风云三号C星微波全球地表温度产品精度评估

地表温度是衡量地表水热平衡的关键参数,微波地表温度因其范围大、全天候等独特的优势,在气候、农业和环境等领域得到广泛应用。基于经质量控制的MODIS地表温度产品对风云三号卫星C星的微波地表温度日产品和月平均产品进行验证评估,结果显示:FY-3C卫星升轨(夜晚)和降轨(白天)微波地表温度日产品平均分别高估8. 7 K、低估13. 2 K,月平均产品平均分别高估7. 9 K、低估12. 0 K,日产品和月平均产品的反演误差都在15K以内。在全球空间分布上,升轨(夜晚)和降轨(白天)月产品误差分别呈现高估和低估,热带雨林区和沙漠、荒漠区域在夜晚分别高估5 K以内和30 K以内,白天则分别低估10 K以内和10～30 K。不同土地覆盖类型间FY-3C微波地表温度反演精度存在差异,总体上升轨(夜晚)比降轨(白天)的精度高,反演精度最高和最低的土地类型分别是常绿阔叶林和荒漠、稀疏植被,不同土地覆盖类型间的地表温度反演精度在季节上存在明显差异。根据分析结果,改进FY-3C微波地表温度的反演算法,可进一步提高微波地表温度的反演精度。

Accuracy Evaluation of the FengYun-3C Global Land Surface Temperature Products Retrieval from Microwave Radiation Imager

Land Surface Temperature (LST) is a key variable in the processes controlling energy, water over the interface between the land surface and atmosphere. LST retrieved from the passive microwave remote sensing was widely used in many applications, such as climate, agriculture and environment, etc. LST products retrieved from FengYun (FY) 3C Microwave Radiation Imager (MWRI) were evaluated based on the MODIS LST products through quality control. The results showed that the ascending (night) and descending (day) daily LST products of FY 3C overestimate 8.7 K, underestimate 13.2 K, respectively, and mean monthly LST overestimate 7.9 K, underestimate 12.0 K. The RMSE of ascending (night) and descending (day) daily and mean monthly LST products were within 15 K. The ascending (night) and descending (day) mean monthly LST overestimated and underestimated respectively all over the world.Ascending (night) mean monthly LST overestimate within 5 K, and 30 K respectively, descending (day) mean monthly LST underestimate within 10 K, 1030 K respectively over tropical rain forest, barren and sparsely vegetated area. Generally, LST of the different land cover classes differ significantly, accuracy of ascending (night) mean monthly LST products were more higher than that of descending (day). The tropical rain forest, barren and sparsely vegetated had the highest and lowest accuracy of mean monthly LST, respectively. There were obviously seasonal differences of the LST accuracy among the land cover classes. According to the analysis results, modifying the retrieval algorithm could improve the accuracy of FY 3C MWRI LST products in the future.