摘要
针对地形对积雪堆积与合成孔径雷达(SAR)散射能量的影响,该文提出一种通过C波段Sentinel-1 SAR的HH与HV极化后向散射系数和入射角数据以及海拔、坡度和坡向等地形数据反演南北极雪密度的方法。以欧洲中期天气预报中心(ECMWF)提供的南北极雪密度数据作为地面真值,通过XGBoost模型反演雪密度。使用2019—2021年的数据训练反演模型,通过2022年数据验证模型的泛化性能,平均绝对误差(MAE)为25.889 kg/m^(3),均方根误差(RMSE)为36.497 kg/m^(3);以东南极洲比利时伊丽莎白公主站、加拿大北极群岛与北冰洋实测数据进一步验证模型性能,MAE为37.514 kg/m^(3),RMSE为43.287 kg/m^(3)。结合地形信息后南北极雪密度反演精度得到大幅度改善,MAE降低24.219 kg/m^(3),RMSE降低28.25 kg/m^(3)。所提方法具有大规模雪密度反演的潜力,有利于南北极雪水当量和表面物质平衡的评估。
Aiming at the influence of terrain on snow density and backscattering of synthetic aperture radar(SAR),a method to invert snow density of Arctic and Antarctica using C-band HH and HV polarization data from Sentinel-1SAR along with elevation,slope,and aspect information was proposed in this paper.The snow density data of the Arctic and Antarctica provided by the European Centre for Medium-Range Weather Forecasts(ECMWF)were used as ground truth to invert the snow density using the XGBoost model.The inversion model was trained using data from 2019 to 2021,and the generalization performance of the model was verified by the data from 2022,with an MAE of 25.889 kg/m^(3) and an RMSE of 36.497 kg/m^(3).The model performance was further verified by the field-measured data from the Belgium Princess Elisabeth Station in East Antarctica,the Canadian Arctic Archipelago and the Arctic Ocean,with an MAE of 37.514 kg/m^(3) and an RMSE of 43.287 kg/m^(3).Incorporating terrain information enhances the inversion accuracy,reducing MAE by 24.219 kg/m^(3) and RMSE by 28.25 kg/m^(3).This approach enabled large-scale snow density inversion,crucial for assessing snow water equivalent and surface mass balance in polar regions.
作者
杨树瑚
王斌
张云
洪中华
韩彦岭
王静
YANG Shuhu;WANG Bin;ZHANG Yun;HONG Zhonghua;HAN Yanling;WANG Jing(School of Information Science and Technology,Shanghai Ocean University,Shanghai 201306,China;Shanghai Marine Intelligent Information and Navigation Remote Sensing Engineering Center,Shanghai 201306,China)
出处
《测绘科学》
CSCD
北大核心
2024年第12期114-123,共10页
Science of Surveying and Mapping
基金
国家自然科学基金项目(42271335,42176175)。
作者简介
杨树瑚(1983-),男,江苏盐城人,副教授,主要研究方向为极地科学和人工智能技术应用。E-mail:shyang@shou.edu.cn;通信作者:张云,教授,E-mail:y-zhang@shou.edu.cn。
