考虑物候期的广义极化分解定量反演高植被覆盖和复杂地表的土壤湿度

Polarimetric radar is efficient to monitor the spatiotemporal distribution of soil moisture, which plays an important role in crop growth and yield. However, over the agricultural area covered by high vegetation, the soil moisture retrieval is challenging, due to the complex variability of vegetation structure with the phenological growth. In addition, current polarimetric decomposition did not sufficiently account for the depolarization from large surface roughness. These obstacles hinder the timely monitoring of soil moisture. In order to solve aforementioned problems, this project proposes to (1) establish dynamic volume scattering model which considers the crop phenological stages to capture the variability in crop shape and spatial distribution; (2) improve the surface scattering component by coupling the bare soil scattering model and compensating the microwave power attenuation during the propagation through vegetation layer. Moreover, the depolarization effect of surface roughness is introduced into the dihedral scattering component by rotating the soil surface around the radar line of sight; (3) develop the generalized polarimetric decomposition to remove the dynamic volume scattering component from the full coherency matrix. Then, the cost functions from the surface and dihedral scattering components are built to retrieve the soil moisture, which is validated using the ground measurements. This project improves the soil moisture retrieval over the complex scenario (high vegetation amount and large roughness), and provides scientific basis for predicting the drought and flood disaster over the agricultural fields.

土壤水直接影响农作物生长和粮食产量，极化雷达是监测土壤水时空分布的有效手段。但是，由于植被的形态随着物候生长而动态变化，且当前极化分解对地表粗糙度去极化效应考虑不足，使高植被覆盖和大粗糙度地表的土壤水信息严重缺失，阻碍了对土壤水的适时监测。针对上述问题，本项目将（1）构建考虑物候期的体散射模型，捕捉植被生长引起的形态和空间分布动态变化；（2）通过耦合极化分解和裸地散射模型构建表面散射分量，并用透射率补偿植被对下层土壤散射的二次衰减。绕雷达视线旋转二面角散射分量的土壤平面引入地表大粗糙度的去极化效应；（3）建立广义极化分解模型从RADARSAT-2时间序列中剔除植被动态影响，分别基于表面散射和二面角散射分量构建代价函数反演高植被覆盖和大粗糙度地表的土壤水分，用地面实测值验证精度。本项目的开展对改进复杂地表土壤水的动态监测，研究气候变化背景下农田水分布格局、旱涝灾害风险评估具有重要的科学意义。

土壤水影响农作物生长和粮食产量，极化雷达因所用微波的穿透性，成为监测土壤水时空分布的有效工具。 但是，由于植被的形态随着物候生长而动态变化，且当前极化分解和辐射传输模型对地表粗糙度去极化效应考虑不足，使高植被覆盖和大粗糙度地表的土壤水分布信息严重缺失，阻碍了对土壤水的适时监测。针对上述问题，本项目（1）提出了随植被物候期变化的植被结构参数，构建了动态体散射模型，捕捉植被生长期内形态和空间分布的动态变化，进一步探索了植被层对微波信号的衰减和散射效应，更好的分离了微波信号中来自植被层和土壤层的信号。（2）将裸地散射模型耦合于极化雷达辐射传输模型、目标分解模型。通过透射率补偿植被对下层土壤散射的二次衰减，通过绕雷达视线旋转二面角散射分量的土壤平面引入地表大粗糙度的去极化效应，将传统的只适用于高频波段（C、X波段）的两分量（表面散射、体散射）水云模型，拓展为L波段下的三分量（表面散射、二面角散射、体散射）水云模型，并分别基于C波段Sentinel-1和L波段的ALOS2-ScanSAR多角度极化雷达反演了青藏高原土壤水分的时空变化。（3）基于动态体散射模型、考虑地表大粗糙度去极化效应的二面角散射分量模型、裸露地表散射模型，构建了广义极化分解模型和雷达后向散射系数分解模型，并从二面角散射和表面散射分量中反演土壤水分；开发了对地面站点数据的升尺度算法，将点尺度的站点数据匹配于微波遥感反演的不同空间分辨率的土壤湿度，以对反演结果进行精度验证和评价。（4）分析了不同气候区植被对ESA-CCI 和SMAP土壤水分产品反演精度的影响，针对SMAP土壤水分数据在青藏高原地区的缺失问题，开发了SMAP像元重构算法, 并耦合了微波和光学遥感影像，以改进对土壤湿度和植被动态结构的反演能力。本项目所取得的研究成果对高植被区、复杂地表土壤水的动态监测具有科学意义，对研究气候变化背景下农田水分含量的时空变异、旱涝灾害风险评估、智慧农业具有应用价值。