Abstract:
Abstract: Agricultural drought caused by soil water deficit exerts great influence on ecosystems and growth of crops. Accurate monitoring and detection of spatio-temporal characteristics of agricultural drought are meaningful for food security. However, agricultural drought is often characterized by current water demand-supply conditions, without considering the rarity of drought event in the historical period. In order to overcome the limitations of using crop water deficit indicator or dryness anomaly indicator only, an integrated evapotranspiration deficit index (IEDI) combining water deficit and dryness probability was proposed in this paper. To calculate the IEDI, crop water stress index (CWSI) ranging from 0 to 1 was calculated firstly based on actual evapotranspiration and potential evapotranspiration by remote sensing to reflect the actual level of crop water stress. Secondly, the drought event rarity index (RI) was derived on the basis of CWSI to reflect how often the current water stress occurred during the study period. The RI quantified the probability of the occurrence of an unusually dry event compared to normal state during the study period, and it was obtained by standardizing the cumulative density via the median of CWSI values. The calculation was based on the assumption that the statistical structure of CWSI follows Beta distribution and the median of CWSI time series represents normal water deficit state. In order to get an equal-interval value ranging from 0 to 1 quantifying how dry this crop water stress is compared to usual state, the RI was further derived using an empirical fitting method based on the standardized index. Finally, the proposed IEDI was derived, which was the square root of the product of CWSI and RI. On the basis of IEDI, temporal variations of agricultural drought in Northeast China, which is potentially threatened by climate extreme events, were analyzed. The impacts of meteorological factors on agricultural drought and the impacts of agricultural drought occurring period on grain yield were further investigated using the frequency analysis and the linear regression approach. Results showed that: 1) The proposed index was better for capturing the abnormal water stress state than the indicator based on current moisture deficit only, and the variations of peak value for IEDI showed high similarities to the RI. 2) High annual value of IEDI indicated severe drought condition. Thus, droughts in 2009 and 2014 in Liaoning Province, in 2007 and 2009 in Jilin Province, and in 2003, 2007 and 2009 in Heilongjiang Province were recognized as the most severe drought events during the study period, which were consistent with historical drought records. 3) IEDI was highly correlated with drought disaster area and grain yield in Northeast China. The correlation coefficients between drought disaster area and IEDI were all above 0.75, with the highest value of 0.88 in Jilin Province. The correlation coefficients between grain yield and IEDI were all above 0.60, with the highest value of 0.78 in Liaoning Province. 4) The correlation coefficients between grain yield and IEDI were higher than those between grain yield and CWSI in 4 major grain production cities: Liaoyuan, Siping, Songyuan and Changchun, manifesting a higher feasibility of IEDI to represent agricultural drought condition during study period. 5) Western Jilin and western Liaoning were the most sensitive regions to meteorological drought and were easily exposed to severe or extreme agricultural drought. 6) The correlation coefficients between IEDI and grain yield first increased and then decreased with the increase of drought duration when start month was fixed. And they were highly related to the start month when drought duration was fixed. Conclusively, the proposed index in this study is able to indicate agricultural drought effectively, which provides an effective way for agricultural drought monitoring.