Performance of multivariate and multiscalar drought indices in identifying impacts on crop production

The assessment of drought characteristics often depends on drought indices, geographic location, hydro-climatic condition, and timescale. In this study, we examined the spatiotemporal characteristics of drought events using the Standardized Precipitation Evapotranspiration Index (SPEI) and the Multivariate Standardized Drought Index (MSDI). We developed a novel framework using hydro-climatic variables from a high-resolution process-based hydrologic model to understand factors that alter drought indices at various timescales, and their impact on crop yields in a large agricultural region of western Canada. These indices were used to quantify droughts for each month of the year by examining 1-12-month drought timescales in 2255 sub-basins, simulated in 17 main river basins across Alberta, for 1981-2017. Temporal variations of the Standardized Yield Residuals Series (SYRS) of three major cereal crops (spring wheat, barley, and canola) were analysed for 1981-2017. Drought characteristics resulting from two indices varied due to differences in the input variables and timescales. The MSDI-based results showed more frequent droughts during the fall and winter for shorter timescales, while the SPEI-based results showed more during spring and summer. Comparing drought frequencies at the decadal scale, we found more droughts during 1996-2005 than during 1986-1995 and 2006-2015. The spatial evolution of drought events based on the MSDI showed more sub-basins with increasing dryness during the study period than did results from the SPEI. The relationship between detrended drought indices and the SYRS varied depending on timescale, geographic location, and growth stage of crops. Overall, both indices performed similarly for agricultural impact assessment; however, the MSDI performed better early in the growing season for wheat and barley, indicating high crop production sensitivity to soil moisture deficiency.