Estimation of Vegetation Parameters of Water Cloud Model for Global Soil Moisture Retrieval Using Time-Series L-Band Aquarius Observations
Estimation of Vegetation Parameters of Water Cloud Model for Global Soil Moisture Retrieval Using Time-Series L-Band Aquarius Observations
复制标题
使用时间序列 L 波段 Aquarius 观测数据反演全球土壤水分的水云模型植被参数估计
DOI：
10.1109/jstars.2016.2596541
复制
发表时间：
2016
影响因子：
5.5
通讯作者：
J. Shi
中科院分区：
文献类型：
--
作者：
C. Liu;J. Shi
利用宝瓶座中波束散射计观测资料，估算了大尺度水云模式的植被参数，并将其应用于全球土壤水分的反演。植被后向散射采用了两种模型：用OH模型描述裸露地表的散射，用水云模型来考虑植被冠层的影响。通过最小化宝瓶座散射计观测值与水云模式模拟的后向散射系数之间的偏差来估计植被参数。两种共聚反应的均方根误差均小于2分贝，且在大多数地区相关性很强(CC>；0.6)。植被参数被用来从宝瓶座雷达数据中反演全球土壤湿度。与宝瓶座辐射计观测得到的宝瓶座土壤水分产品的比较表明，在世界大部分地区，ubRMSE(0.06cm3/cm3)很低，相关性很强(CC>；0.6)。利用蒙特卡罗模拟方法研究了水云模型输入参数误差对植被参数估计的影响。当输入数据无噪声或仅引入雷达测量误差时，该算法收敛于参数的真值。研究发现，植被参数的误差对输入土壤湿度的误差很敏感。两个植被参数的误差相互抵消，减小了后向散射模拟的误差。这项研究表明，如果植被参数设置得当，水云模式可以应用于全球散射计观测，以反演土壤水分。
Using Aquarius middle beam scatterometer observations, the vegetation parameters of the water cloud model at large scale are estimated and applied to global soil moisture retrieval. Vegetation backscattering is derived using two models: Oh model is used to describe the scattering from bare soil surface, while the water cloud model is implemented to account for the effect of vegetation canopy. The vegetation parameters are estimated by minimizing the deviations between the Aquarius scatterometer observations and backscatter coefficients simulated by the water cloud model. The RMSE is less than 2 dB for both copolarizations and correlation is strong (CC > 0.6) in most areas. The vegetation parameters were used to retrieve global soil moisture from Aquarius radar data. The comparisons with the Aquarius soil moisture product derived from the Aquarius radiometer observations show low ubRMSE (0.06cm3 /cm3) and strong correlation (CC > 0.6) in most parts of the world. The impact of errors in input parameters of the water cloud model on the vegetation parameter estimation was assessed by using a Monte-Carlo simulation. The algorithm converges to the true values of the parameters when the input data is noise-free or only the radar measurement error is introduced. It was found that the errors in vegetation parameter are sensitive to the errors in input soil moisture. The errors in two vegetation parameters counteract each other to decrease the error of backscattering simulation. This study demonstrates that the water cloud model could be applied to global scatterometer observations to retrieve soil moisture if the vegetation parameters are appropriately set.