Applying support vector regression to water quality modelling by remote sensing data

This article applies a nonlinear machine learning method, support vector regression (SVR), to construct empirical models retrieving water quality variables using remote sensing images. Based on in situ measurements and high-resolution multispectral SPOT-5 (Satellite Pour l'Observation de la Terre) data, a fittest nonlinear function between input and output was obtained from this method, and SVR model parameters were selected automatically using a genetic algorithm (GA). The relationship between water quality variables – permanganate index (COD_Mn), ammonia-nitrogen (NH₃–N) and chemical oxygen demand (COD) – and spectral components of SPOT-5 data for the Weihe River in China was constructed by the proposed method. Spatial distribution maps for the three water quality variables were also developed. The results show that SVR can implement any nonlinear mapping, and produce better predictions than the traditional statistical multiple regression method, especially when samples are limited. With further testing, SVR can also be extended to hyperspectral remote sensing applications in the management of land and water resources.     

Using remote sensing to estimate sea ice thickness in the Bohai Sea,China based on ice type

It is challenging to use traditional remote sensing techniques to accurately determine the extent and thickness of ice in the Bohai Sea, on account of the presence of sea impurities (i.e. mud, salt bubbles and sand) and shape irregularities. Accordingly, we performed a series of reflectance spectra experiments to empirically link remote measurements of surface reflectance with in situ sea ice thickness measurements in the Bohai Sea. Two years of Thematic Mapper (TM) band 2 and Moderate Resolution Imaging Spectroradiometer (MODIS) band 4 data were used to distinguish between the following sea ice types, using spectral reflectance thresholds of 6.4, 9.6, 10.3 and 12.1%: (a) clean nilas ice (a thin elastic crust of ice up to 10 cm thick that, under pressure, may deform by finger rafting; (b) nilas ice and pancake ice (roughly circular accumulations of frazil ice, usually less than about 3 m in diameter, with raised rims caused by collisions); (c) grey and grey–white ice; and (d) cumulative ice (<30 cm). By establishing a relationship between sea ice type and ice thickness, a novel, practical and low-cost remote sensing technique is introduced to estimate the extent and distribution of sea ice thickness over a large spatial scale. The results obtained by remote sensing are validated with in situ ice shape measurements. The MODIS and TM data are used to distinguish between three ice thickness grades (6–9, 10–20 and 20–30 cm).     

A two-step optimization procedure for assessing water constituent concentrations by hyperspectral remote sensing techniques: An application to the highly turbid Venice lagoon waters

Over the past few years, the increased spectral and spatial resolution of remote sensing equipment has promoted the investigation of new techniques for inland and coastal water monitoring. The availability of new high-resolution data has allowed improvements in models based on the radiative transfer theory for assessing optical water quality parameters. In this study, we fine-tuned a physical model for the highly turbid Venice lagoon waters and developed an inversion technique based on a two-step optimization procedure appropriate for hyperspectral data processing to retrieve water constituent concentrations from remote data. In the first step, the solution of a linearized analytical formulation of the radiative transfer equations was found. In the second step, this solution was used to provide the initial values in a non-linear least squares-based method. This effort represents a first step in the construction of a feasible and timely methodology for Venice lagoon water quality monitoring by remote sensing, especially in view of the existing experimental hyperspectral satellite (Hyperion) and the future missions such as PRISMA, EnMap and HyspIRI. The optical properties of the water constituents were assessed on the basis of sea/lagoon campaigns and data from the literature. The water light field was shaped by an analytical formulation of radiative transfer equations and the application of numerical simulations (Hydrolight software). Once the optical properties of the Venice lagoon bio-optical model were validated, the inverse procedure was applied to local radiometric spectra to retrieve concentrations of chlorophyll, colored dissolved organic matter and tripton. The inverse procedure was validated by comparing these concentrations with those measured in the laboratory from in situ water samples, then it was applied to airborne (CASI and MIVIS) and satellite (Hyperion) sensors to derive water constituent concentration maps. The consistent results encourage the use of this procedure using future missions satellite (PRISMA, EnMap and HyspIRI).     

Comparative analysis of four semi-analytical models for estimating chlorophyll-a concentration in case-2 waters using field hyperspectral reflectance

ABSTRACT

Hyperspectral remote sensing can capture the complicated and variable characteristics of inland waters; thus, it is suited for the water quality assessment of Case-2 waters, and it has the potential to attain high estimation accuracy. In the present study, four improved models adapted from published approaches (three-band index, ΔΦ, BNDBI and TCARI) were investigated to estimate chlorophyll-a (chl-a) for the case of Dianshan Lake, China. Calibration and validation were provided from in situ measured chl-a and field hyperspectral measurements. The improved three-band (ITB) model, ΔΦ model, and BNDBI model yielded satisfactory results and enabled the estimation of chl-a for inland Case-2 waters with coefficients of determination (R²) reaching 0.75, 0.76, and 0.86, respectively. In particular, the TCARI/OSAVI model presented the highest accuracy (R² = 0.94) compared to the other models. All of the results provide strong evidence that the hyperspectral models presented in this paper are promising and applicable to estimate chl-a in eutrophic inland Case-2 waters.     

新庙泡叶绿素a浓度高光谱定量模型研究

利用吉林省新庙泡的高光谱实测数据和水质采样分析数据,尝试通过单波段、波段比值、一阶微分和峰谷间距法建立叶绿素a反演模型。结果表明:单波段光谱反射率与叶绿素a浓度的相关性较差,不宜用于该区域的叶绿素a浓度估算;680 nm和700 nm波段反射率之比、700 nm处光谱一阶微分值和两波段峰谷间距反演模型都具有较高的决定系数,分别为0.783 4、0.792 7、0.796 9,验证模型的决定系数为0.651 3、0.431 7、0.756 4,均方根误差分别为8.69μg·L-1、14.50μg·L-1、10.04μg·L-1,显著水平P<0.01。这3种方法皆可以用于新庙泡叶绿素a浓度的定量遥感,其中又以峰谷间距法为最优。     

An optimal model for estimating suspended sediment concentration from Landsat TM images in the Caofeidian coastal waters

Suspended sediment concentration (SSC) is one of the most critical parameters in water quality and environmental evaluations. Remote sensing has the potential for monitoring the dynamics and spatial distribution of SSC efficiently. The primary objective of this study is to develop retrieval models that are reliable and sensitive to SSC levels in the Caofeidian area, a new seaport in northeast China, based on Landsat-5 Thematic Mapper (TM) images and a set of in situ data sets, including spectral reflectance data and water quality data. The study finds that the band reflectance ratio and binary combination factor (i.e. the ratio of the reflectance to the particle size) are more effective than single band reflectance, and a non-linear model is more potent than a linear model for predicting SSC in the Caofeidian waters. A quadratic polynomial regression model of the R_TM3/R_TM2 ratio is proposed as the optimal retrieval model after evaluating various models with respect to different sensitive factors. The accuracy of the model is acceptable with a relative error and a root mean square error of 25.35% and 7.22 mg l^–1, respectively; the correlation coefficient between the observed and estimated SSCs is 0.986. This study also indicates that the band reflectance ratio and binary combination factor are effective in weakening and even partially eliminating the effects of the changes in the sediment type (i.e. particle size and refractive index). And the band reflectance ratio is more efficient. Using the proposed model and TM data, SSC levels for the entire region were estimated. Such results can serve as a baseline for future environmental monitoring efforts.     

水质遥感监测的关键要素叶绿素a的反演算法研究进展

叶绿素a浓度是表征水体富营养化程度的重要指标,通过遥感手段反演叶绿素a浓度是实现水体富营养化监测的一个有效途径,已衍生出了一系列叶绿素a浓度反演算法.这些算法各有所长,适用范围也各自有别.由于水体光学特征差异,盲目套用这些算法难以取得预期效果.为了推动水质遥感的进一步发展,从遥感反演的原理和数据源出发,对国内外利用遥感...     

HSRS-SC:面向遥感场景分类的高光谱图像数据集

目的 场景分类是遥感领域一项重要的研究课题，但大都面向高分辨率遥感影像。高分辨率影像光谱信息少，故场景鉴别能力受限。而高光谱影像包含更丰富的光谱信息，具有强大的地物鉴别能力，但目前仍缺少针对场景级图像分类的高光谱数据集。为了给高光谱场景理解提供数据支撑，本文构建了面向场景分类的高光谱遥感图像数据集（hyperspectral remote sensing dataset for scene classification，HSRS-SC）。方法 HSRS-SC来自黑河生态水文遥感试验航空数据，是目前已知最大的高光谱场景分类数据集，经由定标系数校正、大气校正等处理形成。HSRS-SC分为5个类别，共1 385幅图像，且空间分辨率较高（1 m），波长范围广（380~1 050 nm），同时蕴含地物丰富的空间和光谱信息。结果 为提供基准结果，使用AlexNet、VGGNet-16、GoogLeNet在3种方案下组织实验。方案1仅利用可见光波段提取场景特征。方案2和方案3分别以加和、级联的形式融合可见光与近红外波段信息。结果表明有效利用高光谱影像不同波段信息有利于提高分类性能，最高分类精度达到93.20%。为进一步探索高光谱场景的优势，开展了图像全谱段场景分类实验。在两种训练样本下，高光谱场景相比RGB图像均取得较高的精度优势。结论 HSRS-SC可以反映详实的地物信息，能够为场景语义理解提供良好的数据支持。本文仅利用可见光和近红外部分波段信息，高光谱场景丰富的光谱信息尚未得到充分挖掘。后续可在HSRS-SC开展高光谱场景特征学习及分类研究。     

Measurements and analysis of in situ multi-angle reflectance of turbid inland water: a case study in Meiliang Bay,Taihu Lake,China

The bidirectional reflectance properties of the anisotropic light field above the water surface are important for a range of applications. The bidirectional reflectance distribution function of oceanic waters has been well characterized but there is a lack of information for turbid inland waters. In addition, there is a lack of bidirectional reflectance data measured in turbid inland waters partially due to the difficulty in collecting in situ water-surface multi-angle remote-sensing reflectance data. To facilitate bidirectional reflectance studies of turbid inland waters using in situ multi-angular reflectance data, we have designed and developed a simple hand-held 3D positioning pole to position the spectrometer optical fibre probe and a specific method to collect the multi-angular reflectance data above the water surface with this pole. Using this device, we collected multi-angular reflectance data in Meiliang Bay, Taihu Lake, China, and analysed the uncertainties in this method. We analysed the bidirectional distribution characteristics of the data, and compared the findings to those in the literature. Both uncertainty analysis and bidirectional distribution characteristics analysis showed that our method is effective in collecting multi-angular reflectance above the water surface and can be applied to validate bidirectional correction models in the future.     

Soil water and plant canopy effects on remotely measured surface temperatures

Abstract.

Thermal infrared remote sensing of diurnal crop canopy temperature variations represents a possible method for determining the availability of soil water to plants. This study was performed to assess the effects of soil water and crop canopy on apparent temperatures observed by means of remote sensors, and to determine the impact of these effects on remote soil water monitoring. Airborne thermal scanner and apparent reflectance data (one date) and ground PRT-5 data (three dates) were collected primarily over barley and other small grain canopies. Plant heights, cover, and available soil water for four layers in the top 20 cm were determined. Analysis of the data showed a close inverse linear relationship between the available water and the day minus night temperature difference δT, for thick barley canopies (plant cover above 90 per cent) only. The use of apparent reflectance values in the visible region did not improve available soil water regression equations substantially. These results suggest that the available water or plant stress could only be accurately determined for thick canopies, and that the reflectance data could probably be used to identify such canopies but would not improve regression estimates of soil water from remote sensing data.     

利用MERIS产品数据反演太湖叶绿素a浓度研究

第三代水色传感器MERIS的荧光通道的合理设置为荧光遥感法的应用提供了广阔的发展前景。利用MERIS数据、同步地面光谱和水质监测数据,分别通过基线荧光高度（FLH）、归一化荧光高度（NFH）和最大叶绿素指数（MCI）建立了太湖叶绿素a浓度的荧光遥感估算模型。结果表明：MERIS荧光参数中最大叶绿素指数（MCI）较基线荧光高度（FLH）更适合太湖水体叶绿素a浓度的反演;归一化荧光高度（NFH）与实测叶绿素a浓度间的拟合效果最好。最后选取NFH进行MERIS荧光遥感模型的太湖叶绿素a浓度的反演,其结果客观地反映了太湖水体叶绿素a浓度的空间分布格局。     

Mapping the concentrations of total suspended matter in Lake Taihu,China, using Landsat‐5 TM data

Remote sensing techniques can be used to estimate and map the concentrations of suspended matter in inland water, providing both spatial and temporal information. Although an empirical approach to remote sensing of inland waters has been carried out frequently, satellite imagery has not been incorporated into routine lake monitoring programmes due in part to the lack of a standard prediction equation with multi‐temporal capacity for suspended matter. Empirical and physical models must be developed for each lake and its corresponding turbidity composition if they are to be compared over time, or with other bodies of water.

This study aimed to develop and apply multi‐temporal models to estimate and map the concentrations of total suspended matter (TSM) in Lake Taihu, China. Two Landsat‐5 Thematic Mapper (TM) images and nearly contemporaneous in situ measurements of TSM were used. A modified Dark‐Object Subtraction (DOS) method was used, and appeared to be adequate for atmospheric correction. The relationships were examined between TSM concentrations and atmospherically corrected TM band and band ratios. Results of this study show that the ratio TM4/TM1 has a strong relationship with TSM concentrations for lake waters with relatively low concentrations of phytoplankton algae. However, TM3 provided a strong predictive relationship with TSM concentrations despite varied water quality conditions. Different prediction models were developed and compared using multiple regression analysis. The Akaike Information Criteria (AIC) approach was used to choose the best models. The validation of the multi‐temporal capability of the best models indicated that it is feasible to apply the linear regression model using TM3 to estimate TSM concentrations across time in Lake Taihu, even if no in situ data were available.     

Analytical algorithms for lake water TSM estimation for retrospective analyses of TM and SPOT sensor data

Suspended matter in inland waters is related to total primary production and fluxes of heavy metals and micropollutants such as PCBs. Synoptic information on suspended matter cannot be obtained from an in situ monitoring network since suspended matter is a spatially inhomogeneous parameter. This problem can be solved by the integrated use of remote sensing data, in situ data and water quality models. To enable retrospective model and remote sensing data comparison of suspended matter concentration and distribution, a methodology is required for processing satellite images that is independent of in situ measurements. Analytical optical modelling, based on knowledge of the in situ inherent optical properties, leads to reliable multi-temporal algorithms for estimating suspended matter concentration in lakes for the data from the SPOT and Landsat TM sensors. This methodology allows multi-temporal, multi-site and multi-instrument comparison of TSM maps derived from satellite imagery. This means that satellite sensor data can now become an independent measurement tool for water management authorities. The remote sensing maps showed that large gradients in TSM were observed for the various lakes as well as temporal changes of these spatial gradients. In situ point samples are shown to be not representative for suspended matter in the lakes.     

Satellite-based water quality monitoring in Lake Vänern,Sweden

Lake Vänern, Sweden, is one of Europe’s largest lakes and has a historical, cultural, ecological as well as economic importance. Lake water quality monitoring is required by national and international legislations and directives, but present programmes are insufficient to meet the requirements. To complement in situ based monitoring, the possibility to obtain reliable information about spatial and temporal water quality trends in Lake Vänern from the ENVISAT mission’s MERIS instrument was evaluated. The complete archive (2002–2012) of MERIS (Medium Resolution Imaging Spectrometer) full resolution data was processed using the water processor developed by Free University Berlin (FUB) to derive aerosol optical thickness (AOT), remote-sensing reflectance (R_rs) and water quality parameters: chlorophyll-a (chl-a) concentration, coloured dissolved organic matter absorption at 443 nm (CDOM), and total suspended matter (TSM) concentration. The objective was to investigate if, either, FUB reflectance products in combination with potential lake-specific band ratio algorithms for water quality estimation, or directly, FUB water quality products, could complement the existing monitoring programme.

Application of lake-specific band ratio algorithms requires high-quality reflectance products based on correctly estimated AOT. The FUB reflectance and AOT products were evaluated using Aerosol Robotic Network – Ocean Color (AERONET-OC) match-up data measured at station Pålgrunden in Lake Vänern. The mean absolute percentage differences (MAPDs) of the final reflectance retrievals at 413, 443, 490, 555, and 665 nm were 510%, 48%, 33%, 34%, and 33%, respectively, corresponding to a large positive bias in 413 nm, positive bias in 443–555 nm, and a negative bias in 665 nm. AOT was strongly overestimated in all bands.

The FUB water quality products were evaluated using match-up in situ data of chl-a, filtered absorbance (Abs_F(420)) and turbidity as Abs_F(420) is related to CDOM and turbidity is strongly related to TSM. The in situ data was collected within the Swedish national and regional monitoring programmes. In order to widen the range of water constituents and add more data to the analysis, data from four large Swedish lakes (Vänern, Vättern, Mälaren, and Hjälmaren) was included in the analysis. High correlation (r ≥ 0.85) between in situ data and MERIS FUB derived water quality estimates were obtained, but the absolute levels were over- (chl-a) or under- (CDOM) estimated. TSM was retrieved without bias.

Calibration algorithms were established for chl-a and CDOM based on the match-up data from all four lakes. After calibration of the MERIS FUB data, realistic time series could be derived that were well in line with in situ measurements. The MAPDs of the final retrievals of chl-a, Abs_F(420) and Turbidity in Lake Vänern were 37%, 15%, and 35%, respectively, corresponding to mean absolute differences (MADs) of 0.9 µg l^?1, 0.17 m^?1, and 0.32 mg l^?1 in absolute values.

The partly inaccurate reflectance estimations in combination with both positive and negative bias imply that successful application of band ratio algorithms is unlikely. The high correlation between MERIS FUB water quality products and in situ data, on the other hand, shows a potential to complement present water quality monitoring programmes and improve the understanding and representability of the temporally and spatially sparse in situ observations. The monitoring potential shown in this study is applicable to the Sentinel-3 mission’s OLCI (Ocean Land Colour Instrument), which was launched by the European Space Agency (ESA) in February 2016 as a part of the EC Copernicus programme.     

A current review of empirical procedures of remote sensing in inland and near-coastal transitional waters

The empirical approach of remote sensing has a proven capability to provide timely and accurate information on inland and near-coastal transitional waters. This article gives a thorough review of empirical algorithms for quantitatively estimating a variety of parameters from space-borne, airborne and in situ remote sensors in inland and transitional waters, including chlorophyll-a, total suspended solids, Secchi disk depth (z _SD), turbidity, absorption by coloured dissolved organic matter (a _CDOM) and other parameters, for example, phycocyanin. Current remote-sensing instruments are also reviewed. The theoretical basis of the empirical algorithms is given using fundamental bio-optical theory of the inherent optical properties (IOPs). Bands, band ratios and band arithmetic algorithms that could be used to produce common biogeophysical products for inland/transitional waters are identified. The article discusses the potential role that empirical algorithms could play alongside more advanced model-based algorithms in the future of water remote sensing, especially for near real-time operational monitoring systems. The article aims to describe the current status of empirical remote sensing in inland and near-coastal transitional waters and provide a useful reference to workers. It does not cover ‘inversion’ algorithms.     

基于Hyperion影像的高光谱水汽含量反演的算法比较研究

基于高光谱遥感图像数据的大气参数反演和一体化辐射校正具有重要研究意义和应用价值。首先,通过6S模型辐射传输计算分析了EO-1/Hyperion遥感影像在940和1 130nm附近水汽吸收区域的光谱吸收特点。其次,采用两通道比值法和三通道比值法,比较了不同波段组合的大气含水量高光谱遥感反演精度并进行了敏感性分析,模拟实验结果表明采用三波段比值算法的相关系数和均方根误差均优于对应的两波段算法。最后,利用张掖地区2008年3景EO-1Hyperion高光谱遥感影像,反演了大气含水量,并与地基CE-318太阳分光光度计测量数据进行对比验证,结果表明:1 124nm水汽吸收通道反演精度优于940nm,两通道和三通道比值法的均方根误差分别为0.369和0.128g/cm2,三通道比值方法优于两通道比值方法,与地面观测结果一致。     

Bio‐optical model with optimal parameter suitable for Taihu Lake in water colour remote sensing

Sixty‐seven samplings were collected, almost covering all over the Taihu Lake, in one campaign in October 2004. At each station, the backscattering coefficients and the field spectra were measured in situ, respectively, with a HS‐6 and a FieldSpec 931 spectroradiometer (ASD Inc.). Almost concurrently, water samples were fetched with Niskin water‐fetching equipment and then returned to the laboratory for concentration and absorption measurement. The whole lake was divided into different areas according to some indexes. Three models were used to calculate remote sensing reflectance R _rsc for the waters where the in situ remote sensing reflectance R _rsm was beyond the bottom effect, which was considered as optically deep waters. By comparison of R _rsc and R _rsm, the best model suitable for optically deep waters in Taihu Lake, together with its optimal experiential parameter, were selected and developed, which was very important and helpful to develop a universal model to estimate accurately remote sensing reflectance for the whole lake in the next step.     

Upscaling of sparse in situ soil moisture observations by integrating auxiliary information from remote sensing

Upscaling of sparse in situ soil moisture (SM) observations is essential for the validation of current and upcoming space-borne SM retrievals, and the successful application of SM observations in hydrological models or data assimilation. In this study, we construct a novel method based on Bayesian data fusion to upscale in situ SM observations to the coarse scale of microwave remote sensing. In the framework of Bayesian theory, the valuable auxiliary information obtained in Moderate Resolution Imaging Spectroradiometer (MODIS) apparent thermal inertia (ATI) is integrated into the upscaling process. The method is validated using SM wireless sensor network data in the Tibetan plateau, which covers an area of approximately 30 × 30 km² with 20 in situ stations. Results confirm that the upscaled SM using the method with randomly selected three stations from the 20 stations is extremely close to the mean of the 20 SMs. The mean root mean square error (RMSE) between the upscaled SM and the mean of the 20 in situ SMs was 0.02 m³ m^?3, and the max RMSE was less than 0.05 m³ m^?3. Furthermore, the sensitivity of the upscaling accuracy to the number of in situ observations is discussed. When the number of in situ observations is greater than nine, the increasing accuracy of the Bayesian method is limited by the uncertainty in the ATI of the remote sensing.     

高光谱场景辐亮度模型仿真研究

高光谱遥感的地面场景是高光谱遥感系统中影响因素最复杂多变的部分。首先基于星载高光谱遥感成像的辐射传输过程,对非均匀的朗伯表面的入瞳处大气辐亮度传输模型进行了研究,得到只需要考虑目标与邻近像元反射率,大气传输因子的辐亮度简化模型。之后介绍了大气中光子扩散原理,并采用蒙特卡洛方法对大气点扩散函数进行仿真;联合地表目标像元反射率数据计算得到基于非均匀朗伯面地表的邻近像元反射率;然后总结了大气传输模型软件MODTRAN计算入瞳处辐亮度数据的原理步骤,并利用其反演了朗伯表面的相关大气传输参数。最终利用基于传感器入瞳处的辐亮度数据表征了高光谱地面场景。     

Using hyperspectral data to quantify water-quality parameters in the Wabash River and its tributaries,Indiana

A hand-held spectrometer was used to collect above-water spectral measurements for measuring optically active water-quality characteristics of the Wabash River and its tributaries in Indiana. Water sampling was undertaken concurrent with spectral measurements to estimate concentrations of chlorophyll (chl) and total suspended solids (TSS). A method for removing sky and Sun glint from field spectra for turbid inland waters was developed and tested. Empirical models were then developed using the corrected field spectra and in situ chl and TSS data. A subset of the field measurements was used for model development and the rest for model validation. Spectral characteristics indicative of waters dominated by different inherent optical properties (IOPs) were identified and used as the basis of selecting bands for empirical model development. It was found that the ratio of the reflectance peak at the red edge (704 nm) with the local minimum caused by chl absorption at 677 nm was a strong predictor of chl concentrations (coefficient of determination (R²) = 0.95). The reflectance peak at 704 nm was also a good predictor for TSS estimation (R² = 0.75). In addition, we also found that reflectance within the near-infrared (NIR) wavelengths (700–890 nm) all showed a strong correlation (0.85–0.91) with TSS concentrations and generated robust models. Results suggest that hyperspectral information provided by field spectrometer can be used to distinguish and quantify water-quality parameters under complex IOP conditions.