L波段SMAP卫星数据平静海面亮温信息提取研究

风致海表粗糙度是微波辐射计接收亮温的主要误差来源，通过建立亮温增益模型对海面亮温进行修正是获取平静海面亮温信息的一项关键工作。基于SMAP卫星L2C海面亮温数据和10 m风速数据，分析了亮温增益与风速之间的相关性，并利用最小二乘法根据不同风速区间建立L波段下的粗糙海面亮温增益模型，在考虑白冠对亮温影响的情况下，输入Argo浮标插值得到的温、盐数据集，使用Meissner-Wentz介电常数模型计算得到参考亮温，与亮温增益模型修正得到的平静海面亮温进行回归分析，定量评价了该模型在该海域的适用性。研究表明，风速与亮温增益呈现正相关关系，0～3 m/s风速区间的水平极化亮温灵敏度比垂直极化方式高0.25 K/(m/s)，3～12 m/s风速区间的前者比后者高0.02 K/(m/s)，大于12 m/s风速区间的前者比后者高0.01 K/(m/s)；在靠近陆域的海面亮温偏差比较大，在开阔海域大部分极化亮温误差均能控制在0.2 K以内；模型亮温与参考亮温回归分析的决定系数均在0.5以上，均方根误差均可以控制在0.2 K以内。

Extraction of flat sea surface brightness temperature information from L-band SMAP satellite

Wind-induced sea surface roughness is the main error source for microwave radiometer to receive brightness temperature.Based on SMAP satellite L2C sea surface brightness temperature data and 10 meters wind speed data, analyzes the gain the correlation between wind speed and brightness temperature, and by using least square method based on rough sea surface under different wind speed range build L-band gain brightness temperature model, considering the influence of whitecaps on brightness temperature, input Argo buoy interpolation temperature and salinity data sets, and use Meissner -Wentz dielectric constant temperature model to calculate the reference brightness temperature. Regression analysis is carried out with the flat sea surface brightness temperature corrected by the brightness temperature gain model, and the applicability of the model in the sea area was quantitatively evaluated. The results show that there is a positive correlation between wind speed and brightness temperature gain. The brightness temperature sensitivity of horizontal polarization in the wind speed range of 0～3 m/s is 0.25 K higher than that of vertical polarization mode, the former is 0.02 K higher than the latter in the wind speed range of 3～12 m/s, and the former is 0.01 K higher than later when the wind speed is greater than 12 m/s; In the sea surface brightness temperature deviation near the land area is relatively large, and in the open sea most polarization brightness temperature errors can be controlled within 0.2 K.The determination coefficients of the regression analysis of the model and reference brightness temperature are both above 0.5, and the root-mean-square error can be controlled within 0.2 K.