Neural network estimation of LAI, fAPAR, fCover and LAI×Cab, from top of canopy MERIS reflectance data: Principles and validation

A neural network is developed to operationally estimate biophysical variables over land surfaces from the observations of the ENVISAT-MERIS instrument: the leaf area index (LAI), the fraction of absorbed photosynthetically active radiation (fAPAR), the fraction of vegetation cover (fCover), and the canopy chlorophyll content (LAI×C_ab). The neural network requires as input the geometry of observation and the top of canopy reflectances, corrected from the atmospheric effects, in eleven spectral bands. It is trained on a reflectance database made of radiative transfer model simulations. The principles underlying the generation of the database and the design of the network are first presented. The estimated variables are then compared to other existing products, LAI- and fAPAR-MODIS and MGVI-MERIS, and validated against ground measurements performed in the framework of the VALERI project. Results show remarkable consistency of the temporal dynamics between the several products with however some differences in the range of variation. When compared to actual VALERI ground measurements, the proposed algorithm shows the best performances for LAI (RMSE = 0.47) and fAPAR (RMSE = 0.09).     

Estimating the altitude of aerosol plumes over the ocean from reflectance ratio measurements in the O2 A-band

A methodology is proposed to infer the altitude of aerosol plumes over the ocean from reflectance ratio measurements in the O₂ absorption A-band (759 to 770 nm). The reflectance ratio is defined as the ratio of the reflectance in a first spectral band, strongly attenuated by O₂ absorption, and the reflectance in a second spectral band, minimally attenuated. For a given surface reflectance, simple relations are established between the reflectance ratio and the altitude of an aerosol layer, as a function of atmospheric conditions and the geometry of observation. The expected accuracy for various aerosol loadings and models is first quantified using an accurate, high spectral resolution, radiative transfer model that fully accounts for interactions between scattering and absorption. The method is developed for POLDER and MERIS, satellite sensors with adequate spectral characteristics. The simulations show that the method is only accurate over dark surfaces when aerosol optical thickness at 765 nm is relatively large (> 0.3). In this case, the expected accuracy is on the order of ± 0.5 km or ± 0.2 km for POLDER or MERIS respectively. More accurate estimates are obtained with MERIS, since in this case the spectral reflectance ratio is more sensitive to aerosol altitude. However, a precise spectral calibration is needed for MERIS. The methodology is applied to MERIS and POLDER imagery acquired over marine surfaces. The estimated aerosol altitude is compared with in situ lidar profiles of backscattering coefficient measured during the AOPEX-2004 experiment for MERIS, or obtained with the space-borne lidar CALIOP for POLDER. The retrieved altitudes agree with lidar measurements in a manner consistent with theory. These comparisons demonstrate the potential of the differential absorption methodology for obtaining information on aerosol altitude over dark surfaces.     

太湖水体悬浮颗粒物生物光学模型及MERIS数据反演

根据太湖悬浮颗粒物生物光学特性建立的悬浮颗粒物红外单波段生物光学模型具有明确的机理性.利用2006年至2009年野外实测悬浮颗粒物浓度和水体光学数据对该生物光学模型进行检验和分析.结果表明,悬浮颗粒物比后向散射系数的时空差异是影响总悬浮颗粒物生物光学模型精度的主要因素.总悬浮颗粒物在红外波长的吸收系数对总悬浮颗粒物生物光学模型精度也具有显著的影响.根据中分辨率光谱成像仪(MERIS)悬浮颗粒物 反演结果和野外实测风速数据,太湖沉积物再悬浮将显著增加水体中悬浮颗粒物的含量.     

NIR-red reflectance-based algorithms for chlorophyll-a estimation in mesotrophic inland and coastal waters: Lake Kinneret case study

A variety of models have been developed for estimating chlorophyll-a (Chl-a) concentration in turbid and productive waters. All are based on optical information in a few spectral bands in the red and near-infra-red regions of the electromagnetic spectrum. The wavelength locations in the models used were meticulously tuned to provide the highest sensitivity to the presence of Chl-a and minimal sensitivity to other constituents in water. But the caveat in these models is the need for recurrent parameterization and calibration due to changes in the biophysical characteristics of water based on the location and/or time of the year. In this study we tested the performance of NIR-red models in estimating Chl-a concentrations in an environment with a range of Chl-a concentrations that is typical for coastal and mesotrophic inland waters. The models with the same spectral bands as MERIS, calibrated for small lakes in the Midwest U.S., were used to estimate Chl-a concentration in the subtropical Lake Kinneret (Israel), where Chl-a concentrations ranged from 4 to 21 mg m⁻³ during four field campaigns. A two-band model without re-parameterization was able to estimate Chl-a concentration with a root mean square error less than 1.5 mg m⁻³. Our work thus indicates the potential of the model to be reliably applied without further need of parameterization and calibration based on geographical and/or seasonal regimes.     

Merged satellite ocean color data products using a bio-optical model: Characteristics, benefits and issues

The characteristics and benefits of ocean color merged data sets created using a semi-analytical model and the normalized water-leaving radiance observations from the SeaWiFS, MODIS-AQUA and MERIS ocean color missions are presented. Merged data products are coalesced from multiple mission observations into a single data product with better spatial and temporal coverage than the individual missions. Using the data from SeaWiFS, MODIS-AQUA and MERIS for the 2002-2009 time period, the average daily coverage of a merged product is ∼ 25% of the world ocean which is nearly twice that of any single mission's observations. The frequency at which a particular area is sampled from space is also greatly improved in merged data as some areas can be sampled as frequently as 64% of the time (in days). The merged data presented here are validated through matchup analyses and by comparing them to the data sets obtained from individual missions. Further, a complete error budget for the final merged data products was developed which accounts for uncertainty associated with input water-leaving radiances and provides uncertainty levels for the output products (i.e. the chlorophyll concentration, the combined dissolved and detrital absorption coefficient and the particulate backscattering coefficient). These merged products and their uncertainties at each pixel were developed within the NASA REASON/MEaSUREs and ESA GlobColour projects and are available to the scientific community. Our approach has many benefits for the creation of unified Climate Data Records from satellite ocean color observations.     

Neural network estimation of chlorophyll a from MERIS full resolution data for the coastal waters of Galician rias (NW Spain)

In typical Case 2 waters, accurate remote sensing retrieval of chlorophyll a (chla) is still a challenging task. In this study, focusing on the Galician rias (ΝW Spain), algorithms based on neural network (NN) techniques were developed for the retrieval of chla concentration in optically complex waters, using Medium Resolution Imaging Spectrometer (MERIS) data. There is considerable interest in the accurate estimation of chla for the Galician rias, because of the economic and social importance of the extensive culture of mussels, and the high frequency of harmful algal events. Fifteen MERIS full resolution (FR) cloud-free images paired with in situ chla data (for 2002-2004 and 2006-2008) were used for the development and validation of the NN. The scope of NN was established from the clusters obtained using fuzzy c-mean (FCM) clustering techniques applied to the satellite-derived data. Three different NNs were developed: one including the whole data set, and two others using only points belonging to one of the clusters. The input data for these latter two NNs was chosen depending on the quality level, defined on the basis of quality flags given to each data set. The fitting results were fairly good and proved the capability of the tool to predict chla concentrations in the study area. The best prediction was given for the NN trained with high-quality data using the most abundant cluster data set. The performance parameters in the validation set of this NN were R² = 0.86, mean percentage error (MPE) = − 0.14, root mean square error (RMSE) = 0.75 mg m^− 3, and relative RMSE = 66%. The NN developed in this study detected accurately the peaks of chla, in both training and validation sets. The performance of the Case-2-Regional (C2R) algorithm, routinely used for MERIS data, was also tested and compared with our best performing NN and the sea-truthing data. Results showed that this NN outperformed the C2R, giving much higher R² and lower RMSE values.This study showed that the combination of in situ data and NN technology improved the retrieval of chla in Case 2 waters, and could be used to obtain more accurate chla maps. A local-based algorithm for the chla retrieval from an ocean colour sensor with the characteristics of MERIS would be a great support in the quantitative monitoring and study of harmful algal events in the coastal waters of the Rias Baixas. The limitations and possible improvements of the developed chla algorithms are also discussed.     

Ocean transparency from space: Validation of algorithms estimating Secchi depth using MERIS, MODIS and SeaWiFS data

Ocean transparency, often measured using Secchi disk, is a useful index of water quality or productivity and is used in many environmental studies. The spaceborne ocean color sensors provide synoptic and regular radiometric data and can be used for applying environmental policies if the data is converted into relevant biogeochemical properties. We adapted and developed semi-analytical and empirical algorithms to estimate the Secchi depth from satellite ocean color data in both coastal and oceanic waters. The development of the algorithms is based on the use of a comprehensive in situ bio-optical dataset. The algorithms are validated using an extensive set of coincident satellite estimates and in situ measurements of the Secchi depth (so-called matchups). More than 400 matchups are compiled for the MERIS, MODIS and SeaWiFS sensors. The comparison between Secchi depth retrievals from remote sensing data and in situ measurements yields determination coefficients (R²) between 0.50 and 0.73, depending on the sensor and algorithm. The type II linear regression slopes and intercepts vary between 0.95 and 1.46, and between − 0.8 and 6.2 m, respectively. While semi-analytical algorithms provide the most promising results on in situ data, the empirical one proves to be more robust on remote sensing data because it is less sensitive to error due to erroneous atmospheric corrections. Using ocean color archives, one can derive maps of ocean transparency for different areas. Our climatology of the Secchi depth based on ocean color for the transition zone between the North Sea and Baltic Sea is compared to an historical dataset.     

Atmospheric correction of ENVISAT/MERIS data over inland waters: Validation for European lakes

Traditional methods for aerosol retrieval and atmospheric correction of remote sensing data over water surfaces are based on the assumption of zero water reflectance in the near-infrared. Another type of approach which is becoming very popular in atmospheric correction over water is based on the simultaneous retrieval of atmospheric and water parameters through the inversion of coupled atmospheric and bio-optical water models. Both types of approaches may lead to substantial errors over optically-complex water bodies, such as case II waters, in which a wide range of temporal and spatial variations in the concentration of water constituents is expected. This causes the water reflectance in the near-infrared to be non-negligible, and that the water reflectance response under extreme values of the water constituents cannot be described by the assumed bio-optical models. As an alternative to these methods, the SCAPE-M atmospheric processor is proposed in this paper for the automatic atmospheric correction of ENVISAT/MERIS data over inland waters. A-priori assumptions on the water composition and its spectral response are avoided by SCAPE-M by calculating reflectance of close-to-land water pixels through spatial extension of atmospheric parameters derived over neighboring land pixels. This approach is supported by the results obtained from the validation of SCAPE-M over a number of European inland water validation sites which is presented in this work. MERIS-derived aerosol optical thickness, water reflectance and water pigments are compared to in-situ data acquired concurrently to MERIS images in 20 validation match-ups. SCAPE-M has also been compared to specific processors designed for the retrieval of lake water constituents from MERIS data. The performance of SCAPE-M to reproduce ground-based measurements under a range of water types and the ability of MERIS data to monitor chlorophyll-a and phycocyanin pigments using semiempirical algorithms after SCAPE-M processing are discussed. It has been found that SCAPE-M is able to provide high accurate water reflectance over turbid waters, outperforming models based on site-specific bio-optical models, although problems of SCAPE-M to cope with clear waters in some cases have also been identified.     

Using MERIS full resolution data to monitor coastal waters — A case study from Himmerfjärden, a fjord-like bay in the northwestern Baltic Sea

In this paper we investigate if MERIS full resolution (FR) data (300 m) is sufficient to monitor changes in optical constituents in Himmerfjärden, a fjord-like, north–south facing bay of about 30 km length and 4 km width. The MERIS FR products were derived using a coastal processor (FUB Case-2 Plug-In). We also compared the performance between FUB and standard processor (MEGS 7.4), using reduced resolution (RR) data (1 km resolution) from the open Baltic Sea, and compared the products to sea-truthing data. The optical variables measured for sea-truthing were chlorophyll, suspended particulate matter (SPM), as well as coloured dissolved organic matter (CDOM, also termed yellow substances), and the spectral diffuse attenuation coefficient, K_d(490). The comparison of the RR data to the sea-truthing data showed that, in the open Baltic Sea, the MERIS standard processor overestimated chlorophyll by about 59%, and SPM by about 28%, and underestimated yellow substance by about 81%, whereas the FUB processor underestimated SPM by about 60%, CDOM by about 78%, and chlorophyll a by about 56%. The FUB processor showed a relatively high precision for all optical components (standard deviation: 6–18%), whereas the precision for the MEGS 7.4 was rather low (standard deviation: 43–73%), except for CDOM (standard deviation: 13%). The analysis of the FR data showed that all FR level 2 water products derived from MERIS followed a polynomial decline in concentration when moving off-shore. The distribution of chlorophyll and SPM was best described by a 2nd order polynomial, and the distribution of CDOM by a 3rd order polynomial, verifying the diffusional model described in Kratzer and Tett [Kratzer, S. and Tett, P. (in press). Using bio-optics to investigate the extent of coastal waters — a Swedish case study. Hydrobiologia.]. A new K_d(490) and Secchi depth algorithm based on MERIS channel 3 (490 nm) and channel 6 (620 nm) each was derived from radiometric sea-truthing data (TACCS, Satlantic). Applying the K_d(490) algorithm to the MERIS FR data over Himmerfjärden, and comparing to sea-truthing data the results showed a strong correlation (r = 0.94). When comparing the FR data to the sea-truthing data CDOM and K_d(490) showed a low accuracy, but a high precision with a rather constant off-set. In summary, one may state that the precision of MERIS data improves by applying the FUB Case-2 processor and the accuracy improves with improved spatial resolution for chlorophyll and SPM. Furthermore, the FUB processor can be used off-the-shelf for open Baltic Sea monitoring, provided one corrects for the respective off-set from sea-truthing data which is most likely caused by an inaccuracy in the atmospheric correction. Additionally, the FR data can be used to derive CDOM, K_d(490) and Secchi depth in Himmmerfjärden if one corrects for the respective off-set. We will need to perform more comparisons between sea-truthing and MERIS FR data before the new K_d(490) algorithm can be made operational, including also scenes from other times of year. In order to provide a level 2 product that can be used reliably by the Baltic Sea user community, our recommendation to ESA is to include the spectral attenuation coefficient as a MERIS standard product.