IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 60, 2022 4103015 Integration of Multisource Data to Estimate Downward Longwave Radiation Based on Deep Neural Networks Fuxin Zhu , Xin Li , Senior Member, IEEE, Jun Qin, Kun Yang , Lan Cuo , Wenjun Tang , and Chaopeng Shen Abstract—Downward longwave radiation (DLR) at the surface is a key variable of interest in fields, such as hydrology and climate research. However, existing DLR estimation methods and DLR products are still problematic in terms of both accuracy and spatiotemporal resolution. In this article, we propose a deep convolutional neural network (DCNN)-based method to estimate hourly DLR at 5-km spatial resolution from top of atmosphere (TOA) brightness temperature (BT) of the Himawari- 8/Advanced Himawari Imager (AHI) thermal channels, combined with near-surface air temperature and dew point temperature of ERA5 and elevation data. Validation results show that the DCNN-based method outperforms popular random forest and multilayer perceptron-based methods and that our proposed scheme integrating multisource data outperforms that only using remote sensing TOA observations or surface meteorological data. Compared with state-of-the-art CERES-SYN and ERA5-land DLR products, the estimated DLR by our proposed DCNN-based method with physical multisource inputs has higher spatiotem- poral resolution and accuracy, with correlation coefficient (CC) of 0.95, root-mean-square error (RMSE) of 17.2 W/m 2 , and mean bias error (MBE) of -0.8 W/m 2 in the testing period on the Tibetan Plateau. Index Terms— Deep convolutional neural network (DCNN), downward longwave radiation (DLR), Himawari-8, Tibetan Plateau (TP). Manuscript received February 12, 2021; revised May 26, 2021; accepted June 29, 2021. Date of publication July 15, 2021; date of current version January 14, 2022. This work was supported in part by the Strategic Priority Research Program “Big Earth Data Science Engineering (CASEarth)” Chinese Academy of Sciences under Grant XDA19070104 and in part by the National Natural Science Foundation of China under Grant 41901086. (Corresponding authors: Xin Li; Jun Qin.) Fuxin Zhu and Jun Qin are with the National Tibetan Plateau Data Center, Key Laboratory of Tibetan Environmental Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Acad- emy of Sciences, Beijing 100101, China (e-mail: zhufuxin@itpcas.ac.cn; shuairenqin@itpcas.ac.cn). Xin Li and Wenjun Tang are with the National Tibetan Plateau Data Center, Key Laboratory of Tibetan Environmental Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sci- ences, Beijing 100101, China, and also with the CAS Center for Excellence in Tibetan Plateau Earth Sciences, Chinese Academy of Sciences, Beijing 100101, China (e-mail: xinli@itpcas.ac.cn; tangwj@itpcas.ac.cn). Kun Yang is with the Department of Earth Science System, Tsinghua University, Beijing 10084, China (e-mail: yangk@tsinghua.edu.cn). Lan Cuo is with the Key Laboratory of Tibetan Environmental Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China (e-mail: lancuo@itpcas.ac.cn). Chaopeng Shen is with Civil and Environmental Engineering, Pennsylvania State University, University Park, PA 16802 USA (e-mail: cshen@engr.psu.edu). Digital Object Identifier 10.1109/TGRS.2021.3094321 I. I NTRODUCTION D OWNWARD longwave radiation (DLR) at the surface has a crucial influence on land–atmosphere interaction process, such as evapotranspiration, snowmelt, soil thermal state, and atmospheric circulations [1]–[3]. DLR is also a crucial forcing variable for various land surface models, hydrology models, and climate models [4]. Furthermore, DLR is perhaps the most fundamental variable for understanding the impact of increasing CO 2 and other greenhouse gases on climate change [1], [5]. Therefore, accurate DLR estimates are much desired by the climate change research community. The most direct way to measure DLR is by ground-based instruments (e.g., pyrgeometer). However, due to high costs in instrument and maintenance, the ground-based observations of DLR are available at far fewer stations than that of other meteorological variables such as temperature and humidity, especially on the Tibetan Plateau (TP) with harsh climate conditions. In situ observations of DLR in these areas are too sparse to map surface radiation budget at a large scale, let alone to satisfy hydrometeorological studies. Benefitted from the large spatial coverage and high spatial resolution, remote sensing is widely used to map DLR at a large scale [6], [7]. Currently, some satellite-derived global DLR products have been developed, such as Global Energy and Water cycle Experiment (GEWEX) [8], International Satellite Cloud Climatology Project (ISCCP) [9], and Clouds and Earth’s Radiant Energy System (CERES) [10]. Although some of these existing satellite-derived global DLR products have fine temporal resolutions, their coarse spatial resolutions and great uncertainties can hardly meet scientific research require- ments, especially on the TP [1], [5], [11]. Therefore, it is still necessary to estimate DLR accurately at high spatial–temporal resolution. The Advanced Himawari Imager (AHI) onboard the new geostationary satellite Himawari-8 has 16 spectral bands with 0.5-, 1-, and 2-km resolutions at subsatellite point from visible to thermal infrared [12]. Compared to all earlier geostation- ary imagers, the AHI observes the Earth’s atmosphere with much higher spectral, temporal, and spatial resolutions [13]. AHI/Himawari-8 was designed to observe the Earth’s surface, atmospheric moisture, clouds from the shortwave reflectance, and longwave radiances. Furthermore, ten infrared channels 1558-0644 © 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See https://www.ieee.org/publications/rights/index.html for more information. Authorized licensed use limited to: Tsinghua University. Downloaded on January 16,2022 at 00:57:59 UTC from IEEE Xplore. Restrictions apply.