Impacts of multi-timescale SPEI and SMDI variations on winter wheat yields

Drought is one of the main factors that impacts crop yields. However, determining how different drought types affect winter wheat yields is difficult due to the lack of observation data. This work aimed to investigate the impacts of multi-timescale droughts on winter wheat yields. The winter wheat yields during 1981–2015 were simulated by the DSSAT-CERES-Wheat model. We analyzed the drought characteristics based on the standardized precipitation evapotranspiration index (SPEI) and soil moisture deficit index (SMDI) at timescales of 1 to 9 months at 108 sites in China. The modified Mann-Kendall (MMK) method was used to test the tendency of 1-, 3- and 6-month SPEI, SMDI and winter wheat yields. Pearson correlation analysis was used to explore the relationship between winter wheat yields and SPEI/SMDI at different timescales. The results showed that, DSSAT-CERES-Wheat generally performed well in simulating winter wheat anthesis date, maturity date and yields (0.64 < R2 < 0.97, where R2 is determination of coefficient). The dry or wet status for the 1- to 9-month timescales of SPEI and SMDI were generally consistent in the three subregions. Seasonal drought occurred more frequently in the Huang-Huai-Hai Plain than in the other two subregions. The 4-month SPEI and 1-month SMDI at the 0–10 cm depth affected the winter wheat yield more during the jointing to milk stages. For yields vs. 4-month SPEI, the number of stations with Pearson correlation coefficient r > 0.4 in March, April and May was 29, 35 and 23, respectively. For yields vs. 1-month SMDI, the number of stations with r > 0.4 in March, April and May was 16, 23 and 33, respectively. SMDI and SPEI explained more than 14% and less than 2% of the yield variability, respectively. Closer relations between SMDI and winter wheat yields were observed. This study provides useful references for preventing agricultural drought.