Wheat is the most important Ethiopian crop, and rust one of its greatest antagonists. There is a need for cheap and scalable rust monitoring in the developing world, but existing methods employ costly data collection techniques. We introduce a scalable, accurate, and inexpensive method for tracking outbreaks with publicly available remote sensing data. Our approach improves existing techniques in two ways. First, we forgo the spectral features employed by the remote sensing community in favor of automatically learned features generated by Convolutional and Long Short-Term Memory Networks. Second, we aggregate data into larger geospatial regions. We evaluate our approach on nine years of agricultural outcomes, show that it outperforms competing techniques, and demonstrate its predictive foresight. This is a promising new direction in crop disease monitoring, one that has the potential to grow more powerful with time.

Eradicating hunger and malnutrition is a key development goal of the twenty first century. This paper addresses the problem of optimally identifying seed varieties to reliably increase crop yield within a risk-sensitive decision making framework. Specifically, a novel hierarchical machine learning mechanism for predicting crop yield (the yield of different seed varieties of the same crop) is introduced. This prediction mechanism is then integrated with a weather forecasting model and three different approaches for decision making under uncertainty to select seed varieties for planting so as to balance yield maximization and risk. The model was applied to the problem of soybean variety selection given in the 2016 Syngenta Crop Challenge. The prediction model achieved a median absolute error of 235 kg/ha and thus provides good estimates for input into the decision models. The decision models identified the selection of soybean varieties that appropriately balance yield and risk as a function of the farmer’s risk aversion level. More generally, the models can support farmers in decision making about which seed varieties to plant.

The practice of planting winter cover crops has seen renewed interest as a solution to environmental issues with the modern maize- and soybean-dominated row crop production system of the US Midwest. We examine whether cover cropping patterns can be assessed at scale using publicly available satellite data, creating a classifier with 91.5% accuracy (.68 kappa). We then use this classifier to examine spatial and temporal trends in cover crop occurrence on maize and soybean fields in the Midwest since 2008, finding that despite increased talk about and funding for cover crops as well as a 94% increase in cover crop acres planted from 2008–2016, increases in winter vegetation have been more modest. Finally, we combine cover cropping with satellite-predicted yields, finding that cover crops are associated with low relative maize and soybean production and poor soil quality, consistent with farmers adopting the practice on fields most in need of purported cover crop benefits. When controlling for invariant soil quality using a panel regression model, we find modest benefits of cover cropping, with average yield increases of 0.65% for maize and 0.35% for soybean. Given these slight impacts on yields, greater incentives or reduced costs of implementation are needed to increase adoption of this practice for the majority of maize and soybean acres in the US.

Accurate prediction of crop yields in developing countries in advance of harvest time is central to preventing famine, improving food security, and sustainable development of agriculture. Existing techniques are expensive and difficult to scale as they require locally collected survey data. Approaches utilizing remotely sensed data, such as satellite imagery, potentially provide a cheap, equally effective alternative. Our work shows promising results in predicting soybean crop yields in Argentina using deep learning techniques. We also achieve satisfactory results with a transfer learning approach to predict Brazil soybean harvests with a smaller amount of data. The motivation for transfer learning is that the success of deep learning models is largely dependent on abundant ground truth training data. Successful crop yield prediction with deep learning in regions with little training data relies on the ability to fine-tune pre-trained models.

Millions of people worldwide are absent from their country’s census. Accurate, current, and granular population metrics are critical to improving government allocation of resources, to measuring disease control, to responding to natural disasters, and to studying any aspect of human life in these communities. Satellite imagery can provide sufficient information to build a population map without the cost and time of a government census. We present two Convolutional Neural Network (CNN) architectures which efficiently and effectively combine satellite imagery inputs from multiple sources to accurately predict the population density of a region. In this paper, we use satellite imagery from rural villages in India and population labels from the 2011 SECC census. Our best model achieves better performance than previous papers as well as LandScan, a community standard for global population distribution.

Crop type mapping at the field level is necessary for a variety of applications in agricultural monitoring and food security. As remote sensing imagery continues to increase in spatial and temporal resolution, it is becoming an increasingly powerful raw input from which to create crop type maps. Still, automated crop type mapping remains constrained by a lack of field-level crop labels for training supervised classification models. In this study, we explore the use of random forests transferred across geographic distance and time and unsupervised methods in conjunction with aggregate crop statistics for crop type mapping in the US Midwest, where we simulated the label-poor setting by depriving the models of labels in various states and years. We validated our methodology using available 30 m spatial resolution crop type labels from the US Department of Agriculture's Cropland Data Layer (CDL). Using Google Earth Engine, we computed Fourier transforms (or harmonic regressions) on the time series of Landsat Surface Reflectance and derived vegetation indices, and extracted the coefficients as features for machine learning models. We found that random forests trained on regions and years similar in growing degree days (GDD) transfer to the target region with accuracies consistently exceeding 80%. Accuracies decrease as differences in GDD expand. Unsupervised Gaussian mixture models (GMM) with class labels derived using county-level crop statistics classify crops less consistently but require no field-level labels for training. GMM achieves over 85% accuracy in states with low crop diversity (Illinois, Iowa, Indiana, Nebraska), but performs sometimes no better than random when high crop diversity interferes with clustering (North Dakota, South Dakota, Wisconsin, Michigan). Under the appropriate conditions, these methods offer options for field-resolution crop type mapping in regions around the world with few or no ground labels.

Low-intensity tillage has become more popular among farmers in the United States and many other regions. However, accurate data on when and where low-intensity tillage methods are being used remain scarce, and this scarcity impedes understanding of the factors affecting the adoption and the agronomic or environmental impacts of these practices. In this study, we used composites of satellite imagery from Landsat 5, 7, and 8, and Sentinel-1 in combination with producer data from about 5900 georeferenced fields to train a random forest classifier and generate annual large-scale maps of tillage intensity from 2005 to 2016. We tested different combinations of hyper-parameters using cross-validation, splitting the training and testing data alternatively by field, year, and state to assess the influence of clustering on validation results and evaluate the generalizability of the classification model. We found that the best model was able to map tillage practices across the entire North Central US region at 30 m-resolution with accuracies spanning between 75% and 79%, depending on the validation approach. We also found that although Sentinel-1 provides an independent measure that should be sensitive to surface moisture and roughness, it currently adds relatively little to classification performance beyond what is possible with Landsat. When aggregated to the state level, the satellite estimates of percentage low- and high-intensity tillage agreed well with a USDA survey on tillage practices in 2006 (R2 = 0.55). The satellite data also revealed clear increases in low-intensity tillage area for most counties in the past decade. Overall, the ability to accurately map spatial and temporal patterns in tillage should facilitate further study of this important practice in the United States, as well as other regions with fewer survey-based estimates.