首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 261 全文下载次数: 129
引用本文:

DOI:

10.11834/jrs.20210097

收稿日期:

2020-04-09

修改日期:

2020-07-09

PDF Free   EndNote   BibTeX
基于植被二向性反射统一模型的高分五号FAPAR遥感反演算法
田定方1, 杨斯棋2, 徐大伟3, 任华忠2, 范闻捷2, 刘镕源4
1.北京大学遥感与GIS研究所;2.北京大学地球与空间科学学院遥感与地理信息系统研究所;3.中国科学院农业资源与农业区划研究所/呼伦贝尔草原生态系统国家野外科学研究观测站;4.中国自然资源航空物探遥感中心
摘要:

植被光合有效辐射吸收比率(FAPAR)是描述植被光合作用能量交换过程的重要参数,广泛应用于植被长势监测、植被生产力估算、全球变化等研究领域,遥感是大范围获取FAPAR的唯一途径。与多光谱传感器相比,高光谱传感器能更加精确、细致地观测植被的光谱特征,有利于分析植被冠层反射、吸收特性,进而反演植被冠层FAPAR。为利用高光谱反射率数据实现FAPAR定量反演,本文首先在植被BRDF统一模型和FAPAR-P模型的基础上,构建了BRDF-FAPAR统一模型UBFM(Unified BRDF-FAPAR Model);进而基于高分五号高光谱传感器特征模拟了不同情况下植被冠层反射率和相应的FAPAR;然后运用改进的最佳指数法选择FAPAR反演的特征波段组合;在此基础上,将特征波段组合反射率与FAPAR模拟结果作为神经网络的输入参数,构建FAPAR神经网络反演算法。研究结果表明,改进的最佳指数法能有效地筛选出FAPAR估算的敏感波段,随着输入波段数的增加,神经网络估算的精度逐渐提高,综合考虑波段信息量和实际影像数据噪声影响,本研究针对高分五号高光谱传感器选择8个波段作为FAPAR反演特征波段,基于UBFM模型构建的神经网络反演精度较好,模拟实验算法误差约为0.014。选择内蒙古呼伦贝尔是谢尔塔拉草原为主要研究区,从高分五号高光谱影像中反演了研究区的FAPAR,并利用同步地面实测数据验证,反演误差为0.048。该算法简化了传统模型机理方法的中间环节和多个参数设置,有较好的可行性、稳定性和精度,为国产卫星高光谱传感器地表植被参数定量反演提供了新途径。

FAPAR Retrieval from Gaofen-5 hypserctral images based on Unified BRDF Model
Abstract:

The fraction of absorbed photosynthetically active radiation (FAPAR) is a key parameter to characterize photosynthesis process of vegetation and is widely used in many study areas such as vegetation monitoring, NPP estimation and global changing. Remote sensing provides the only way to get FAPAR at large scale. Comparing with multispectral instrument, hyperspectral instrument has an advantage in analyzing the canopy reflectance and absorption based on the high accuracy of spectrum measurement, which is important to FAPAR retrieval. This study developed a new FAPAR retrieval algorithm for Chinese Gaofen-5(GF-5) Visible-shortwave Infrared Advanced Hyperspectral Imager (AHSI) data, based on BRDF unified model and neural network(NNT). The validation was made in Hulun Buir Xeltala which is a grassland and farming-pastoral area in Inner Mongolia. First, the simulated GF-5 AHSI reflectance-FAPAR datasets were generated by the BRDF unified model and the characteristic of the data set was analyzed. Five groups of input bands of NNT were selected based on Optimal Index Factor (OIF) and a new factor OIFR which is modified by the relevance of band reflectance and FAPAR, separately. Different groups of bands were used to build NNT and the results were assessed by a test set in the simulation dataset. Finally, the best feature bands and NNT were chosen to generate the FAPAR map of the study area from GF-5 AHSI image. Validation with in-situ observations was made. Overall, it was found the new factor OIFR is more efficient than the origin factor OIF in band selection. As the amount of input bands increase, the NNT gets higher accuracy gradually, but the trend stops when the amount reaches a certain level. Considering both band information and instrument noise, total 8 bands were selected as the feature bands of FAPAR retrieval with the FAPAR RMSE of NNT is 0.014. FAPAR map of the study area was generated and the comparison with in-situ FAPAR showed the applicability of the method with RMSE=0.048. It is feasible to analyze the reflectance and absorption by hyperspectral data when the NNT reduced the middle term and parameters of traditional methods simultaneously, which brought a new approach to surface parameters retrieval of domestic satellite hyperspectral instruments.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群 分享按钮