曹引,冶运涛,赵红莉,蒋云钟,王浩,严登明.基于离散粒子群和偏最小二乘的水源地浊度高光谱反演[J].农业机械学报,2018,49(1):173-182.
CAO Yin,YE Yuntao,ZHAO Hongli,JIANG Yunzhong,WANG Hao,YAN Dengming.Satellite Hyperspectral Retrieval of Turbidity for Water Source Based on Discrete Particle Swarm and Partial Least Squares[J].Transactions of the Chinese Society for Agricultural Machinery,2018,49(1):173-182.
摘要点击次数: 2443
全文下载次数: 1109
基于离散粒子群和偏最小二乘的水源地浊度高光谱反演   [下载全文]
Satellite Hyperspectral Retrieval of Turbidity for Water Source Based on Discrete Particle Swarm and Partial Least Squares   [Download Pdf][in English]
投稿时间:2017-04-13  
DOI:10.6041/j.issn.1000-1298.2018.01.022
中文关键词:  水源地  浊度  高光谱反演  离散粒子群  偏最小二乘  HJ-1A HSI
基金项目:国家重点研发计划项目(2017YFC0405801、2017YFC0405804)、国家自然科学基金项目(51309254)、高分辨率对地观测系统重大专项(08-Y30B07-9001-13/15-01)、中国水利水电科学研究院科研专项“十三五”重点科研项目(WR0145B272016)和中国水利水电科学研究院流域水循环模拟与调控国家重点实验室开放基金项目(IWHR-SKL-201517)
作者单位
曹引 东华大学 
冶运涛 中国水利水电科学研究院 
赵红莉 中国水利水电科学研究院 
蒋云钟 中国水利水电科学研究院 
王浩 东华大学
中国水利水电科学研究院 
严登明 东华大学 
中文摘要:随着面源污染的加剧,导致水源地水体富营养化程度日趋严重,浊度作为衡量水体富营养化的一项重要指标,是水质评价的重要参数。为降低浊度偏最小二乘(Partial least squares, PLS) 反演模型建模的不确定性,提高模型反演精度,提出了基于离散粒子群(Discrete binary particle swarm optimization, DBPSO)和偏最小二乘的水体浊度反演模型。以2015年10月在南水北调东线重要水源地微山湖获取的水体浊度和准同步的HJ-1A HSI高光谱数据为例,利用HJ-1A HSI B26-B105(中心波长:518~870nm)全谱段光谱反射率(Original spectral reflectance, OSR)和归一化光谱反射率(Normalized spectral reflectance, NSR)直接构建浊度OSR-PLS和NSR-PLS反演模型,同时利用离散粒子群算法优选输入浊度PLS反演模型的最佳原始波段反射率和归一化光谱反射率,在此基础上提出并构建浊度OSR-DBPSO-PLS和NSR-DBPSO-PLS反演模型;然后对上述模型进行精度评价,分析光谱归一化处理和特征波段优选对PLS模型反演精度的影响,选择精度最高的模型反演微山湖水体浊度分布。结果表明:NSR-PLS模型精度(R2=0.91)高于OSR-PLS模型(R2=0.50),对波段进行归一化处理能提高浊度PLS反演模型精度;DBPSO能够识别浊度PLS反演的最佳波段,浊度PLS建模所需的波段数由80个分别减少为44个(OSR波段)和36个(NSR波段),在此基础上构建的OSR-DBPSO-PLS模型(R2=0.96)和NSR-DBPSO-PLS模型(R2=0.97)均具有较高精度,显著高于直接利用全谱波段构建的浊度PLS模型反演精度;选择综合误差最小的NSR-DBPSO-PLS模型反演微山湖水体浊度,反演结果符合实际,该模型适用于HJ-1A HSI数据和内陆水体浊度反演。
CAO Yin  YE Yuntao  ZHAO Hongli  JIANG Yunzhong  WANG Hao  YAN Dengming
Donghua University,China Institute of Water Resources and Hydropower Research,China Institute of Water Resources and Hydropower Research,China Institute of Water Resources and Hydropower Research,Donghua University;China Institute of Water Resources and Hydropower Research and Donghua University
Key Words:water source  turbidity  hyperspectral retrieval  discrete particle swarm  partial least squares  HJ-1A HSI
Abstract:With the increase of non-point source pollution emissions, the degree of eutrophication in water source is becoming seriously increase. Turbidity is an important parameter of water quality assessment, as an indicator of eutrophication. A discrete binary particle swarm optimization-partial least squares (PLS) model was proposed to reduce modeling uncertainty of turbidity retrieval using PLS and improve retrieval accuracy. A discrete binary particle swarm optimization was used to select original spectral reflectance and normalized spectral reflectance of concurrent HJ-1A HSI hyperspectral data with the turbidity obtained from October, 2015 in Weishan Lake as the input of partial least squares model. OSR-PLS and NSR-PLS model retrieving turbidity were developed by using original spectral reflectance and normalized spectral reflectance in full spectrum (HJ-1A HSI B26-B105 with 518nm to 870nm (central wavelength)). Meanwhile, the OSR-DBPSO-PLS and NSR-DBPSO-PLS models were developed to retrieve turbidity by using the selected original spectral reflectance and normalized spectral reflectance. The influence of spectral normalized and the characteristic band optimized on PLS model retrieval accuracy were analyzed based on the four models’ elevation. Finally, the highest accuracy model was used to retrieve the turbidity distribution in Weishan Lake. The results indicated that the accuracy of NSR-PLS (R2=0.91) model was better than that of OSR-PLS model (R2=0.50). The normalization of reflectance can improve PLS accuracy of turbidity retrieval. DBPSO can identify the optimal original and normalize spectral reflectance. The number of bands required for turbidity PLS modelling was reduced from 80 to 44 (OSR) and 36 (NSR), respectively. The OSR-DBPSO-PLS (R2=0.96) and NSR-DBPSO-PLS (R2=0.97) modelling based on 44 OSR and 36 NSR, respectively, had high accuracies, which were significantly higher than the OSR-PLS and NSR-PLS modelling by full spectrum. The NSR-DBPSO-PLS model with minimal comprehensive error was selected to retrieve turbidity in Weishan Lake, which was suitable for inland water turbidity retrieval based on HJ-1A HSI data.

Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.

   下载PDF阅读器