ways. We use the first when Hypatia has no history on the job type in the database, i.e., when the job is run for the first time. knowledge of job’s tasks is available, Hypatia uses the job’s input and output dataset sizes and the number of concurrent connections, to estimate transfer time using the iPerf lookup table with values representing time in seconds needed to transfer data from the edge to remote cloud for a dataset of a size given in kilobytes and a number of concurrent connections. The lookup table provides fast access but may introduce error because the data in the table is a “snapshot” in time. Table 5.1 shows a snapshot of a Hypatia lookup table. Files of different sizes, listed on the left, are sent over the network with a variable number of concurrent connections, listed across the top. The data is produced using iPerf to measure the network performance between the EC and PC , when Hypatia is first started. Each of the numbers in the table is an average of over 10 profiling runs. For job types that Hypatia has seen , Hypatia uses the average time measured across past tasks of the same type to estimate the transfer time. This computation takes longer than a simple table lookup but uses recent history to makes predictions . Hypatia launches a set of virtual machine instance types to the EC and PC,plant pot with drainage the number of each is specified by the user in the job description. Instance type names map to the amount of memory and compute resources that each provides. We deploy one Hypatia worker to each processor. Thus, we view each compute resource in terms of the number of workers it can support. The Hypatia queue is placed on the local instance and has two queues: local and remote. Based on the load split ratio, Hypatia places tasks in the local or remote queues. Workers then pull tasks from their assigned queue for execution on a first-come-first-served basis.

Given a Job J with n tasks and D dataset, di is the amount of data that must be transferred if the job is to use the PC. The scheduling plan uses this value, the network bandwidth between the EC and PC , the number of concurrent connections required to saturate the link, the number of available workers in each cloud, and the current state of the queues. The scheduler computes the ratio of the number of tasks that will execute locally and remotely such that time to completion for the job is minimized. Hypatia estimates the time to completion for EC tasks as the average time to complete past tasks of a similar type , and the state of the EC queue, which is expressed as the lag caused by unfinished tasks in the EC queue . The estimate for PC tasks includes a time estimate for data transfer. In addition, Hypatia uses the lag in the PC queue instead of the EC queue for this estimate.To evaluate the efficacy of the Hypatia scheduler, we execute multiple workloads across multi-tier deployments and measure the time to completion per job. For the experiments, we consider one edge instance and two sets of public cloud instances. On the edge, we use an m3.xlarge instance with 4CPUs and 2GiB memory. In the public cloud, our first instance set consists of a single “free tier” t2.medium instance type with 2 CPUs and 4GiB of memory. Our second public cloud set consists of an m5ad.xlarge instance type with 4 CPUs and 16GiB of memory. Our experiments consider deployments with 2, 4, 8, and 12 CPUs in the public cloud. Our workloads consist of two machine learning applications that we developed in the previous chapters of this dissertation: linear regression and k-means clustering. For each experiment, we evaluate the performance of the mixed load against executing all of the jobs on the edge cloud or all of the jobs in the public cloud . We refer to Hypatia deployments as mixed because Hypatia will schedule job tasks on both the edge cloud and the public cloud when doing so reduces time to completion.Since our jobs consist of tasks with different parameters, their time to transfer data and to perform the computation differs across tasks. To empirically evaluate this effect, for each set of machine learning models, we use two sets of jobs: uniform containing tasks having the same parameters, and variable containing tasks having different parameters . The statistics for each machine learning algorithm, for the variable or uniform sets, for EC and PC workers are listed in Table 5.3.

We present mean and standard deviation for the time it takes to transfer the data needed for the task and to process the task .The first experiment uses jobs that have 900 tasks, where each task is given a set of parameters that it then uses to compute the coefficients of a linear regression model based on two time series. The length of the time series defines the dataset size, and thus the computation time per task. Jobs are either uniform or variable task sets as defined above. The uniform linear regression tasks take as input a one month time series of 5-minute interval measurements . The variable job tasks take as input time series that vary between one day and one month worth of 5-minute measurements. For this experiment, the EC has 4 workers, and the PC has 2, 4, or 6 workers. Figure 5.3 shows the average time in seconds it took to complete a job with 900 uniform tasks using 4 EC workers and either 2 , 4 , or 6 PC workers , on average across 5 jobs. For each set of results, the first three bars show the average time in seconds that it takes to complete a task using mixed , local , and remote deployments, respectively. The second three bars per set show the average time in seconds that Hypatia estimates that each deployment should have taken, which we discuss later in this chapter. In every case, the mixed Hypatia workload performs best for time to completion for the workload. For uniform jobs, the mixed load using 2 remote workers finishes in 176 seconds on average, while EC-only takes 207s and PC-only takes 1160s on average, respectively. With 6 PC workers, the mixed workload takes 144s, EC-only takes 199s, and PC-only takes 387s on average. The results also show that as the number of PC workers increases, Hypatia is able to accurately split jobs between the two resource sets and time to completion decreases: 176s for PC2, 160s for PC4, and 144s for PC6 on average. Data for linear regression experiments is stored in the database running on a separate instance within the local edge cloud . This makes data transfer from an EC worker much faster than the transfer from the PC worker. The average data transfer time for EC workers is 0.85s and for PC workers is 2.42s. Processing time is 0.03s for EC and 0.13s for PC workers on average. For the variable jobs in Figure 5.3 ,black plastic planting pots the mixed workload takes an average of 102s with 2 remote workers, 91s with 4 remote workers, and 80s with 6 remote workers. Even though we expect EC workload to be the same across all 3 experiments in this figure since local workers don’t change, we still see some variation with workloads taking on average 110s in PC2, 105s in PC4 and 105s in PC6. The total time of the workload is significantly less than for uniform workloads because many jobs employ much smaller input data sets sizes.

The average data transfer time is 0.47s for EC and 1.6s for PC workers, while computation takes 0.02s for EC and 0.12s for PC workers, respectively. We see that for this particular job type, the PC workers take more time to fetch and process tasks. To investigate this further, we next change the instance type from the free tier t2.medium to use m5ad.xlarge instead – which we note is 4.4 times more expensive monetarily. Figure 5.4 shows results for the linear regression application with uniform jobs and with remote instances having 4, 8, and 12 workers, respectively. Only using 12 PC workers, can we observe remote experiments outperforming EC-only workloads. Similarly, with the increased number of workers the scheduler picked a correct split to minimize the time to completion, which was 158s, 122s, 100s, for 4, 8 and 12 PC workers, respectively. Local only experiments took on average 195s, while remote took 566s, 296s, and 187s respectively.The mean computation time of the uniform workload is 0.19s for EC and 0.31s for PC workers, with a standard deviations of 0.04s and 0.05s respectively. For the variable workload, the mean computation time for EC workers was 2.86s with a standard deviation of 5.09s, while PC workers have a mean of 4.36s and a standard deviation of 8.13s. Like for the linear regression application, Hypatia is able to deploy the Kmeans workload to achieve the shortest time to completion on average. As the number of PC workers increases, Hypatia adapts to employ the extra computational power by using the high-overhead communication link sparingly. The average time to completion is 306s, 238s, and 198s for mixed variable workload for 4, 8, and 12 PC workers, respectively. For EC-only variable workload, we see 423s, 430s, and 424s for three different repeats of the same experiment. PC-only workloads took 778s with PC4, 423s with PC8, and 316s with PC12. For uniform mixed workloads, the average time to completion is 69s, 56s, and 43s for 4, 8, and 12 PC workers, respectively. EC-only workload took 84s on average while PC-only workloads took 250s, 121s, and 82s on average.With more remote workers we observe that PC-only outperforms EC-only.

These results reflect the importance of considering both data transfer and computation time when deciding when to use remote, public/private cloud resources in multi-tier settings. Moreover, they show that Hypatia is able to adapt to different resource configurations to achieve the best time to completion for different machine learning applications automatically.With this dissertation, we present Hypatia – a scalable system for distributed, data-driven IoT applications. Hypatia ingresses data from disparate sensors and systems , and provides a wide range of analytics, visualization, and recommendation services with which to process this data and extract actionable insights. With a few examples of commonly used machine learning algorithms, like clustering and regression, we provide abstractions that make it easy to plug in different algorithms that are of interest to agronomists and other specialists who work with datasets that can benefit from their locality. Hypatia integrates an intelligent scheduler that automatically splits analytics applications and workloads across the edge, private, and public cloud systems to minimize the time to completion, while accounting for the cost of data transfer and remote computation. We use Hypatia to investigate K-means clustering consider different methods for computing correlation, using large numbers of trials , and cluster degeneracy.It then scores the clustering results using Bayesian Information Criterion to provide users with recommendations as to the “best” clustering. We validate the system using synthesized data sets with known clusters for validation, and then use it to analyze and scale measurements of electrical conductivity of soil from a large number of farms. We compare our approach to the state of the art in clustering for EC data and show that our work significantly outperforms it. We also show that the system is easy to use by experts and novices and provides a wide range of visualization options for analysts. We next extend the system with support for data ingress from sensors and develop a new approach for “virtualizing” sensors to extend their capability. Specifically, we show that it is possible to estimate outdoor temperature accurately from the processor temperature of simple, low-cost, single-board computers .

Posted on

Doing so poses challenges for k-means clustering and all variants mis-classified some points

Advanced users can modify the following Centaurus parameters: maximum number of clusters to fit to the data ; number of experiments per K to run; number of times to initialize the k-means clustering ; type of covariance matrix to use for the analysis ; whether to scale the data so that each dimension has zero mean and unit standard deviation . Centaurus considers each parameterization that the user chooses as a “job”. Each job consists of multiple tasks that Centaurus deploys. Users can check the status of a job or view the report for a job . The status page provides an overview of all the tasks with a progress bar for the percentage of tasks completed and a table showing task parameters and outcomes available for download and visualization. Centaurus has a report page to provide its recommendation. The recommendation consists of the number of clusters and k-means variant that produced the best score. This page also shows the cluster assignments and spatial plots using longitude and latitude . For additional analysis, users can select “advanced report” to see the correlation among features in the dataset, scores for each k-means variant, best clusterings for each one of the variants, etc. We implement Centaurus using Python and integrate a number of opensource software, packages, and cloud services. These services include the Python Flask Flask  web framework, RabbitMQ RabbitMQ  and Python Celery Celery  for messaging and queuing support, and an PostgreSQL SQL database PostgreSQL and MongoDB Community Edition NoSQL database MongoDB , which we use to store parameters and results for jobs and tasks. Other packages include Numpy Walt et al. , Pandas McKinney et al. , SciKit-Learn Pedregosa et al. , and SciPy Jones et al. for data processing and Matplotlib Hunter and Seaborn Seaborn for data visualization. Centaurus can execute on any virtualized cluster or cloud system and autoscales deployments by starting and stopping virtual servers as required by the computation. In our evaluation in this chapter, we deploy Centaurus on a private cloud that runs Eucalyptus v4.4 Nurmi et al. , Aristotle ,growing strawberries vertical system which has multiple virtual servers with different CPU, memory, and storage capabilities. We build upon, generalize, and extend this sys- tem in Chapter 5 of this dissertation.

Centaurus stores the cluster assignments for each experiment, which is the result with the largest log-likelihood value across initial assignments. This Centaurus instance only considers clustering results when all clusters have at least 30 points, in its computation of BIC and AIC. Finally, as described above, Centaurus reports the result with the highest average BIC score the “best” clustering across every K considered for all variants. Note that Dataset-1 was generated using a GMM where all dimensions are independent of each other and are identically distributed. Thus the “perfect” classification results generated by the Full and Diagonal methods indicate that they correctly disregard any observed sample variance or covariance. The results for Full-Untied with Dataset-2 and Dataset-3 illustrate Centaurus ’s ability to correct for cross-dimensional correlation. The generating GMM in both cases is untied . Also, unlike in Dataset-1 where there are three distinct clusters with separated centers, we purposefully placed the cluster centers of Dataset-2 and Dataset-3 near each other and generated distributions that overlap.To visualize the effect of different k-means variants on BIC score, we perform 2048 single k-means runs for each variant for synthetic datasets described in 3.4Figure 3.2 shows histograms of the BIC scores for each the of three synthetic datasets. We divide the scores among 100 bins. For each dataset, we present six histograms, one for each of the k-means variants, represented in different colors, where each variant has a total of 2048 single k-means runs. The X-axis depicts BIC scores from experiments – farther right corresponds to larger BIC and thus higher quality clusterings. This corresponds to a clustering with four clusters having cardinality of 2188, 531, 308, and 205, respectively. The second-best clustering has BIC score of -8925.4 and three clusters with cardinality 1733, 973, and 526, respectively, as shown in . Figure 3.4c shows the difference between these two clusterings. A specific data point is shown if it has a different cluster number assignment when we rank clusters by cardinality. For this data, clearly these clusterings differ. Thus, doubling the number of experiments from 1024 to 2048 allows Centaurus to find a clustering with a better BIC score.

The Sedgwick dataset has a more stable outcome in terms of the best BIC score when increasing the number of experiments. Figure-3.3b shows that even with 256 experiments , we achieve the same maximum BIC score as with 2048 experiments. The best result has a BIC score of -7468.0 and three clusters with 1111, 996, and 568 elements . This result is consistent over many repeated jobs with a sufficiently large number of experiments i.e. any job with more than 256 experiments produced this same clustering as the one corresponding to the largest score. The second-best clustering agrees with the best result on the number of clusters with cluster cardinalities of 963, 879, and 833, and a BIC score of -7529.8 . While these clusters do differ, Figure 3.5c shows thatthe differences are scattered spatially. Thus the best and second-best clusterings may not differ in terms of actionable insight. For the UNL field, the best and second-best clusterings are shown in Figure 3.6. These are both from job-2048. The best clustering has six clusters with cardinalities 2424, 1493, 1138, 561, 111, and 70, respectively. The second-best clustering has four clusters with cardinalities 2730, 1615, 838, and 614, respectively. From these features and the differences shown in Figure 3.6c it is clear the best and second-best clustering are dissimilar. Further, the second-best clustering from job-2048 is the best clustering in job-64, job-512, and job-1024 respectively. As with the Cal Poly data , doubling the number of experiments from 1024 to 2048 “exposed” a better and significantly different clustering. Unlike the results for the synthetic datasets, the best clustering for the Veris EC datasets is produced by the Full Untied variant for sufficiently large job sizes. This result is somewhat surprising since the Full Untied variant incurs the largest score penalty in the BIC score computation among all of the variants. The score is penalized for the mean, variance, and covariance estimates from each cluster. The other variants require fewer parameter estimates . Related work has also argued for using fewer estimated parameters to produce the best clustering Fridgen et al. leading to an expectation that a simpler variant would produce the best clustering, but is not the case for these datasets.

Because Centaurus considers all variants, it will find the best clustering even if this effect is not general to all Veris data.To evaluate the differences among clustering variants as applied to farm datasets, we present three largest jobs with their best BIC scores and the variant that produced the best score . For each farm dataset we present results from the three largest experiments: Job-512, Job-1024, and Job-2048. The results show that for most of the multivariate datasets,growing vegetables in vertical pvc pipe the best clusterings came from the Full-Untied variant with a very small number of exceptions: The ALM dataset and Job-1024 for TC1 dataset . For the univariate datasets , the only variant possible is spherical since there are no additional dimensions with which to compute the covariance. For these datasets, the Spher-Untied variant performs best.Degenerate clusters are a surprisingly frequent, yet under-studied, the phenomenon when clustering farm datasets. We next investigate the frequency with which degeneracy occurs for different datasets and different numbers of clusters. Figure3.7 illustrates the search space for the best BIC score with the estimated joint distribution of BIC scores and the number of elements in the smallest cluster . The figure represents a Job-2048 for CAP dataset, with all six variants and all values of K . Darker colors on the graph represent higher density regions. Our system uses all of the k values and all of the variants when choosing the model with the highest BIC score. Per-component distributions are available on the sides of the graph. The graph indicates that the highest BIC scores often come from the clusterings that have one or more almost empty clusters. Particularly for variants that rely on an estimate of co-variance between dimensions, inferences made about the means of these clusters are suspect when their sizes are small. Note that for larger values of k, such clusterings can be common. For example, this particular job had 50961 or 41.5% degenerate and 71916 non-degenerate experiments. To illustrate this effect more fully, we divide the experiments based on the number of clusters, k, and illustrate how degeneracy behaves with increasing k in Table 3.6. The total number of experiments for multivariate datasets was 12288 for each k and all six variant types. The univariate datasets had 4096 experiments and include only two variants for each k. For each farm, the results show the number of non-degenerate clusterings for each k. In some cases, the number of non-degenerate clusterings decreases as k increases. To emphasize the overall degeneracy across all Job-2048 experiments for each dataset, we summarize the percentages of experiments with fewer than 30 elements in their smallest cluster in Table 3.7. The smallest percentage of degenerate clusters is 6% for the GR1 dataset and the largest percentage was 72% for TC2 dataset. We have chosen 30 as a reasonable rule of thumb for a cluster size from which to make an inference about the mean of each cluster in the experiments having three dimensional data. Because k-means converges to a locally optimum solution for non-convex solution spaces, the choice of initial assignment can effect the clustering it finds. Often, users of k-means will run it once, or a small number of times assuming that the local minimum it finds is “close” to the global minimum. In this subsection, we investigate the validity of this conjecture for soil EC data across farms. Figure 3.8 presents the best BIC scores for different experiment sizes for all of the farm datasets. In some of the jobs, the best BIC score occurs only once amongst all of the experiments while in others the best BIC score is more common among multiple experiments . Thus a small number of trials is likely to result in the most common clustering rather than the best one. More importantly, an increase in the number of experiments increases the chance of finding the best BIC score.

The x-axis for each graph in Figure 3.8 is the power of 2 in the number of experiments. For most of the graphs, k-means finds the best BIC consistently beyond some large number. However, for a few of them, it appears that an even greater number of experiments may be necessary before a single consistently large BIC is determined. The graph in Figure 3.3c, for example, seems to indicate that an even larger number of experiments may yield a larger BIC. Thus, for the EC available to our study, it is clear that the best clustering is often rare and thus requires a large number of independent trials to determine. Even though the best clustering may be rare, it may also be that it differs from the most common clustering by so little as to make the effort required to find it unnecessary or wasteful. Table 3.8 compares the clustering determined by the best BIC scored cluster to the most common clustering for each of the data sets across their largest jobs. We limit the clusterings to those with at least 30 elements in each cluster to prevent degenerate clusterings from clouding the results. For each data set we show two rows. The row marked “B” shows the BIC score, the value of k, the cardinality of each of the k clusters, and the number of occurrences of this clustering for the clustering having the best BIC score. The row marked “MC” shows the same information for the most common clustering.

Posted on

Community Supported Agriculture connects farmers and the consumers of their products

The Mask R-CNN model produced an AP of 0.8996. Results were not stratified by size category. This result is 8 AP percentage points higher than the Mask R-CNN results on the Nebraska dataset for the medium sized category. This could be due to a number of factors. First, the imagery used was 0.26 meter, which is much higher resolution than Landsat 5’s 30 meter resolution. Therefore, each building instance could have a higher amount of pixels with which to compute informative features. A larger model with more parameters was also used, Resnet-101, which increases learning capacity as well as the tendency to overfit. Finally, some figures referenced in Wen et al. indicate that bounding boxes encompass multiple individual buildings. If these groups of buildings were used to represent a single instance to calculate mAP, the score would be higher than if each building was considered as a separate instance.Deines et al used a random forest model to classify irrigated pixels across the HPA using spectral indices, GRIDMET precipitation, SSURGO soil water content, a topographic DEM, and other features. The random forest model was trained on 40% of the data, validated on 30%, and tested on the remaining 30%. Since this method classified pixels and not objects, metrics are not directly comparable, however, the result shows that the model was quite successful in accurately mapping irrigated area across a wide climatological gradient. Pixelwise, overall accuracy for classified irrigated area was 91.4%, and results were visually assessed to correspond well with GCVI computed from Landsat scenes. Given that this model performed so well and that we are able to inspect feature importance for random forest models more easily than with CNN models, it is useful to determine if the features that were successful in the random forest model were or were not shared by the Mask R-CNN model so future models may take advantage of both approaches. The top six most important features in the random forest model that led to this accurate classification were,lettuce vertical farming in order of importance: latitude, slope, longitude, growing degree days , minimum NDWI, and the day-ofyear at peak greenness.

The authors note that the random forest model uses latitude and longitude to separate scenes by climatological gradients, which can improve detection as different climatological areas contain different agricultural patterns. Of these features, only minimum Normalized Difference Water Index can be directly learned by Mask R-CNN, though the model likely recognizes scene qualities such as bare soil that are correlated with a drier climatology. This indicates that pixel-wise features besides reflectances can contribute, and be even more important, than reflectance alone. Future approaches based on CNN’s could make use of not just reflectance information, but also the features named above. However, this would currently require training CNN models from scratch. This could be prohibitively expensive, increase training times, or lead to lower accuracies since weights are initialized randomly and training from scratch may not lead to as informative features as those that are arrived at after pretraining. The results from Deines et al. do not separate individual fields from each other; many clearly distinct fields are mapped as a single irrigated area unit. The method used here is useful for tracking pixel level changes in irrigation, yet it cannot be used for tracking field level statistics from year to year. The amount of training data was the second largest factor in determining the resulting performance metrics. Decreasing the amount of training data by 50% decreased performance metrics by a substantial amount. The large performance drop for the small size category, which had an 11.4% difference in terms of AP percentage points, indicates that more training data is especially important for more uncommon center pivot size categories, which highlights the importance of using as much training data as possible when training CNN-based models. Using a NIR-R-G vs RGB composite did not affect the validation AP, which is expected since center pivots are visually defined by shape rather than spectra. Correct preprocessing choices were also important in order to achieve good results, and the most important of these was to convert image values to 8-bit integers before normalizing by the mean. This step clips the very large values present in Landsat 5 OLI scene chips from high reflectance from snow and clouds and adjusts the range of values of the training set to more closely match that of the pretrained model’s original dataset, which in this case was Imagenet. Models trained on the original TIFF imagery performed very poorly even when these images were normalized by the mean .

Adjusting other hyperparameters did not affect the model performance as much as converting the data type of each image sample to 8 bit integer and using the largest training data size. Examples of hyperparameters tested include doubling the number of region proposals generated during the training stage and increasing the amount of non-max suppression to prune low-confidence region proposals. I expected that more region proposals and more aggressive pruning of low confidence proposals would lead to a better performing model, however making these adjustments did not impact model performance, which is likely because the default number of regions generated was sufficient to intersect with most potential instances of center pivots in each sample.These results clearly show that segmenting small center pivots will be a challenge for Landsat RGB imagery, given it’s coarse resolution and the less resolving power for smaller boundaries. However, Figure 13 shows that in a scene where there are many fields in full cultivation, they are for the most part all accurately mapped. This implies that inaccuracies from misdetected small center pivots in a stage of pre-cultivation could be remedied by applying the model to multiple dates and merging the output results in order to capture center pivots in a stage of cultivation during the growing season. The results from the medium category are overall much more accurate, though even in this size category, Figure 11 shows that there are many false negatives due to many pivots being in a stage of pre-cultivation or showing half in cultivation and half pre-cultivation. Figure 11 also highlights another source of error in segmenting center pivots, where corner areas surrounding center pivots are also cultivated. While these regions are not annotated in the Nebraska dataset as belonging to part of the center pivot, they could either be irrigated along with the center pivot as a single field using an extension to the irrigation apparatus or separately. This has relevance for accurately estimating individual field water use, and highlights the difficulty in resolving individual units using satellite imagery. Center pivots that are composed of multiple sections will be hard to segment, given the limited amount of training data that contains similar representations. As with small center pivots in stages of pre cultivation, one approach to segmenting these could be to apply a model to multiple dates in a growing season and then merge the highest confidence detections in order to reduce the amount of false negatives.

Because the dataset contained 52,127 instances of center pivots, it was infeasible to examine each with respect to 32 Landsat scenes as an image reference. In some cases existing center pivots went unlabeled . This impacted both the model training, leading to lower performance due to greater within class variability and similarity between the center pivots category and background category,vertical grow shelf and less certain evaluation metrics, since the results were evaluated on an independent test subset of the reference dataset. Furthermore, center pivots as a detection target represent some of the most visually distinct agricultural features, and therefore it is expected that these models would perform more poorly on small agricultural fields that do not conform to such a consistent shape and range of sizes. Finally, the reference dataset for Nebraska used a broad interpretation of center pivots to include center pivots in multiple stages of cultivation . This led to a considerable amount of within class variability during training which impacted model performance. This limitation could be handled better by assigning more specific semantic categories to the center pivot labels based on greenness indices, in order to distinguish between different developmental stages. Or, the model produced from the original dataset could be applied for multiple dates throughout the seasons and results could be merged based on detection confidence in order to map pivots when they are at their most discernible, i.e. cultivated.In the original CSA model, members support a farm by paying in advance, and in return they receive a share of the farm’s produce; members also share in production risks, such as a low crop harvest following unfavorable weather. An important social invention in industrialized countries, Community Supported Agriculture addresses problems at the nexus of agriculture, environment and society. These include a decreasing proportion of the “food dollar” going to farmers, financial barriers for new farmers, large-scale scares from food borne illness, resource depletion and environmental degradation. Together with farmers markets, farm stands, U-picks and agritourism, CSAs constitute a “civic agriculture” that is re-embedding agricultural production in more sustainable social and ecological relationships, maintaining economic viability for small- and medium-scale farmers and fulfilling the non–farm-based population’s increasing desire to reconnect with their food . The first two CSAs in the United States formed in the mid-1980s on the East Coast . By 1994, there were 450 CSAs nationally , and by 2004 the number had nearly quadrupled to 1,700 . There were an estimated 3,637 CSAs in the United States by 2009 . This rapid expansion left us knowing little about CSA farmers and farms and raised questions about their social, economic and environmental characteristics. Knowing these features of CSAs would allow for more-precise policy interventions to support and extend these kinds of operations, and could inform more in-depth analyses, in addition to giving farmers and the public a better understanding of them.We conducted a study of CSAs in 25 counties in California’s Central Valley and its surrounding foothills — from Tehama in the north to Kern in the south, and Contra Costa in the west to Tuolomne to the east. The valley’s Mediterranean climate, combined with its irrigation infrastructure, fertile soil, early agrarian capitalism and technological innovation have made it world renowned for agricultural production . In addition to its agricultural focus, we chose this region because we wanted to learn about how CSAs were adapting to the unique context of the Central Valley. Many of the region’s social characteristics — relatively low incomes, high unemployment rates and conservative politics — differ from those in other regions where CSAs are popular, such as the greater San Francisco Bay Area and Santa Cruz . An initial list was compiled from seven websites that list CSAs in the state: Biodynamic Farming and Gardening Association, California Certified Organic Farmers, Community Alliance with Family Farmers, Eat Well Guide, LocalHarvest, the Robyn Van En Center and Rodale Institute. Of the 276 CSAs that we found, 101 were in our study area. We contacted them by e-mail and phone. It became evident that some did not correspond, even loosely, to the definition of a CSA in which members share risks with the farm and pay in advance for a full season of shares. As the study progressed, we revised our definition of a CSA to mean an operation that is farm based and makes regular direct sales of local farm goods to member households. We removed some CSAs that did not meet the revised definition, based on operation descriptions on their websites or details provided by phone or e-mail if a website was not available. Some interviews that we had already completed could not be used for our analysis because the operations did not meet the revised definition. As the study progressed, we augmented the initial list with snowball sampling by asking participating farmers about other CSAs, which added 21 CSAs. Of these 122 farms, 28 were no longer operating as CSAs, seven turned out to be CSA contributors without primary responsibility for shares and 13 did not meet our revised CSA definition. We called the 28 CSAs no longer operating “ghost CSAs” because of their continued presence on online lists.

Posted on

The federal government has had an agricultural guest worker program for most of the past century

In response to the plaintiff’s standing argument, the defendant food company or grocery store can reply that the WTO Agreements specifically contemplate allowing compensation and retaliation for injuries inflicted upon private commercial interests. Defendant would argue that it is only presenting a defense based on explicit WTO language. Moreover, defendant would argue that, if the doctrine of standing blocks the raising of the WTO-based defenses, it would face administrative actions or consumer damages under a law that very likely violates either the SPS Agreement or the TBT Agreement. Defendants would argue that such a result is unjust and legally indefensible because nobody should be held legally accountable under a law that may be itself demonstrably invalid.The hired workers who do most of California’s farm work are primarily employed to produce crops, which accounted for almost three-fourths of the state’s $37 billion in farm sales in 2010. Within crops, most hired workers are employed to produce the so-called FVH commodities, fruits and nuts , vegetables and melons , and horticultural specialties such as flowers and nursery products that generated almost 90 percent of California’s crop sales and two-thirds of the state’s total farm sales. California requires all employers with $100 or more in quarterly wages to pay unemployment insurance taxes on worker earnings. Over the past two decades, UI data show stable average annual agricultural employment of just under 400,000, but crop support employment, primarily employees of farm labor contractors, recently surpassed the number of workers hired directly by crop employers . Average annual employment is a measure of the number of year-round job slots, not the number of farm workers, because of seasonal peaks and worker turnover. For example, agricultural employment averaged 389,500 in 2011, with a peak of 452,800 in June and a second peak of 449,600 in September; the low was 311,700 in January. The employment peak-trough ratio was almost 1.5,nft vertical farming meaning that 50 percent more workers were employed in June than in January. According to Khan et al.,an analysis of individual social security numbers reported by agricultural establishments in the 1990s found almost three individuals for each year-round farm job, suggesting 1.1 million unique farm workers.

Even though the analysis removed SSNs reported by 50 or more employers in one year and jobs that generated less than $1 or more than $75,000 in quarterly earnings, some observers believe that UI data may exaggerate the number of unique farm workers. If the three-to-one ratio of workers to year-round jobs is correct, there are about 1.1 million farm workers; at two-to-one, there are almost 800,000. For most workers, farm work is a job rather than a career. A conservative estimate is that at least 10 percent of farm workers leave the farm work force each year, so that farmers rely on an influx of new entrants to replace those who leave for non-farm jobs or return to Mexico. If California has a million unique farm workers, this means 100,000 newcomers are required to replace those who exit; for the 2.5 million unique hired farm workers across the United States, 250,000 newcomers a year are required. Mexico-U.S. migration has slowed, providing fewer new entrants to replace farm workers who exit. About 10 percent of the people born in Mexico have moved to the United States, some 12 million, and 30 percent of the 40 million foreign-born U.S. residents were born in Mexico, making Mexico the largest source of U.S. immigrants. Mexican born U.S. residents have spread throughout the United States, but almost 60 percent live in California and Texas. Between 2005 and 2010, the Pew Hispanic Center estimated zero net Mexico-U.S. migration; that is, almost 1.4 million Mexicans moved to the United States over this five-year period and 1.4 million Mexicans moved to Mexico . Many of those who returned to Mexico were deported, and some took their U.S.-born children with them. There are still Mexicans moving to the United States, but returns to Mexico outnumbered new Mexican entrants to the United States by four to one in recent years. Reasons for the slowdown in Mexico-US migration include high U.S. unemployment, border violence and more fences and agents that raise smuggling costs and risks, and improving conditions in Mexico. California agriculture is feeling the effects of slowing Mexico-U.S. migration because of its revolving-door labor market, which relies on newcomers from abroad to replace workers who exit.

If Mexico-U.S. migration does not increase with the expected U.S. economic recovery, where will California farmers get replacement farm workers? The answer depends on immigration policy: will currently unauthorized farm workers be legalized and required to continue to work in agriculture or will replacement workers be guest workers from abroad? The current H-2A program certified 7,200 U.S. farmers to fill over 90,000 farm jobs with guest workers in FY11, including 250 California farmers to fill 3,000 farm jobs. The H-2A program requires farm employers to try to recruit U.S. workers under federal and state supervision, offer guest workers free housing, and pay them a super minimum wage called the Adverse Effect Wage Rate of $10.24 an hour in California in 2012. California farm employers assert that the H-2A program is too cumbersome and bureaucratic because of the state’s diverse and perishable crops. They urge three major employer-friendly changes. First, employers would like attestation to replace certification, meaning that employers would attest or assert that they tried and failed to recruit U.S. workers while offering appropriate wages, and their attestations would allow them to recruit and employ guest workers. Second, farm employers would like to offer housing vouchers worth $200 to $300 a month instead of free housing, adding $1 to $2 an hour to current wages. Third, to help offset the cost of housing vouchers, the AEWR would be rolled back by $1 or more an hour and studied to determine how well it achieves its goal of protecting U.S. workers from any wage-depressing effects of guest workers. The Clinton Administration blocked efforts to enact these employer-friendly changes to the H-2A program during the 1990s. However, in December 2000, farm employer and worker advocates negotiated the Agricultural Job Opportunity Benefits and Security Act , which would legalize currently unauthorized farm workers and make these three employer-friendly changes to the H-2A program. They hoped that Congress would enact AgJOBS in the waning days of the Clinton Administration, but AgJOBS was blocked by those opposed to “amnesty.” Most farm employers and worker advocates continue to urge enactment of the 12-year old AgJOBS bill.

Senator Dianne Feinstein introduced a version in 2009 that would grant Blue Card temporary legal status to up to 1.35 million unauthorized foreigners who did at least 150 days or 863 hours of farm work in the 24-month period ending December 31, 2008. If Blue Card holders continued to do farm work over the next three to five years, they and their families could become legal immigrants. AgJOBS’s employer-friendly changes to the H-2A program include the Big 3 desired by farm employers: attestation, housing vouchers, and a reduced AEWR. Farm employers and workers today are in a period of uncertainty. There is unlikely to be any immigration reform that includes earned legalization or amnesty in 2012, and legalization may face continued obstacles in a Republican controlled House in the next Congress. However, employer-friendly changes to the H-2A program or a new guest worker program could accompany a federal mandate that all employers use E-Verify to check the legal status of newly hired workers. Representatives who favor mandatory E-Verify have proposed new guest worker programs administered by USDA rather than the Department of Labor that include attestation, reduced or no housing requirements,indoor vertical farming and lower minimum wages without legalizing currently unauthorized workers. Just months after tainted cantaloupes caused the deadliest U.S. outbreak of food borne illness in a century, 270 consumers were sickened and three killed this summer by cantaloupes carrying Salmonella. The FDA has linked the contaminated cantaloupe to unsanitary conditions at Chamberlain Farms of Owensville, Indiana. Like the Holly, Colorado farm implicated in the deadly Listeria outbreak in cantaloupes last fall, Chamberlain is a relatively small grower in a region with no obvious comparative advantage in cantaloupe production. Moreover, climatic conditions predispose the area to bacterial contamination. Recent outbreaks of food borne illness and an increase in infections from most pathogens monitored by the CDC in 2011 have occurred alongside growth in consumer demand for locally sourced produce that has led even multinational discount retailers like Walmart to seek out local suppliers. Growing reliance on local production sacrifices the benefits of specialization according to comparative advantage and scale economies that more concentrated production affords.

In addition, dependency on smaller, local producers may come at the expense of food safety.Voluntary and regulation-induced investments in food safety, including process control, inspection, and traceability, often include fixed-cost components and “lumpiness” that gives an advantage to larger firms, which can spread the costs over larger quantities of production. High fixed costs of food safety can cause small firms to exit. Food safety processes and technologies appear to impose higher costs per unit of output on small operations. This fear was articulated during the 1998 implementation of pathogen reduction and Hazard Analysis and Control Point systems in the red meat- and poultry-slaughter industries. More recently, small farms successfully lobbied for exceptions to similar rules for produce growers under the Food Security Modernization Act, citing concern that the higher cost of regulatory compliance to small operations would force local and direct to-consumer operations to exit. Though there are few empirical analyses of the cost of food safety investments among produce growers, studies of the 1998 regulatory changes in the meat packing industry have documented a considerable disadvantage to small firms. An ex ante analysis of the planned regulations, for instance, estimated the costs to small beef slaughter plants as a share of the value of shipments would be 100 times greater than the cost to large firms. Many of the costs associated with the regulations, including monitoring, record keeping, and sanitation equipment are fixed, at least over a range of outputs, so that the cost per unit of production decreases in the scale of production. Moreover, the Economic Research Service of the USDA estimated that smaller plants had variable and fixed costs of regulatory compliance that were three and six times higher, respectively, than those of larger firms. Output inspection costs, for instance, may vary proportionally to output over a range of outputs, so that inspection technology exhibits constant returns to scale and does not disadvantage small firms. Plant inspection costs, however, are largely fixed, so that average plant inspection costs decline with output, favoring larger operations. Large firms, for instance, may find it optimal to operate their own testing laboratories and, thus, exploit increasing returns to scale. Small firms, however, are likely to contract testing out to third parties and, consequently, bear a relatively high cost per unit output. An ERS survey of the meat- and poultry-processing sectors concluded that testing costs, in particular, were higher for smaller firms. Likewise, the cost of regulatory enforcement exhibits increasing returns to scale because of the fixed costs associated with inspection of a firm’s facilities and procedures. These include travel costs and testing that increase in the number of facilities but not in the level of output. Thus, given a fixed enforcement budget, a decline in industry concentration resulting from demands for local production is likely to lower the level of safety.Food producers face food safety costs imposed by regulation, but may also make voluntary investments in pathogen control in order to meet contractual demands of buyers, reduce the risk of crop losses, avoid liability judgments, and protect brand value. Optimal private investment in exante food safety mechanisms equates the marginal expected reduction in damage and liability costs to the marginal cost of private risk reduction. This condition is likely to induce disproportionately greater investments from large operations than from small ones. The USDA estimated that as much as two-thirds of the reduction in Salmonella contamination in meat and poultry facilities is attributable to private voluntary investments rather than regulation-induced effort. When outbreaks of food borne illness occur, those firms responsible for contamination are typically obligated to recall all or some of the affected product.