基于Landsat8的海冰表面温度反演算法研究  

Research on Sea Ice Surface Temperature Retrieval Algorithm Based on Landsat8

作　　者：范沛 赵羲[1,2] 庞小平[1,2] 刘玥 FAN Pei

机构地区：[1]武汉大学中国南极测绘研究中心,湖北武汉430079 [2]极地测绘科学国家测绘地理信息局重点实验室,湖北武汉430079

出　　处：《地理空间信息》2019年第11期52-56,I0002,共6页Geospatial Information

基　　金：国家自然科学基金面上项目资助项目(41876223、41576188)

摘　　要：Landsat 8影像被广泛应用于陆地表面温度的提取,但基于Landsat 8的海冰表面温度产品依旧空缺,且无标准算法可循。基于Landsat 8卫星搭载的TIRS传感器获取的热红外波段影像,利用3种常用的劈裂窗算法分别进行海冰表面温度反演,并将结果与浮标实测海冰表面温度数据和MODIS海冰表面温度产品进行了对比和验证。与浮标数据相比,DU C等的算法平均绝对误差和均方根误差最小,分别为1.7186 K和2.6213 K,可作为海冰表面温度反演的标准算法;而Jimenez-Muooz J C等的算法结果与MODIS冰温产品最为接近,适用于需与已有的MODIS冰温产品联合使用的情况,且两种算法的精度均优于已有的MODIS冰温产品;JIN M J等的算法在研究区域表现不佳。同时,与浮标数据相比,所有基于卫星影像反演的冰温产品在数值上总体偏低。Landsat8 imagery has been widely used in land surface temperature retrieval. However, there is no sea ice surface temperature(IST) products based on Landsat8 imagery yet, nor is a standard algorithm to follow. In this paper, we tried to use the thermal imagery captured by TIRS carried on Landsat8 satellite to derive sea ice surface temperature based on three mainstream split window algorithms, and validated the results by MODIS IST products and drifting buoy data. Compared to drifting buoy data, the method introduced by DU C has the best accuracy with a mean absolute deviation of 1.718 6 K and a root mean square error of 2.621 3 K, which can be treat as the standard algorithm to retrieve sea ice surface temperature. The method proposed by Jimenez-Munoz J C coincides with MODIS IST products best and suggested to be used together with MODIS IST products. Both of them have better accuracy than MODIS IST products, but the result by JIN M J’s method is not ideal in our study area. All the satellite IST products tend to underestimate the real IST measured by drifting buoys.

关 键 词：海冰表面温度 劈裂窗算法 LANDSAT 浮标 

分 类 号：P237[天文地球—摄影测量与遥感]