Agricultural Drought Prediction Model In China Based On Meteorological Drought And High Temperatures

With the intensification of global climate change and the rapidly increasing population,the demand for food and water resource will do further expand.Food and water security mostly are affected by agricultural droughts.The agriculture drought(soil moisture deficit)is influenced by precipitation deficit(meteorological drought)and high temperatures.Thus,the development of efficient and reliable methods for agricultural drought prediction by considering these factors is crucial to crop yield security,water resources allocation,and drought mitigation deployment.The diversity temporal scales of Standardized Soil Moisture Index(SSI)are applied to monitor agricultural drought and may result in inconsistent drought situations.The Joint Standardized Soil Moisture Index(JSSI),which is used to assess the comprehensive situation of agricultural drought,is constructed by combing the SSI(calculated by GLDAS-Noah root zone soil moisture)over 1-,3-,6-,9-,and 12-month timescales based on Kendall Copula.The6-month timescale Standardized Precipitation Index(SPI)and 3-month timescale Standardized Temperature Index(STI)are utilized to depict meteorological drought and high temperatures,respectively.Accordingly,the Standardized Compound Events Indicator(SCEI)and Standardized Dry or Hot Events Index(SDHEI)are constructed based on SPI and STI.The joint and conditional probability of agricultural drought is analyzed under the conditions of meteorological drought and high temperature with different severity in China from June to August.For the 1–3-month lead,the meta-Gaussian(MG)model and Pair Copula Constructions(PCC)model is proposed under considering two cases as predictors:1)antecedent SPI(SPI_t–i;t denotes the target month,and i indicates the lead time)combined persistence JSSI(JSSI_t–i),and 2)previous SPI,antecedent STI(STI_t–i),and JSSI persistence.The leave-one-out cross validation(LOOCV)is adopted to predict the agricultural drought in summer season of each year over China.The prediction performance of these two agricultural drought prediction models under the two cases in different climate regions of China is compared.The main results listed below:(1)The JSSI is capable of capturing both emerging and prolonged agricultural drought in a timely manner,which can reflect the comprehensive agricultural drought situation.(2)Compared with the Standardized Compound Events Indicators(SCEI)depicted compound dry-hot events(i.e.,the conditions of meteorological drought and high temperatures concurrent or consecutive),the SDHEI represented the situation that meteorological drought and high temperatures at least one appears.Therefore,SDHEI can better represent the comprehensive situation of dry and hot events,which can be used as a powerful tool for monitoring dry and hot events.(3)The probability of the compound events(i.e.,meteorological drought,high temperatures,and agricultural drought simultaneously occurrence)with the level of light,moderate,and severe is 0?20%,0?10%,and 0?6%in June,July,and August,respectively.The spatial distributions with a relatively higher probability value of the compound events mainly lied in northwestern China desert climate region,southwestern China,Tibet,and southeast coastal areas.In Southwestern China,Qinghai-Tibet Plateau,Inner Mongolia,and Eastern China,the high temperature is the most important factor to trigger the agricultural drought in these regions.The areas of meteorological drought significantly influenced agricultural drought mainly located in the Northwestern China desert,southeast coastal zone,east regions of Northern China,and the north and south part of the Qinghai-Tibet Plateau.(4)For the 1–3-month lead agricultural drought prediction performance in June–August,in comparison with the MG model combined high temperatures(used predictors including the SPT_t–i,STI_t–i,and JSSI_t–i,i.e.,SPT_t–i+STT_t–i+JSSI_t–i),the MG model without considering the high temperatures(i.e.,SPT_t–i+JSSI_t–i)performed better in each climate region.This implied the MG model could not improve the prediction performance when considering the effect of high temperature on the agricultural drought in high dimensional.(5)In terms of the typical agricultural droughts selected and performance evaluation metrics of Nash–Sutcliffe efficiency coefficient(NSE)and Coefficient of Determination(R²),the prediction skills of the PCC model combined the SPT_t–i+STT_t–i+SSI_t–i outperformed the MG model combined the SPT_t–i+SSI_t–i,further demonstrated that the PCC model is more suitable for predicting agricultural drought in summer over China.