摘要：
In this study, two hybrid residual deep learning models coupled with physical knowledge are proposed for improving daily transpiration (Ec) estimation. A Hybrid-Physics-Data-Residual Learning (HPDRL) model is constructed by mixing a Hybrid-Physics-Data (HPD) model with a Physics-based Residual Learning (PRL) model. To this purpose, the HPD model is first formed by adding a complementary physical variable (EcPHY), which is generated by a recently presented physics-based Ec model (hereafter "BTA-& psi;"), to a deep learning (DL) model along with driving variables to regress Ec. Then, the PRL model is developed by using the residual learning method to integrate the BTA-& psi; and DL models. Three hybrid models, HPD, PRL, and HPDRL, are used to estimate daily Ec for the three species of trees and compared with two baseline models, the BTA-& psi; and pure DL models. The results show that the PRL and HPDRL models benefit from the integration of the BTA-& psi; and DL models via the residual learning method, and thus effectively improve daily Ec estimation. In contrast, the HPD model, limited by the flawed physics-based BTA-& psi; model, exhibits the weakest estimation skill among all three hybrid models. Moreover, the HPDRL model further exhibits better generalization capability than the PRL and pure DL models. Although both hybrid residual learning models can capture the range between the minimum and maximum observed Ec more completely than the pure DL model, the HPDRL model extrapolates better than the PRL model in unseen scenarios with limited training samples.
作者机构：
[Zhou, Jialu; Wang, Cheng] Yunnan Normal Univ, Fac Geog, Kunming, Peoples R China.;[Zhou, Jialu; Nie, Sheng; Wang, Cheng] Int Res Ctr Big Data Sustainable Dev Goals, Beijing, Peoples R China.;[Zhou, Jialu; Nie, Sheng; Wang, Cheng] Chinese Acad Sci, Aerosp Informat Res Inst, Key Lab Digital Earth Sci, Beijing, Peoples R China.;[Zheng, Wenwu; Deng, Yunyuan; Fu, Jing] Hengyang Normal Univ, Coll Geog & Tourism, Hengyang, Peoples R China.;[Sun, Yue] Natl Forestry & Grassland Adm, East China Survey & Planning Inst, Hangzhou, Peoples R China.
通讯机构：
[Nie, S ] I;Int Res Ctr Big Data Sustainable Dev Goals, Beijing 100094, Peoples R China.;Chinese Acad Sci, Aerosp Informat Res Inst, Key Lab Digital Earth Sci, Beijing 100094, Peoples R China.
关键词：
Forest height;terrain height;ICESat-2;space-borne LiDAR;photon-counting;leaf-on and leaf-off
摘要：
Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) provides effective photon-counting light detection and ranging (LiDAR) data for estimating forest height across extensive geographical areas. Although prior studies have illustrated canopy conditions during leaf-on and leaf-off phases may influence ICESat-2 derived forest heights, a comprehensive understanding of this effect remains incomplete. This study seeks to comprehensively assess how varying canopy conditions (leaf-on/leaf-off) affect ICESat-2 forest height retrieval and modelling. First, the accuracies of ICESat-2 terrain and canopy heights under leaf-on and leaf-off conditions were validated. Second, random forest algorithm was utilized to model forest height by integrating ICESat-2, Sentinel-2, and other ancillary datasets. Finally, we evaluated the influence of leaf-on and leaf-off conditions on forest height retrieval and modelling. Results reveal higher consistency between ICESat-2 and airborne LiDAR-derived terrain heights compared to the agreement between two canopy height datasets. Accuracies of ICESat-2 terrain and canopy heights are higher under leaf-off conditions in contrast to leaf-on conditions. Notably, the accuracies of ICESat-2 terrain and canopy heights under various conditions are closely linked to canopy cover. Furthermore, the accuracy of forest height modelling can be enhanced by combining ICESat-2 data collected during both leaf-on and leaf-off seasons with further eliminating low-quality samples.
作者机构：
[Guo, Binbin; Dai, Zhong; Yang, Qin; Deng, Yunyuan; Zou, Jun] Hengyang Normal Univ, Coll Geog & Tourism, Hengyang 421002, Peoples R China.;[Xu, Tingbao] Australian Natl Univ, Fenner Sch Environm & Soc, Canberra, ACT 2601, Australia.;[Yang, Qin] China Univ Geosci, Sch Informat Engn, Beijing 100083, Peoples R China.;[Zhang, Jing] Capital Normal Univ, Beijing Lab Water Resources Secur, Beijing 100048, Peoples R China.;[Dai, Zhong] Int Ctr Space Technol Nat & Cultural Heritage HIST, Hengyang Base, Hengyang 421002, Peoples R China.
通讯机构：
[Binbin Guo] C;College of Geography and Tourism, Hengyang Normal University, Hengyang 421002, China<&wdkj&>Author to whom correspondence should be addressed.
关键词：
precipitation;satellite observation;evaluation;hydrologic modelling;Xiangjiang River basin
摘要：
Satellite precipitation products (SPPs) have emerged as an important information source of precipitation with high spatio-temporal resolutions, with great potential to improve catchment water resource management and hydrologic modelling, especially in data-sparse regions. As an indirect precipitation measurement, satellite-derived precipitation accuracy is of major concern. There have been numerous evaluation/validation studies worldwide. However, a convincing systematic evaluation/validation of satellite precipitation remains unrealized. In particular, there are still only a limited number of hydrologic evaluations/validations with a long temporal period. Here we present a systematic evaluation of eight popular SPPs (CHIRPS, CMORPH, GPCP, GPM, GSMaP, MSWEP, PERSIANN, and SM2RAIN). The evaluation area used, using daily data from 2007 to 2020, is the Xiangjiang River basin, a mountainous catchment with a humid sub-tropical monsoon climate situated in south China. The evaluation was conducted at various spatial scales (both grid-gauge scale and watershed scale) and temporal scales (annual and seasonal scales). The evaluation paid particular attention to precipitation intensity and especially its impact on hydrologic modelling. In the evaluation of the results, the overall statistical metrics show that GSMaP and MSWEP rank as the two best-performing SPPs, with KGE(Grid) >= 0.48 and KGE(Watershed) >= 0.67, while CHIRPS and SM2RAIN were the two worst-performing SPPs with KGE(Grid) <= 0.25 and KGE(Watershed) <= 0.42. GSMaP gave the closest agreement with the observations. The GSMaP-driven model also was superior in depicting the rainfall-runoff relationship compared to the hydrologic models driven by other SPPs. This study further demonstrated that satellite remote sensing still has difficulty accurately estimating precipitation over a mountainous region. This study provides helpful information to optimize the generation of algorithms for satellite precipitation products, and valuable guidance for local communities to select suitable alternative precipitation datasets.
期刊：
Annals of the American Association of Geographers,2023年113(5):1190-1206 ISSN：2469-4452
作者机构：
[Wu, Bo; Yan, Jinbiao] Jiangxi Normal Univ, Sch Geog & Environm, Nanchang, Peoples R China.;[Yan, Jinbiao] Hengyang Normal Univ, Coll Geog & Tourism, Hengyang, Peoples R China.;[Cao, Kai] East China Normal Univ, Sch Geog Sci, Shanghai, Peoples R China.
关键词：
coefficient optimization;geographically weighted regression;l0-norm;splicing algorithm;MBIC;variable selection;algoritmo de empalme;10-norm;MBIC;regresión geográficamente ponderada;selección de variables
摘要：
Forest aboveground biomass (AGB) retrieval using synthetic aperture radar (SAR) backscatter has received extensive attention. The water cloud model (WCM), because of its simplicity and physical significance, has been one of the most commonly used models for estimating forest AGB using SAR backscatter. Nevertheless, forest AGB estimation using the WCM is usually based on simplified assumptions and empirical fitting, leading to results that tend to overestimate or underestimate. Moreover, the physical connection between the model and the polarimetric synthetic aperture radar (PolSAR) is not established, which leads to the limitation of the inversion scale. In this paper, based on the fully polarimetric SAR data from the Advanced Land Observing Satellite-2 (ALOS-2) Phased Array-type L-band Synthetic Aperture Radar (PALSAR-2), the relative contributions of the three major scattering mechanisms were first analyzed in a hilly area of southern China. On this basis, the traditional WCM was extended by considering the secondary scattering mechanism. Then, to establish the direct relationship between the vegetation scattering mechanism and forest AGB, a new relationship equation between the PolSAR decomposition model and the improved water cloud model (I-WCM) was constructed without the help of external data. Finally, a nonlinear iterative method was used to estimate the forest AGB. The results show that volume scattering is the dominant mechanism, accounting for more than 60%. Double-bounce scattering accounts for the smallest fraction, but still about 10%, which means that the contribution of the double-bounce scattering component is not negligible in forested areas because of the strong penetration capability of the long-wave SAR. The modified method provides a correlation coefficient R2 of 0.665 and a root mean square error (RMSE) of 21.902, which is an improvement of 36.42% compared to the traditional fitting method. Moreover, it enables the extraction of forest parameters at the pix scale using PolSAR data without the need for low-resolution external data and is thus helpful for high-resolution mapping of forest AGB.
