Monitoring drought impacts on crop productivity of the U.S. Midwest with solar-induced fluorescence: GOSIF outperforms GOME-2 SIF and MODIS NDVI, EVI, and NIRv

The frequency and severity of drought are increasing in the context of global warming. Elucidating the responses of crop productivity to drought is essential for informing agricultural management and ensuring food security. Here we used satellite-derived solar-induced chlorophyll fluorescence (SIF) data and vegetation indices to evaluate the impacts of the 2012 drought on crop productivity in the U.S. Midwest. We used SIF from the global, OCO-2 based SIF product (GOSIF; 0.05°, 8-day), GOME-2 SIF product (0.5°, monthly), and three MODIS-derived vegetation indices (NDVI, EVI, and NIRv). We compared the seasonal cycles and anomalies of SIF and VIs from 2008 to 2018. We also examined to what extent these proxies could capture the variations of gross primary production (GPP) for eddy covariance flux sites. SIF and VIs were able to capture the seasonal cycle in drought and normal years. SIF better captured the photosynthesis changes due to water and heat stresses than vegetation indices. In particular, GOSIF data with the finer spatio-temporal resolution was a good monitor of crop response to drought. Crop yield decreased by 25% in the 2012 drought relative to the multi-year mean, while GOSIF, NDVI, EVI, and NIRv reduced by 22%, 4%, 10%, and 8%, respectively. GOSIF had the strongest relationship with crop yield (R2 = 0.91), followed by NIRv (R2 = 0.89), EVI (R2 = 0.68) and NDVI (R2 = 0.48). Compared to the crop yield data, the mean difference of the yield estimates based on GOSIF, EVI, and NIRv were 379.32, 328.43, and 503.67 kg/ha, respectively. For both corn and soybeans, yield anomalies were better correlated with GOSIF anomalies than with NIRv and EVI anomalies. Our study demonstrated that SIF with finer spatio-temporal resolution has great potential for monitoring the responses of crop productivity to drought.