Inspired by concerns of the effects of a warming climate, drought variation and its impacts have gained much attention in China. Arguments about China's drought persist and little work has utilized agricultural drought survey area to evaluate the impact of natural drought on agriculture. Based on a newly revised self‐calibrating Palmer Drought Severity Index (PDSI) model driven with air‐relative‐humidity‐based two‐source (ARTS) E₀ (PDSI_ARTS; Yan et al., 2014), spatial and temporal variations of drought were analyzed for 1982–2011 in China, which indicates that there was nonsignificant change of drought over this interval but with an extreme drought event happened in 2000–2001. However, using air temperature (T_a)‐based Thornthwaite potential evaporation (E_{P_Th}) and Penman‐Monteith potential evaporation (E_{P_PM}) to drive the PDSI model, their corresponding PDSI_Th and PDSI_PM all gave a significant drying trend for 1982–2011. This suggests that PDSI model was sensitive to E_P parameterization in China. Annual drought‐covered area from agriculture survey was initially adopted to evaluate impact of PDSI drought on agriculture in China during 1982–2011. The results indicate that PDSI_ARTS drought area (defined as PDSI_ARTS < −0.5) correlated well with the agriculture drought‐covered area and PDSI_ARTS successfully detected the extreme agriculture drought in 2000–2001 during 1982–2011, i.e., climate factors dominated the interannual changes of agriculture drought area, while PDSI_Th and PDSI_PM drought areas had no relationship with the agriculture drought‐covered area and overestimated the uptrend of agriculture drought This study highlights the importance of coupling PDSI with drought survey data in evaluating the impact of natural drought on agriculture.

Air temperature, T _a, with high spatial and temporal resolution is desired for global change, agricultural disaster, land surface studies, and modelling applications. A statistical algorithm for Moderate Resolution Imaging Spectroradiometer (MODIS) data is developed for daytime T _a retrievals over east China at a resolution of 0.05° × 0.05°. The approach first applies a statistical regression of the first guess, i.e. the preliminary estimate, of T _a to MODIS 11 μm and 12 μm brightness temperature (T _11μm and T _12μm) and site data (longitude, latitude and altitude) for east China. Then the first guess of T _a is further corrected with a series of bias equations for different latitude zones in east China. Further quantitative validation with measured T _a using 335 synoptic weather stations for the whole of 2006 indicates that the algorithm performs well with overall statistics of R = 0.96, RMSE = 3.23°C, and bias = −0.09°C. 75% of the estimated T _a is within 3°C of the actual T _a and 92% of the estimated T _a is within 5°C of the actual T _a. This bias correction algorithm can be applied to other geostationary and sun-synchronous satellite instruments for T _a retrieval.