Combining satellite imagery with machine learning (SIML) has the potential to address global challenges by remotely estimating socioeconomic and environmental conditions in data-poor regions, yet the resource requirements of SIML limit its accessibility and use. We show that a single encoding of satellite imagery can generalize across diverse prediction tasks (e.g., forest cover, house price, road length). Our method achieves accuracy competitive with deep neural networks at orders of magnitude lower computational cost, scales globally, delivers label super-resolution predictions, and facilitates characterizations of uncertainty. Since image encodings are shared across tasks, they can be centrally computed and distributed to unlimited researchers, who need only fit a linear regression to their own ground truth data in order to achieve state-of-the-art SIML performance.

Significance: There is interest in whether COVID-19 cases respond to environmental conditions. If an effect is present, seasonal changes in local environmental conditions could alter the global spatial pattern of COVID-19 and inform local public health responses. Using a comprehensive global dataset of daily COVID-19 cases and local environmental conditions, we find that increased daily ultraviolet (UV) radiation lowers the cumulative daily growth rate of COVID-19 cases over the subsequent 2.5 wk. Although statistically significant, the implied influence of UV seasonality is modest relative to social distancing policies. Temperature and specific humidity cumulative effects are not statistically significant, and total COVID-19 seasonality remains to be established because of uncertainty in the net effects from seasonally varying environmental variables.

Year-to-year variance of agricultural productivity is an important determinant of food security. Previous global analyses described increased yield volatility from warming, but it has become increasingly clear that changes in water availability are also a key determinant of yields. Here, we provide the first global quantification of climate change impacts on maize, soybean, and sorghum yield variance due to changes in temperature and soil moisture. Pairing an empirical crop model with CMIP6 simulations indicates that changes in temperature and soil moisture increase interannual weather-induced yield variance by 7 to 19 % ° C across crops. This increase is driven roughly equally by previously quantified increased temperature stress variance, as well as by increased covariance between temperature and soil moisture stresses. Results using a simple land surface model are consistent with those using CMIP6 simulations in indicating that this increased covariance is driven by warmer air more quickly drying soils and by dry soils promoting greater warming.

Understanding the impact of climate change on human migration is critical for policymakers. Yet climate change can both incentivize people to migrate and reduce their ability to move, making its effect on human migration ambiguous. We propose an approach to studying migration that combines causal inference methods with cross-validation techniques to reliably estimate effects of weather on migration within and across borders. This approach highlights the key role of migrant demographics in the weather-migration relationship. We show that allowing weather effects to differ by age and education improves out-of-sample performance by a factor of five or more compared with a homogeneous effect. Demographic heterogeneity is critical in explaining this discrepancy. Projections based on our empirical estimates indicate that the effects of climate change on future cross-border migration will be an order of magnitude larger for most demographics than the average effect, but differing responses across groups largely offset one another.