知识与财富的链接
土壤水分是量度干旱程度最重要的指标,如何对其有效监测与预警一直是各界致力解决的重大科学问题。基于Suomi NPP/VIIRS数据的温度植被干旱指数TVDI、归一化植被水分指数NDWI、植被状况指数VCI,分别构建了青海省东部农业区3种土壤水分监测模型,利用连续的野外定点观测数据及生态站点观测数据进行模型检验,并在2017年夏旱过程进行了应用检验。结果表明：2012—2016年模型回代检验中,TVDI指数模型表现最优(RMSE为4.4%),其次为VCI指数模型(RMSE为4.7%),NDWI指数模型表现最差(RMSE为5.2%);2018—2020年夏季互助遥感检验场定点观测检验中,TVDI指数模型表现最好(RMSE为3.8%),VCI指数模型次之(RMSE为5.0%),NDWI指数模型表现最差(RMSE为8.8%);2017年夏季干旱过程中,TVDI指数模型反演的旱情发展过程及分布范围与实际旱情情况相符,而NDWI指数模型反演的旱情分布范围明显偏小,VCI指数模型甚至不能反映旱情缓解、解除期的变化。
Soil moisture is a most important index to measure the degree of drought and continues to bea major scientific problem for effective monitoring and early warning about soil moisture for all walks of life. The normalized difference water index, vegetation condition index and temperature-vegetation dryness index were calculated based on Suomi NPP/VIIRS to construct three soil moisture monitoring models in the eastern agricultural region of Qinghai Province.Models were tested using continuous field fixed-point observation data and site observation data, and application tests were performed in the summer drought of 2017. The results showed that the TVDI model was the best (RMSE was 4.4%), the VCI model was the second (RMSE was 4.7%) and the NDWI model was the worst (RMSE was 5.2%) in the back-generation test during 2012-2016. When the three models applied to the field survey points in Huzhu County of Qinghai Province using the remote sensing inspection during 2018 to 2020, TVDI model performed best (RMSE was 3.8%), VCI model took the second place (RMSE was 5.0%)and NDWI model was the worst (RMSE was 8.8%).The drought distribution range retrieved by TVDI model was consistent with the actual occurrence and development and mitigation process.The changes of drought mitigation or relief period were not reflectedby VCI model in drought distribution and the drought area retrieved by NDWI model was smaller in the summer drought of 2017.
