Characterizing variability in recent drought trends across Central America using different observational data sets

Smallholder farmers across Central America are dependent on seasonal precipitation patterns for rainfed agriculture. Recently, the region has experienced prolonged droughts and shifts in precipitation patterns, dramatically increasing food insecurity. As on-ground observational networks are sparse, we seek to quantify recent changes in the timing and intensity of recent precipitation and drought patterns across the region using established indices, such as Standardized Precipitation Index and Standardized Precipitation-Evapotranspiration Index , and drought indicators identified by smallholder farmers in the region. We use several different gridded observational data sets that span the past four decades, including a reanalysis product (ERA5), a reanalysis product adjusted using satellite and gauge observations (MERRA-2), and a combination of satellite and gauge observations (CHIRPS V2). The sensitivity of the trends and their statistical significance are compared across datasets. Particular focus is on the Mesoamerican dry corridor, a tropical dry zone extending along the Pacific side of Central America. This area features a bi-modal rainy season with an intervening dry period of weeks to over a month, often referred to as the mid-summer drought, the characteristics of which define planting and harvesting of crops. We quantify recent changes in the timing, onset, duration, and intensity of the mid-summer drought. Our findings show that while temperatures have significantly increased throughout the region, changes in the direction and magnitude of precipitation are highly variable and exhibit a large sensitivity to the dataset used to quantify changes. By understanding better which metrics are less sensitive to observational data source, this will help reduce uncertainty when extending this analysis to future drought projections.