1 April 2017 in Chemosphere
20 January 2023 in Environmental Research Letters
Ordinary least squares linear regression (LR) has long been a popular choice among researchers interested in using historical data for estimating crop yield response to climate change. Today, the rapidly growing field of machine learning (ML) offers a wide range of advanced statistical tools that are increasingly being used for more accurate estimates of this relationship. This study compares LR to a popular ML technique called boosted regression trees (BRTs). We find that BRTs provide a significantly better prediction accuracy compared to various LR specifications, including those fitting quadratic and piece-wise linear functions. BRTs are also able to identify break points where the relationship between climate and yield undergoes significant shifts (for example, increasing yields with precipitation followed by a plateauing of the relationship beyond a certain point). Tests we performed with synthetically simulated climate and crop yield data showed that BRTs can automatically account for not only spatial variation in climate–yield relationships, but also interactions between different variables that affect crop yields. We then used both statistical techniques to estimate the influence of historical climate change on rice, wheat, and pearl millet in India. BRTs predicted a considerably smaller negative impact compared to LR. This may be an artifact of BRTs conflating time and climate variables, signaling a potential weakness of models with excessively flexible functional forms for inferring climate impacts on agriculture. Our findings thus suggest caution while interpreting the results from single-model analyses, especially in regions with highly varied climate and agricultural practices.
12 October 2020 in Nature Sustainability
Water scarcity is a global issue that disproportionately affects small-scale farmers in low- and middle-income countries (LMICs). Through geospatial analysis, we estimated that less than 37% of small-scale farms probably have irrigation in water scarce regions across LMICs, compared with 42% of non-small-scale farms. Through a literature synthesis assisted by machine learning, we then systematically mapped the existing research for on-farm interventions that improve the incomes or yields of small-scale farmers in water scarce regions. We mapped over 888 on-farm interventions used to combat water scarcity from 560 publications and showed a research bias towards yields rather than livelihoods. We found gaps in evidence for many commonly proposed solutions, including livestock management, digital technology and solutions to protect natural resources at the farm-level, such as buffer strips. Our findings can be used to set a funding agenda for research on the geographies that are most at risk of water scarcity and the interventions that most lack evidence.
21 September 2023 in Environmental Research Letters
The 2022 Indian heatwave impacted key geographies of wheat production in northwestern and central India. It coincided with this crop’s harvest season in the region, and is expected to have considerably reduced national wheat production. Here we provide spatially disaggregated estimates of the likely impact on wheat yield using statistical relationships derived from historical climate and wheat yield data. Compared to a normal year (median of 1992–2021 climate), national wheat production in 2022 is expected to have fallen by 4.5%; some regions may have experienced wheat yield losses of up to 15%. Our analysis can also be analyzed as a proof of concept for using current and forecasted weather data to estimate real-time impact of short-term weather variability on end-of-season crop yields, even while the crop season is ongoing.
1 December 2018 in Journal of Water Process Engineering
18 July 2022 in Climatic Change