多时间尺度温度-植被指数特征空间旱情监测的差异性

采用2011年5月28日MODIS多时间尺度数据产品和土壤相对含水量RSM数据,对河北省多时间尺度Ts-EVI特征空间旱情遥感监测的差异性进行了研究,结果表明：① 多时间尺度Ts-EVI特征空间的TVDI与土壤表层RSM具有较高的相关性,1 d尺度Ts和1 d、8 d尺度EVI构建的RSM-TVDI决定系数较高,8 d尺度的Ts和8 d、16 d尺度EVI的RSM-TVDI决定系数较低。② 多时间尺度Ts-EVI特征空间的旱情监测结果在空间分布上具有较好的一致性,但其面积存在一定的差异;旱情监测应用中Ts-EVI特征空间构建应首选1 d时间尺度的Ts和EVI,其次是1 d尺度的Ts和8 d尺度的EVI,再次为8 d尺度的Ts和8 d尺度的EVI。

Diversities of Drought Monitoring Based on Temperature-vegetation Index Featured Spaces with Multi-time Scale

Agricultural drought is one of the most serious meteorological disasters in China. Temperature (Ts)-vegetation index (VI) featured space is an important method for soil moisture estimation and drought monitoring. In this paper, multi-time scales AQUA- MODIS products such as MYD11A1 with the day of year (DOY)148, MYD09GA (DOY:148), MYD11A2 (DOY:145), MYD09A1 (DOY:145) and MYD13A2 (DOY:137) acquired on May 28, 2011 were used to establish Ts-enhanced vegetation index (EVI) featured space with 1 d, 8 d and 16 d time scales. Besides, 36 in situ relative soil moisture (RSM) on May 28, 2011 and ASTER-GDEM2 data in Hebei Province were also used to study the diversity of drought monitoring. Results showed that: 1) Correlations between temperature vegetation drought index (TVDI) derived from multi-time scale Ts-EVI featured spaces and RSM varied significantly in different soil depths. Among them, the correlation of RSM₂₀-TVDI was the highest, correlation of RSM₁₀-TVDI was higher and that of RSM₅₀-TVDI was the lowest. Relationships between RSM₂₀ and TVDI derived from Ts-EVI featured spaces with multi-time scales showed that coefficients of determination of RSM-TVDI with Ts₁₄₈-EVI₁₄₈ and Ts₁₄₈-EVI₁₄₅ scales were relatively higher and those with Ts₁₄₈-EVI₁₃₇, Ts₁₄₅-EVI₁₄₅ and Ts₁₄₅-EVI₁₃₇ scales were relatively lower. 2) Among the 36 in situ RSM samples, 10 samples were used as test samples and the rest ones were used to establish the soil moisture estimation model RSM₂₀-TVDI. Moreover, mean absolute error (MAE), mean relative error (MRE) and root mean square error (RMSE) were used as indices to evaluate RSM estimation abilities of the RSM₂₀-TVDI model. Error tests showed that the maximums of MAE, MRE and RMSE of the RSM₂₀-TVDI models from different time scales featured spaces were no more than 4.85%, 8.33% and 0.054, respectively. So the RSM₂₀-TVDI model could estimate soil moisture successfully. 3) The spatial distributions of drought retrieved from Ts with 1d (Ts₁₄₈) and 8d (Ts₁₄₅) scales and EVI with 1 d (EVI₁₄₈), 8 d (EVI₁₄₅) and 16 d (EVI₁₃₇) scales, respectively, showed good consistencies. Drought monitoring on May 28, 2011 showed that the mild drought regions mainly distributed in the northern parts of Langfang, northern of Cangzhou, southern of Hengshui, middle and southern of Xingtai and northern of Handan. Moderate drought regions mainly located in southwestern of Cangzhou and eastern of Xingtai. Severe and extreme drought regions mainly distributed in the southwestern parts of Baoding, southern of Shijiazhuang and the middle and northern of Handan. 4) Diversity existed in drought areas derived from Ts-EVI featured spaces with multi-time scales. For drought monitoring based on Ts-EVI featured space with MODIS products, on the condition that the data were available and with ideal quality, Ts and EVI with 1 d scale should be the first choice, then Ts with 1 d scale and EVI with 8 d scale, both Ts and EVI with 8 d scale, and Ts with 1 d or 8 d and EVI with 16 d scales should be the last choice.