作者:Yan, GJ (Yan, Guangjian)[ 1 ] ; Wang, TX (Wang, Tianxing)[ 1,2 ] ; Jiao, ZH (Jiao, Zhonghu)[ 1 ] ; Mu, XH (Mu, Xihan)[ 1 ] ; Zhao, J (Zhao, Jing)[ 1 ] ; Chen, L (Chen, Ling)[ 3 ]

REMOTE SENSING OF ENVIRONMENT

卷: 172  頁: 15-27

DOI: 10.1016/j.rse.2015.10.026

出版年: JAN 2016

摘要

Longwave radiation (5-100 mu m) is a critical component of the Earth's radiation budget. Most of the existing satellite-based retrieval algorithms are valid only for flat surfaces without accounting for topographic effects. This causes significant errors. Meanwhile, the fixed spatial resolution of remote sensing data makes it difficult to link the satellite-derived longwave radiation to different land models running on various scales. These deficiencies result in an urgent need for topographic modeling and spatial scaling studies of longwave radiation. In this paper, a longwave topographic radiation model (LWTRM) is proposed that quantifies all possible radiation-affecting factors over rugged terrain. For driving the LWTRM, a hybrid method for simultaneously deriving multiple components of longwave radiation from MODIS data is suggested based on artificial neuron networks (ANN) and the radiative transfer simulation. Topographically corrected longwave radiation is then derived by coupling the ANN outputs and LWTRM. Based on this, a general upscaling strategy for longwave radiation is presented. The results demonstrate that: (1) both the proposed LWTRM and the upscaling strategy are rather effective and work well over rugged areas; (2) the ANN-based retrieval method can produce longwave radiation with better accuracy(RMSE <23 W/m(2), bias <9 W/m(2)). More importantly, it can simultaneously derive multiple components of longwave radiation in a consistent manner; (3) over mountainous areas, the radiation cannot be accurately characterized in terms of either spatial distribution or specific values if topographic effects are neglected, for instance, the induced error can reach up to 100 W/m(2) for the longwave net flux; and (4) the topographic effects cannot be ignored below spatial scale of approximately 5 km in the selected study area. (C) 2015 Elsevier Inc All rights reserved.

通訊作者地址: Wang, TX (通訊作者)

Beijing Normal Univ, Sch Geog, State Key Lab Remote Sensing Sci, Beijing 100875, Peoples R China.

地址:

[ 1 ] Beijing Normal Univ, Sch Geog, State Key Lab Remote Sensing Sci, Beijing 100875, Peoples R China

[ 2 ] Chinese Acad Sci, Inst Remote Sensing & Digital Earth, Beijing 100101, Peoples R China

[ 3 ] Beijing Forestry Univ, Acad Forestry, Beijing 100083, Peoples R China