基于优化温度植被干旱指数干湿边界的淮河流域麦田墒情反演

Soil Moisture Estimation of Wheat Fields by Optimizing Dry and Wet Boundaries of Temperature Vegetation Drought Index in Huaihe Basin

摘要:

根据温度植被干旱指数（Temperature vegetation drought index, TVDI）的区域土壤含水率反演对于流域旱情监测、水资源规划等具有极大潜力，但TVDI特征空间干湿边量化的经验性和不确定性易导致反演精度受限。提出了TVDI干湿边多目标优化求解方法，通过最大化TVDI与地表反照率（Albedo, A）、土壤部分红波反射率（Soil red band reflectance, Rs,red）和土壤部分近红外波反射率（Soil near-infrared band reflectance, Rs,nir）的相关性实现特征空间干湿边定量求解，并对淮河流域麦田墒情进行了反演分析。结果表明：TVDI干湿边优化求解时，地表反照率对墒情反演精度的提升占主导作用，权重为0.5~0.8，其次是土壤红波和土壤近红外波反射率，分别为0.1~0.2和0.1~0.3；优化后TVDI对生育期内气象干旱变化具有更好的响应，特征空间涵盖范围增加了24.05%~54.02%，干边截距增加了1.72%~5.69%，干边斜率减小了8.04%~66.51%；优化后TVDI与实测土壤含水率的决定系数（Coefficient of determination, R2）增加了33.12%~82.61%，反演土壤含水率时的平均绝对误差（Mean absolute error, MAE）、均方根误差（Root mean square error, RMSE）、归一化均方根误差（Normalized root mean square error, NRMSE）降低了5.09%~20.52%、7.73%~21.16%、7.69%~21.27%，在不同生育期和土层深度均能保持较高精度；2023年淮河流域冬小麦拔节期、孕穗期、开花期和灌浆期0~40cm平均土壤含水率分别为0.242、0.255、0.259、0.237cm3/cm3，流域内河南省和山东省麦地墒情较低，适宜在拔节期、开花期和灌浆期进行补充灌溉。综上，干湿边多目标优化求解方法提升了TVDI在区域尺度麦田墒情反演的适应性和准确性，可为旱情监测及防控研究提供理论依据和可靠工具。

Abstract:

The regional soil moisture estimation based on the temperature vegetation dryness index (TVDI) holds significant potential for drought monitoring and water resource planning in basins. However, the empirical nature and uncertainty in quantifying the dry and wet boundaries within the TVDI feature space can easily limit the accuracy of the estimation. A multi-objective optimization method was introduced to address the determination of TVDI’s dry-edge and wet-edge. By maximizing the correlation between TVDI and both surface albedo as well as soil red/near-infrared reflectance, a quantitative solution for the dry and wet edges in the feature space was achieved, enabling an analysis of soil moisture inversion in wheat fields within the Huai River Basin. Results indicated that during the optimization of the TVDI dry and wet edges, surface albedo significantly enhanced the accuracy of soil moisture inversion, accounting for a weight of 0.5~0.8. The contributions from soil red wave and near-infrared reflectivity were relatively smaller, at 0.1~0.2 and 0.1~0.3, respectively. The optimized TVDI demonstrated improved responsiveness to changes in meteorological drought during the wheat growing period. The coverage range of the feature space was expanded by 24.05%~54.02%, with the intercept of the dry edge increased by 1.72%~5.69% and the slope decreased by 8.04%~66.51%. After optimization, the coefficient of determination (R2) between TVDI and measured soil moisture content was increased by 33.12%~82.61%. Meanwhile, the average absolute error (MAE), root mean square error (RMSE), and normalized root mean square error (NRMSE) during soil moisture estimation was decreased by 5.09%~20.52%, 7.73%~21.16%, and 7.69%~21.27%, respectively, ensuring high accuracy across different growth stages and soil layer depths. In 2023, the average soil moisture content at 0~40cm depth during the jointing, booting, flowering, and grain-filling stages of winter wheat in the Huai River Basin was 0.242cm3/cm3, 0.255cm3/cm3, 0.259cm3/cm3, and 0.237cm3/cm3, respectively. Wheat fields in Henan Province and Shandong Province exhibited relatively low soil moisture levels, making supplementary irrigation advisable during the jointing, flowering, and grain-filling stages. In conclusion, the multi-objective optimization method for determining the dry and wet edges improved the adaptability and precision of TVDI for regional-scale soil moisture inversion in wheat fields. The research result can provide a novel theoretical foundation and reliable tools for drought monitoring and prevention research.