Crop type mapping at the field level is critical for a variety of applications in agricultural monitoring, and satellite imagery is becoming an increasingly abundant and useful raw input from which to create crop type maps. Still, in many regions crop type mapping with satellite data remains constrained by a scarcity of field-level crop labels for training supervised classification models. When training data is not available in one region, classifiers trained in similar regions can be transferred, but shifts in the distribution of crop types as well as transformations of the features between regions lead to reduced classification accuracy. We present a methodology that uses aggregate-level crop statistics to correct the classifier by accounting for these two types of shifts. To adjust for shifts in the crop type composition we present a scheme for properly reweighting the posterior probabilities of each class that are output by the classifier. To adjust for shifts in features we propose a method to estimate and remove linear shifts in the mean feature vector. We demonstrate that this methodology leads to substantial improvements in overall classification accuracy when using Linear Discriminant Analysis (LDA) to map crop types in Occitanie, France and in Western Province, Kenya. When using LDA as our base classifier, we found that in France our methodology led to percent reductions in misclassifications ranging from 2.8% to 42.2% (mean = 21.9%) over eleven different training departments, and in Kenya the percent reductions in misclassification were 6.6%, 28.4%, and 42.7% for three training regions. While our methodology was statistically motivated by the LDA classifier, it can be applied to any type of classifier. As an example, we demonstrate its successful application to improve a Random Forest classifier.
Stanford health law experts Michelle Mello and David Studdert discuss the ongoing pandemic, proof of vaccination “passports” at the state and federal levels, and a July 19 ruling that Indiana University could require that its students be vaccinated.
Crop productivity is potentially affected by several air pollutants, although these are usually studied in isolation. A significant challenge to understanding the effects of multiple pollutants in many regions is the dearth of air quality data near agricultural fields. Here we empirically estimate the effect of four key pollutants (ozone (O3), particulate matter (PM), sulfur dioxide (SO2), and nitrogen dioxide (NO2)) on maize and soybean yields in the United States using a combination of administrative data and satellite-derived yield estimates. We identify clear negative effects of exposure to O3, PM, and SO2 in both crops, using yields measured in the vicinity of monitoring stations. We also show that while stations measuring NO2 are too sparse to reliably estimate a yield effect, the strong gradient of NO2 concentrations near power plants allows us to more precisely estimate NO2 effects using satellite measured yield gradients. The presence of some powerplants that turn on and others that shut down during the study period are particularly useful for attributing yield gradients to pollution. We estimate that total yield losses from these pollutants averaged roughly 5% for both maize and soybean over the past two decades. While all four pollutants have statistically significant effects, PM and NO2 appear more damaging to crops at current levels than O3 and SO2. Finally, we find that the significant improvement in air quality since 1999 has halved the impact of poor air quality on major crops and contributed to yield increases that represent roughly 20% of overall yield gains over that period.
Quantification of the sector-specific financial impacts of historical global warming represents a critical gap in climate change impacts assessment. The multiple decades of county-level data available from the U.S. crop insurance program – which collectively represent aggregate damages to the agricultural sector largely borne by U.S. taxpayers – present a unique opportunity to close this gap. Using econometric analysis in combination with observed and simulated changes in county-level temperature, we show that global warming has already contributed substantially to rising crop insurance losses in the U.S. For example, we estimate that county-level temperature trends have contributed $US2017 23.9 billion – or 17% – of the national-level crop insurance losses over the 1991-2017 period. Further, we estimate that observed warming contributed approximately one third of total losses in the most costly single year (2012). In addition, analyses of a large suite of global climate model simulations yield very high confidence that anthropogenic climate forcing has increased U.S. crop insurance losses. These sector-specific estimates provide important quantitative information about the financial costs of the global warming that has already occurred (including the costs of individual extreme events), as well as the economic value of mitigation and/or adaptation options.