Answers are grounded in this paper and cite their sources.
Drought is among the most common but least understood phenomena that affect an increasing number of people in the context of climate change. To understand underlying drought dynamics affecting the local agricultural production in Botswana, a broad database comprising climatic and remote-sensing data together with socioeconomic indicators was set up. A data science approach that includes statistical and machine learning methods was chosen to retrieve information applicable in a drought early-warning system. The aim of the study was to examine how data science can contribute to the understanding of drought risk through the integration of various data sources. Different regression models (including linear and OLS) were applied. Naïve Bayes classification and Random Forest regression were included, as was a change point analysis. The impacts of two variables in particular, the Standardized Precipitation Index (SPI) and the Southern Oscillation Index (SOI), on crop productivity could be observed, highlighting possible national and regional thresholds. Further development of the early warning system, including validation, should be accompanied by ground-truth information and work with local partners.
No takes yet. Share an insight, caveat, or question.
H. Kemper (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: