首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 459 全文下载次数: 399
引用本文:

DOI:

10.11834/jrs.20221748

收稿日期:

2021-11-19

修改日期:

2022-03-22

PDF Free   EndNote   BibTeX
积雪季森林冠层微波透过率半经验模型
杨建卫1, 蒋玲梅1, 武胜利2, 栾英宏3, 潘金梅4, 施建成5
1.北京师范大学/中国科学院空天信息创新研究院遥感科学国家重点实验室;2.国家卫星气象中心;3.上海航天电子技术研究所;4.中国科学院空天信息创新研究院;5.中国科学院国家空间科学中心
摘要:

星载被动微波遥感是获取宏观尺度雪深时空分布的重要手段。森林冠层不仅衰减了来自地表的微波辐射,同时自身也是一个热辐射源,因此森林冠层增加了被动微波遥感反演雪深的不确定性。本研究基于植被辐射传输tau-omega模型(τ-ω)提出了相邻像元(森林和非森林)的冠层微波透过率提取方法,探索冠层微波透过率模型在卫星尺度的应用。该方法假设相邻的森林和非森林像元存在相同的积雪和环境参数,通过联立相邻像元的辐射传输方程从理论上推算冠层微波透过率,进而借助森林生物量建立森林透过率的半经验估算模型。通过对比微波辐射模型模拟亮温和AMSR2卫星观测亮温,发现未经过森林辐射校正的亮温(18.7GHz和36.5GHz)往往存在低估现象,而经过森林辐射校正后的模拟亮温更接近于卫星观测;通过留一法(Leave-One-Out Cross Validation)对发展的透过率半经验模型验证,发现反演的透过率与参考值相关性高于0.7,在18.7GHz和36.5GHz频段的均方根误差RMSE分别为0.0589和0.0787;通过分析高低频亮温差(Tb18.7V-Tb36.5V)与雪深的关系,发现相关系数由森林辐射校正前的0.26提高到校正后的0.46;最后利用传统的经验性雪深反演算法对森林辐射校正效果进行测试,发现雪深反演误差(unRMSE)由原来的8.9cm降低到7.8cm,相关系数由0.32提高到0.49。本研究发展的冠层微波透过率半经验模型可以实现卫星遥感尺度的应用,为提高林区的雪深反演精度提供了参考和支撑。

A semi-empirical microwave transmissivity model for forest canopy in snow season
Abstract:

Space-borne passive microwave remote sensing is a crucial technique to monitor the global temporal-spatial distribution of snow depth. Forest canopy not only attenuates the microwave radiation from soil but also emits some radiation into the sensor. Therefore forest canopy increases the uncertainness of snow depth retrievals with satellite passive microwave remote sensing. This paper aims to develop a microwave transmissivity model at the scale of satellite observations (25km×25km) to realize forest correction to satellite observations. This paper proposes a novel method (hereafter referenced as adjacent pixel approach) to estimate canopy transmissivity, which combines the radiative transfer functions of adjacent forest and open pixels. Then, a semi-empirical transmissivity model depending on the forest biomass is built to correct satellite observed brightness temperatures. To demonstrate the function of proposed transmissivity model in snow depth retrieving, the modeling brightness temperature data are compared to AMSR2 observations in Northeast China. The results indicate that the microwave emission model tends to underestimate brightness temperature against satellite observations due to ignoring forest canopy effects. After correction to AMSR2 observations using the proposed method, model simulations are much closer to AMSR2 observations. We further verify the proposed semi-empirical microwave transmissivity model using Leave-One-Out cross-validation method. Results show that the correlation coefficient between estimates and reference values is as high as 0.7, and the RMSE values are 0.0589 and 0.0787 for 18.7 GHz and 36.5 GHz, respectively. The relationship between brightness temperature spectral difference (Tb18.7V-Tb36.5V) and ground-based snow depth are also improved after forest correction relatively to before, from 0.26 to 0.46. To demonstrate the improvement of forest radiation correction on snow depth retrievals, an empirical retrieval algorithm is selected to test. Results present the RMSE is 7.8 cm, while it is 8.9 cm without forest correction. Moreover, the correlation coefficient increases from 0.32 to 0.49. The proposed semi-empirical transmissivity method significantly improves the performance of the microwave radiative transfer model in forested areas. What"s more, this method can be directly applied to correct satellite-based brightness temperature, which reduces the uncertainness of estimated snow depth values. Thus, this study provides the reference and support for improving snow depth under the forest canopy.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群