Hyperspectral Imager For The Coastal Ocean Research Articles

Spectral band-shifting of multispectral remote-sensing reflectance products: Insights for matchup and cross-mission consistency assessments

Comparing or integrating remote-sensing reflectance, Rrsλsr−1, data from multispectral sensors is hindered by differences in their nominal bands and relative spectral-response functions. Similarly, assessing satellite-to-in situ-matchup assessments of Rrs using autonomous ground radiometers, such as the Ocean Color component of the Aerosol Robotic Network (AERONET-OC), faces the same challenges. Therefore, spectral band-shifting techniques are required to approximate Rrs in spectral bands that are mutually unavailable. We present a spectral-matching technique (SMTH) using a novel look-up-table (LUT) generated from Rrs obtained from the Hyperspectral Imager for the Coastal Ocean (HICO) sensor. This LUT consists of nearly 6.3 million unique Rrs spectra, processed from 8893 Level-2 HICO scenes. An extensive in situ hyperspectral dataset (n = 8289) was used to compare the performance of SMTH with three commonly used techniques, namely Melin and Sclep (2015) “MS-QAA” model, linear (LNR) and cubic spline (CS) interpolations. The comprehensive evaluation of eight sensors using in situ dataset revealed that the SMTH has significantly better overall accuracy with respect to spectral band-shifting than the existing approaches. The Bias was improved 7-fold, whereas the root-mean-square difference (RMSD) and mean absolute difference (MAD) were ∼ 15–60% lower. Even so, each model exhibits outstanding performance in certain bands. For instance, the LNR model excels in the 380 nm band, while the MS-QAA emerges as the optimal choice for the 400 and 412 nm bands. SMTH outperforms other models in the green-red region, notably above 600 nm, where most existing ocean-color sensors have a limited number of spectral bands. Despite this variability, all models, showcase satisfactory performance in the 443–560 nm region. Significantly, both SMTH and LNR demonstrated a 100% retrieval rate, the ratio of band-shifted samples to the input samples. In contrast, MS-QAA and CS returned invalid values for 7% and 39% of the input samples, respectively. Further evaluation of regional, global, and composite satellite data suggests that the trophic state significantly influences the model performance. The MS-QAA model exhibited the smallest differences within the oligotrophic waters, characterized by low chlorophyll-a concentrations. In contrast, the SMTH model demonstrates the lowest differences in mesotrophic/eutrophic waters and regions subject to harmful algal blooms. While HICO primarily sampled coastal waters, estuaries, and shallow regions from mid- to low latitudes, there is an opportunity to enhance the representativeness of our LUT by incorporating open-ocean and high-latitude Rrs spectra from the upcoming hyperspectral missions, such as NASA's Plankton, Aerosol, Clouds, ocean Ecosystem (PACE).

A hyperspectral inversion framework for estimating absorbing inherent optical properties and biogeochemical parameters in inland and coastal waters

The simultaneous remote estimation of biogeochemical parameters (BPs) and inherent optical properties (IOPs) from hyperspectral satellite imagery of globally distributed optically distinct inland and coastal waters is a complex, unsolved, non-unique inverse problem. To tackle this problem, we leverage a machine-learning model termed Mixture Density Networks (MDNs). MDNs outperform operational algorithms by calculating the covariance between the simultaneously estimated products. We train the MDNs on a large (N = 8237) dataset of co-aligned, in situ measured, hyperspectral remote sensing reflectance (Rrs), BPs, and absorbing IOPs from globally representative optically distinct inland and coastal waters. The estimated IOPs include absorption due to phytoplankton (aph), chromophoric dissolved organic matter (acdom), and non-algal particles (anap). The estimated BPs include chlorophyll-a, total suspended solids, and phycocyanin (PC). MDNs dramatically reduce uncertainty in the retrievals, relative to operational algorithms, when using a 50/50 dataset split, where the MDNs are trained on a randomly selected half of the in situ dataset and validated on the other half. Our model is shown to have higher, or equivalent, generalization performance than the calculated operational algorithms available for all BPs and IOPs (except PC) via a leave-one-out cross-validation assessment. The MDNs are sensitive to uncertainties in the hyperspectral satellite Rrs, resulting from instrument noise and atmospheric correction; there is a difference of ∼37.4–62.8% (using median symmetric accuracy) between the MDNs' estimates derived from co-located satellite-derived Rrs and in situ Rrs. Of the IOPs, acdom and anap are less sensitive to uncertainties in hyperspectral satellite imagery relative to aph, with remote estimates of aph exhibiting incorrect spectral shape and magnitude relative to in situ measured IOPs. Despite the uncertainties in satellite derived Rrs, the spatial distributions of BPs and IOPs in MDN-derived product maps of Lake Erie and the Curonian Lagoon, based on imagery taken with the Hyperspectral Imager for the Coastal Ocean (HICO) and PRecursore IperSpettrale della Missione Applicativa (PRISMA), are confirmed via co-aligned in situ measurements and agree with the literature's understanding of these well-studied regions. The consistency and accuracy of the model on HICO and PRISMA imagery, despite radiometric uncertainties, demonstrate its applicability to future hyperspectral missions, such as the Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) mission, where the simultaneous estimation model will serve as a key part of phytoplankton community composition analysis.

Characterising retrieval uncertainty of chlorophyll-a algorithms in oligotrophic and mesotrophic lakes and reservoirs

Remote sensing product uncertainties for phytoplankton chlorophyll-a (chla) concentration in oligotrophic and mesotrophic lakes and reservoirs were characterised across 13 existing algorithms using an in situ dataset of water constituent concentrations, inherent optical properties (IOPs) and remote-sensing reflectance spectra Rrsλ collected from 53 lakes and reservoirs (346 observations; chla concentration < 10 mg m-3, dataset median 2.5 mg m-3). Substantial shortcomings in retrieval accuracy were evident with median absolute percentage differences (MAPD) > 37% and mean absolute differences (MAD) > 1.82 mg m-3. Using the Hyperspectral Imager for the Coastal Ocean (HICO) band configuration improved the accuracies by 10–20% compared to the Ocean and Land Colour Instrument (OLCI) configuration. Retrieval uncertainties were attributed to optical and biogeochemical properties using machine learning models through SHapley Additive exPlanations (SHAP). The chla retrieval uncertainty of most semi-analytical algorithms was primarily determined by phytoplankton absorption and composition. Machine learning chla algorithms showed relatively high sensitivity to light absorption by coloured dissolved organic matter (CDOM) and non-algal pigment particulates (NAP). In contrast, the uncertainties of red/near-infrared algorithms, which aim for lower uncertainty in the presence of CDOM and NAP, were primarily explained through the total absorption by phytoplankton at 673 nm (aϕ(673)) and variables related to backscatter. Based on these uncertainty characterisations we discuss the suitability of the evaluated algorithm formulations, and we make recommendations for chla estimation improvements in oligo- and mesotrophic lakes and reservoirs.

QWIP: A Quantitative Metric for Quality Control of Aquatic Reflectance Spectral Shape Using the Apparent Visible Wavelength

The colors of the ocean and inland waters span clear blue to turbid brown, and the corresponding spectral shapes of the water-leaving signal are diverse depending on the various types and concentrations of phytoplankton, sediment, detritus and colored dissolved organic matter. Here we present a simple metric developed from a global dataset spanning blue, green and brown water types to assess the quality of a measured or derived aquatic spectrum. The Quality Water Index Polynomial (QWIP) is founded on the Apparent Visible Wavelength (AVW), a one-dimensional geophysical metric of color that is inherently correlated to spectral shape calculated as a weighted harmonic mean across visible wavelengths. The QWIP represents a polynomial relationship between the hyperspectral AVW and a Normalized Difference Index (NDI) using red and green wavelengths. The QWIP score represents the difference between a spectrum’s AVW and NDI and the QWIP polynomial. The approach is tested extensively with both raw and quality controlled field data to identify spectra that fall outside the general trends observed in aquatic optics. For example, QWIP scores less than or greater than 0.2 would fail an initial screening and be subject to additional quality control. Common outliers tend to have spectral features related to: 1) incorrect removal of surface reflected skylight or 2) optically shallow water. The approach was applied to hyperspectral imagery from the Hyperspectral Imager for the Coastal Ocean (HICO), as well as to multispectral imagery from the Visual Infrared Imaging Radiometer Suite (VIIRS) using sensor-specific extrapolations to approximate AVW. This simple approach can be rapidly implemented in ocean color processing chains to provide a level of uncertainty about a measured or retrieved spectrum and flag questionable or unusual spectra for further analysis.

Hyperspectral reflectance spectra of floating matters derived from Hyperspectral Imager for the Coastal Ocean (HICO) observations

Abstract. Using data collected by the Hyperspectral Imager for the Coastal Ocean (HICO) on the International Space Station between 2010–2014, hyperspectral reflectance spectra of various floating matters in global oceans and lakes are derived for the spectral range of 400–800 nm. Specifically, the entire HICO archive of 9411 scenes is first visually inspected to identify suspicious image slicks. Then, a nearest-neighbor atmospheric correction is used to derive surface reflectance of slick pixels. Finally, a spectral unmixing scheme is used to derive the reflectance spectra of floating matters. Analysis of the spectral shapes of these various floating matters (macroalgae, microalgae, organic particles, whitecaps) through the use of a spectral angle mapper (SAM) index indicates that they can mostly be distinguished from each other without the need for ancillary information. Such reflectance spectra from the consistent 90 m resolution HICO observations are expected to provide spectral endmembers to differentiate and quantify the various floating matters from existing multi-band satellite sensors and future hyperspectral satellite missions such as NASA's Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) mission; Geosynchronous Littoral Imaging and Monitoring Radiometer (GLIMR) mission; and Surface Biology and Geology (SBG) mission. All spectral data are available at https://doi.org/10.21232/74LvC3Kr (Hu, 2021b).

Development of an App and Teaching Concept for Implementation of Hyperspectral Remote Sensing Data into School Lessons Using Augmented Reality

For the purpose of expanding STEM (science, technology, engineering, mathematics) education with remote sensing (RS) data and methods, an augmented reality (AR) app was developed in combination with a worksheet and lesson plan. Data from the Hyperspectral Imager for the Coastal Ocean (HICO) was searched for topics applicable to STEM curricula, which was found in the example of a harmful algal bloom in Lake Erie, USA, in 2011. Spectral shape algorithms were applied to differentiate between less harmful green and more harmful blue algae in the lake. The data was pre-processed to reduce its size significantly without losing too much information and then integrated into an app that was developed in Unity with the Vuforia extension. It was designed to let students browse and understand the raw data in RGB and a tangible hyperspectral cube, as well as to analyze algae maps derived from it. The app runs on Android smartphones with minimized data usage to make it less dependent on school funding and the socioeconomic background of students. Using educational concepts, such as active and collaborative learning, moderate constructivism, and scientific inquiry, the data was integrated into a lesson about environmental problems that was enhanced by the AR app. The app and worksheet were evaluated in two advanced geography courses (n = 36) and found to be complex, but doable and understandable, for the target group of German high school students in their final two school years. Thus, hyperspectral data can be used for STEM lessons using AR technology on students’ smartphones with several limitations both in the technology used and gainable knowledge.

Cyanobacterial pigment concentrations in inland waters: Novel semi-analytical algorithms for multi- and hyperspectral remote sensing data

Cyanobacteria are notorious for producing harmful algal blooms that present an ever-increasing serious threat to aquatic ecosystems worldwide, impacting the quality of drinking water and disrupting the recreational use of many water bodies. Remote sensing techniques for the detection and quantification of cyanobacterial blooms are required to monitor their initiation and spatiotemporal variability. In this study, we developed a novel semi-analytical approach to estimate the concentration of cyanobacteria-specific pigment phycocyanin (PC) and common phytoplankton pigment chlorophyll a (Chl a) from hyperspectral remote sensing data. The PC algorithm was derived from absorbance-concentration relationship, and the Chl a algorithm was devised based on a conceptual three-band structure model. The developed algorithms were applied to satellite imageries obtained by the Hyperspectral Imager for the Coastal Ocean (HICO™) sensor and tested in Lake Kinneret (Israel) during strong cyanobacterium Microcystis sp. bloom and out-of-bloom times. The sensitivity of the algorithms to errors was evaluated. The Chl a and PC concentrations were estimated with a mean absolute percentage difference (MAPD) of 16% and 28%, respectively. Sensitivity analysis shows that the influences of backscattering and other water constituents do not affect the estimation accuracy of PC (~2% MAPD). The reliable PC/Chl a ratios can be obtained at PC concentrations above 10 mg m−3. The computed PC/Chl a ratio depicts the contribution of cyanobacteria to the total phytoplankton biomass and permits investigating the role of ambient factors in the formation of a complex planktonic community. The novel algorithms have extensive practical applicability and should be suitable for the quantification of PC and Chl a in aquatic ecosystems using hyperspectral remote sensing data as well as data from future multispectral remote sensing satellites, if the respective bands are featured in the sensor.

Advancing cyanobacteria biomass estimation from hyperspectral observations: Demonstrations with HICO and PRISMA imagery

Retrieval of the phycocyanin concentration (PC), a characteristic pigment of, and proxy for, cyanobacteria biomass, from hyperspectral satellite remote sensing measurements is challenging due to uncertainties in the remote sensing reflectance (∆Rrs) resulting from atmospheric correction and instrument radiometric noise. Although several individual algorithms have been proven to capture local variations in cyanobacteria biomass in specific regions, their performance has not been assessed on hyperspectral images from satellite sensors. Our work leverages a machine-learning model, Mixture Density Networks (MDNs), trained on a large (N = 939) dataset of collocated in situ chlorophyll-a concentrations (Chla), PCs, and remote sensing reflectance (Rrs) measurements to estimate PC from all relevant spectral bands. The performance of the developed model is demonstrated via PC maps produced from select images of the Hyperspectral Imager for the Coastal Ocean (HICO) and Italian Space Agency's PRecursore IperSpettrale della Missione Applicativa (PRISMA) using a matchup dataset. As input to the MDN, we incorporate a combination of widely used band ratios (BRs) and line heights (LHs) taken from existing multispectral algorithms, that have been proven for both Chla and PC estimation, as well as novel BRs and LHs to increase the overall cyanobacteria biomass estimation accuracy and reduce the sensitivity to ∆Rrs. When trained on a random half of the dataset, the MDN achieves uncertainties of 44.3%, which is less than half of the uncertainties of all viable optimized multispectral PC algorithms. The MDN is notably better than multispectral algorithms at preventing overestimation on low (<10 mg m−3) PC. Visibly, HICO and PRISMA PC maps show the wider dynamic range that can be represented by the MDN. The available in situ and satellite-derived Rrs matchups and measured in situ PC demonstrate the robustness of the MDN for estimating low (<10 mg m−3) PC and the reduced impact of ∆Rrs on medium-to-high in situ PC (>10 mg m−3). According to our extensive assessments, the developed model is anticipated to enable practical PC products from PRISMA and HICO, therefore the model is promising for planned hyperspectral missions, such as the Plankton Aerosol and Cloud Ecosystem (PACE). This advancement will enhance the complementary roles of hyperspectral radiometry from satellite and low-altitude platforms for quantifying and monitoring cyanobacteria harmful algal blooms at both large and local spatial scales.

Use of Hyper-Spectral Visible and Near-Infrared Satellite Data for Timely Estimates of the Earth’s Surface Reflectance in Cloudy Conditions: Part 2- Image Restoration With HICO Satellite Data in Overcast Conditions

Satellite-based visible and near-infrared imaging of the Earth’s surface is generally not performed in moderate to highly cloudy conditions; images that look visibly cloud covered to the human eye are typically discarded. Here, we expand upon previous work that employed machine learning (ML) to estimate underlying land surface reflectances at red, green, and blue (RGB) wavelengths in cloud contaminated spectra using a low spatial resolution satellite spectrometer. Specifically, we apply the ML methodology to a case study at much higher spatial resolution with the Hyperspectral Imager for the Coastal Ocean (HICO) that flew on the International Space Station (ISS). HICO spatial sampling is of the order of 90 m. The purpose of our case study is to test whether high spatial resolution features can be captured using hyper-spectral imaging in lightly cloudy and overcast conditions. We selected one clear and one cloudy image over a portion of the panhandle coastline of Florida to demonstrate that land features are partially recoverable in overcast conditions. Many high contrast features are well recovered in the presence of optically thin clouds. However, some of the low contrast features, such as narrow roads, are smeared out in the heavily clouded part of the reconstructed image. This case study demonstrates that our approach may be useful for many science and operational applications that are being developed for current and upcoming satellite missions including precision agriculture and natural vegetation analysis, water quality assessment, as well as disturbance, change, hazard, and disaster detection.

Assessment of Polymer Atmospheric Correction Algorithm for Hyperspectral Remote Sensing Imagery over Coastal Waters.

Spaceborne imaging spectroscopy, also called hyperspectral remote sensing, has shown huge potential to improve current water colour retrievals and, thereby, the monitoring of inland and coastal water ecosystems. However, the quality of water colour retrievals strongly depends on successful removal of the atmospheric/surface contributions to the radiance measured by satellite sensors. Atmospheric correction (AC) algorithms are specially designed to handle these effects, but are challenged by the hundreds of narrow spectral bands obtained by hyperspectral sensors. In this paper, we investigate the performance of Polymer AC for hyperspectral remote sensing over coastal waters. Polymer is, in nature, a hyperspectral algorithm that has been mostly applied to multispectral satellite data to date. Polymer was applied to data from the Hyperspectral Imager for the Coastal Ocean (HICO), validated against in situ multispectral (AERONET-OC) and hyperspectral radiometric measurements, and its performance was compared against that of the hyperspectral version of NASA’s standard AC algorithm, L2gen. The match-up analysis demonstrated very good performance of Polymer in the green spectral region. The mean absolute percentage difference across all the visible bands varied between 16% (green spectral region) and 66% (red spectral region). Compared with L2gen, Polymer remote sensing reflectances presented lower uncertainties, greater data coverage, and higher spectral similarity to in situ measurements. These results demonstrate the potential of Polymer to perform AC on hyperspectral satellite data over coastal waters, thus supporting its application in current and future hyperspectral satellite missions.

Hyperspectral retrievals of phytoplankton absorption and chlorophyll-a in inland and nearshore coastal waters

Following more than two decades of research and developments made possible through various proof-of-concept hyperspectral remote sensing missions, it has been anticipated that hyperspectral imaging would enhance the accuracy of remotely sensed in-water products. This study investigates such expected improvements and demonstrates the utility of hyperspectral radiometric measurements for the retrieval of near-surface phytoplankton properties11Codes are accessible via https://github.com/STREAM-RS/STREAM-RS, i.e., phytoplankton absorption spectra (aph) and biomass evaluated through examining the concentration of chlorophyll-a (Chla). Using hyperspectral data (409–800 nm at ~5 nm resolution) and a class of neural networks known as Mixture Density Networks (MDN) (Pahlevan et al., 2020), we show that the median error in aph retrievals is reduced two-to-three times (N = 722) compared to that from heritage ocean color algorithms. The median error associated with our aph retrieval across all the visible bands varies between 20 and 30%. Similarly, Chla retrievals exhibit significant improvements (i.e., more than two times; N = 1902), with respect to existing algorithms that rely on select spectral bands. Using an independent matchup dataset acquired near-concurrently with the acquisition of the Hyperspectral Imager for the Coastal Ocean (HICO) images, the models are found to perform well, but at reduced levels due to uncertainties in the atmospheric correction. The mapped spatial distribution of Chla maps and aph spectra for selected HICO swaths further solidify MDNs as promising machine-learning models that have the potential to generate highly accurate aquatic remote sensing products in inland and coastal waters. For aph retrieval to improve further, two immediate research avenues are recommended: a) the network architecture requires additional optimization to enable a simultaneous retrieval of multiple in-water parameters (e.g., aph, Chla, absorption by colored dissolved organic matter), and b) the training dataset should be extended to enhance model generalizability. This feasibility analysis using MDNs provides strong evidence that high-quality, global hyperspectral data will open new pathways toward a better understanding of biodiversity in aquatic ecosystems.

Benefit of the Potential Future Hyperspectral Satellite Sensor (BIODIVERSITY) for Improving the Determination of Water Column and Seabed Features in Coastal Zones

Most of the studies dealing with seabed mapping from hyperspectral images have been carried out using airborne data although hyperspectral satellite sensors have already been or are planned to be launched for the near future (hyperspectral imager for the coastal ocean (HICO) environmental mapping and analysis program (ENMAP) or BIODIVERSITY). The objective of this study is to evaluate the benefit of a BIODIVERSITY-like sensor to determine the bio-optical properties of the water column, namely the Chlorophyll-a concentration, the suspended particulate matter (SPM) concentration, the absorption coefficient of the colored dissolved organic matter (CDOM), the bathymetry and the composition of the seabed, according to its spatial resolution and spectral resolution and its signal to noise ratio. For this purpose, radiative transfer simulations are analyzed together with remote sensing hyperspectral airborne data acquired above the Porquerolles Island (France). The retrieval performance of all in-water and seabed parameters derived from the inversion of BIODIVERSITY-like data is compared with the performance obtained using ENMAP and HICO spatial and radiometric specifications. It is shown that a BIODIVERSITY-like sensor significantly improves the estimation performance of the water column parameters. Furthermore, BIODIVERSITY-like sensor is highly appropriate for seabed mapping when bottom pixels are composed of pure material (e.g., Sand or Posidonia ) in shallow waters when seabed depth is less than 10 m. Conversely, the performance of the inversion deteriorates when seabed pixels are composed of mixed materials (e.g., Sand mixed with Posidonia ). It is also shown that the concentration of chlorophyll, SPM and CDOM absorption are less sensitive to noise level than depth and seabed abundance.

An ultraviolet to visible scheme to estimate chromophoric dissolved organic matter absorption in a Case-2 water from remote sensing reflectance

In a typical Case-2 coastal water environment (here, the Pearl River Estuary (PRE), China), chromophoric dissolved organic matter (CDOM) and suspended particulates dominate the water optical properties, and CDOM fluorescence contributes considerably to surface water reflectance. In this paper, an ultraviolet (UV) to visible scheme to retrieve CDOM absorption (ag)is developed based on a set of in situ observations. First, the CDOM UV absorption and spectral slope (Sg) are derived directly from the visible remote sensing reflectance; then the Sg is extrapolated to obtain the spectrum from UV to visible spectral range. This algorithm performs well, with an overall mean absolute percent difference (MAPD) of ∼30%, ∼5% and ∼6% for the estimation of ag in 250–450 nm, Sg over 250–400 nm, and 250–700 nm, respectively. The effectiveness and stability of the algorithm is further demonstrated in capturing the distribution pattern of CDOM absorption in the PRE from satellite ocean color imagery with multiple spatial and spectral resolution, namely: the Visible Infrared Imaging Radiometer Suite (VIIRS) (750 m/Multispectral), the Ocean and Land Color Instrument (OLCI) (300 m/Multispectral), the Hyperspectral Imager for the Coastal Ocean (HICO) (100 m/Hyperspectral), and the Landsat 8 Operational Land Imager (OLI) (30 m/Multispectral). The UV to visible scheme can benefit the CDOM absorption estimation in two aspects: 1) it is free from the disturbance of suspended matter; 2) it avoids uncertainties caused by the low signal-to-noise ratio (SNR) of ag measurements in the visible range. The algorithm is effective in revealing multiple scales of variation of CDOM absorption from ocean color observations.

Spectral index-based dynamic threshold technique for detecting cloud contamination in ocean colour data

ABSTRACT Coastal and lagoon water colour products are often greatly contaminated by clouds and accompanying shadows. Adjacency effects due to the strong reflection from contiguous cloud pixels can introduce further uncertainty into the water-leaving radiance retrieval process. This study aims to propose a robust cloud-index dynamic-threshold (CIDT) algorithm based on the spectral and spatial characteristics of the clouds and evaluate its performance in comparison with results obtained from the existing algorithms using multispectral and hyperspectral images from the lagoon and coastal water zones. Detection of cloud contamination from these images is performed before applying atmospheric correction in order to prevent significant loss of valuable data and ensure the quality of higher-level products. The CIDT starts with data pre-processing (in order to avoid misclassification across the clouds and non-cloud regions) by excluding signatures other than the cloud and non-cloud (water and land) pixels. Based on the spectral and spatial variability characteristic of clouds, a new cloud index is then introduced that uses the top-of-atmosphere reflectance in conjugation with the dynamic threshold values to detect the cloud pixels. For quantitative and qualitative analyses, CIDT is tested on a number of scenes provided by Landsat 8 Operational Land Imager (OLI) and Hyperspectral Imager for the Coastal Ocean (HICO) sensors over the coastal and lagoon zones. According to the comparison of the algorithm results with the reference cloud flags, CIDT shows a significant improvement in identifying clouds of varying opacity, achieves a higher cloud detection accuracy, and reduces the number of the previously misclassified pixels by other algorithms in different coastal and lagoon water regions. The CIDT is extensively tested in mid-latitude and tropical areas and CIDT can be applied to both multispectral and hyperspectral images over the lagoon and coastal zones without the need for ancillary information.

Hyperspectral Remote Sensing of Phytoplankton Species Composition Based on Transfer Learning

Phytoplankton species composition research is key to understanding phytoplankton ecological and biogeochemical functions. Hyperspectral optical sensor technology allows us to obtain detailed information about phytoplankton species composition. In the present study, a transfer learning method to inverse phytoplankton species composition using in situ hyperspectral remote sensing reflectance and hyperspectral satellite imagery was presented. By transferring the general knowledge learned from the first few layers of a deep neural network (DNN) trained by a general simulation dataset, and updating the last few layers with an in situ dataset, the requirement for large numbers of in situ samples for training the DNN to predict phytoplankton species composition in natural waters was lowered. This method was established from in situ datasets and validated with datasets collected in different ocean regions in China with considerable accuracy (R2 = 0.88, mean absolute percentage error (MAPE) = 26.08%). Application of the method to Hyperspectral Imager for the Coastal Ocean (HICO) imagery showed that spatial distributions of dominant phytoplankton species and associated compositions could be derived. These results indicated the feasibility of species composition inversion from hyperspectral remote sensing, highlighting the advantages of transfer learning algorithms, which can bring broader application prospects for phytoplankton species composition and phytoplankton functional type research.

Evaluating ultraviolet (UV) based photochemistry in optically complex coastal waters using the Hyperspectral Imager for the Coastal Ocean (HICO)

Knowledge of light partitioning into different optically active constituents, particularly chromophoric dissolved organic matter (CDOM) in the ultraviolet (UV) is indispensable for understanding UV dependent biogeochemical issues including photochemical processes in optically complex waters. Herein a new approach is presented to investigate photochemistry by blending two ocean color algorithms, namely the composite SeaUV (Cao et al., 2014) and the SeaCDOM (Cao and Miller, 2015) algorithms, and applying them to visible remote sensing reflectance (Rrs) measured using the Hyperspectral Imager for the Coastal Ocean (HICO). As illustrated using photochemical carbon monoxide (CO) production from CDOM, this model approach allows high resolution examination of UV optical details with estimates of both depth-specific and depth-integrated photoproduction rates in a dynamic estuarine/coastal environment. Decoupled retrievals of inherent and apparent optical properties (i.e. diffuse attenuation coefficient (Kd) and CDOM absorption coefficient (ag)) using two distinct ocean color algorithms over the entire UV spectrum allow a synoptically dynamic view of CDOM's contribution to light attenuation (ag/Kd). This provides new potential to probe UV processes in complex coastal waters on regional as well as global scales using remote sensing of ocean color.

A Reconstruction Method for Hyperspectral Remote Sensing Reflectance in the Visible Domain and Applications

AbstractA reconstruction method was developed for hyperspectral remote sensing reflectance (Rrs) data in the visible domain (400–700 nm) based on in situ observations. A total of 2,647 Rrs spectra were collected over a wide variety of water environments including open ocean, coastal and inland waters. Ten schemes with different band numbers (6 to 15) were tested based on a nonlinear model. It was found that the accuracy of the reconstruction increased with the increase of input band numbers. Eight of these schemes met the accuracy criterion with the mean absolute error (MAE) and mean relative error (MRE) values between reconstructed and in situ Rrs less than 0.00025 sr−1 and 5%, respectively. We chose the eight‐band scheme for further evaluation because of its decent performance. The results revealed that the parameterization derived by the eight‐band scheme was efficient for restoring Rrs spectra from different water bodies. In contrast to the previous studies that used a linear model with 15 spectral bands, the nonlinear model with the eight‐band scheme yielded a comparable reconstruction performance. The MAE and MRE values were generally less than 0.00016 sr−1 and 3% respectively; much lower than the uncertainties in satellite‐derived Rrs products. Furthermore, a preliminary experiment of this method on the data from the Hyperspectral Imager for the Coastal Ocean (HICO) showed high potential in the future applications for reconstructing Rrs spectra from space‐borne optical sensors. Overall, the eight‐band scheme with our nonlinear model was proven to be optimal for hyperspectral Rrs reconstruction in the visible domain.

Estimation of suspended particulate matter in turbid coastal waters: application to hyperspectral satellite imagery.

An empirical algorithm is proposed to estimate suspended particulate matter (SPM) ranging from 0.675 to 25.7 mg L-1 in the turbid Pearl River estuary (PRE). Comparisons between model predicted and in situ measured SPM resulted in R2s of 0.97 and 0.88 and mean absolute percentage errors (MAPEs) of 23.96% and 29.69% by using the calibration and validation data sets, respectively. The developed algorithm demonstrated the highest accuracy when compared with existing ones for turbid coastal waters. The diurnal dynamics of SPM was revealed by applying the proposed algorithm to reflectance data collected by a moored buoy in the PRE. The established algorithm was implemented to Hyperspectral Imager for the Coastal Ocean (HICO) data and the distribution pattern of SPM in the PRE was elucidated. Validation of HICO-derived reflectance data by using concurrent MODIS/Aqua data as a benchmark indicated their reliability. Factors influencing variability of SPM in the PRE were analyzed, which implicated the combined effects of wind, tide, rainfall, and circulation as the cause.

Improving the analysis of biogeochemical patterns associated with internal waves in the strait of Gibraltar using remote sensing images

High Amplitude Internal Waves (HAIWs) are physical processes observed in the Strait of Gibraltar (the narrow channel between the Atlantic Ocean and the Mediterranean Sea). These internal waves are generated over the Camarinal Sill (western side of the strait) during the tidal outflow (toward the Atlantic Ocean) when critical hydraulic conditions are established. HAIWs remain over the sill for up to 4 h until the outflow slackens, being then released (mostly) towards the Mediterranean Sea. These have been previously observed using Synthetic Aperture Radar (SAR), which captures variations in surface water roughness. However, in this work we use high resolution optical remote sensing, with the aim of examining the influence of HAIWs on biogeochemical processes. We used hyperspectral images from the Hyperspectral Imager for the Coastal Ocean (HICO) and high spatial resolution (10 m) images from the MultiSpectral Instrument (MSI) onboard the Sentinel-2A satellite. This work represents the first attempt to examine the relation between internal wave generation and the water constituents of the Camarinal Sill using hyperspectral and high spatial resolution remote sensing images. This enhanced spatial and spectral resolution revealed the detailed biogeochemical patterns associated with the internal waves and suggests local enhancements of productivity associated with internal waves trains.

Atmospheric correction for hyperspectral ocean color retrieval with application to the Hyperspectral Imager for the Coastal Ocean (HICO)

The classical multi-spectral Atmospheric Correction (AC) algorithm is inadequate for the new generation of spaceborne hyperspectral sensors such as NASA's first hyperspectral Ocean Color Instrument (OCI) onboard the anticipated Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) satellite mission. The AC process must estimate and remove the atmospheric path radiance contribution due to the Rayleigh scattering by air molecules and scattering by aerosols from the measured top-of-atmosphere (TOA) radiance, compensate for the absorption by atmospheric gases, and correct for reflection and refraction of the air-sea interface. In this work, we present and evaluate an improved AC for hyperspectral sensors developed within NASA's SeaWiFS Data Analysis System software package (SeaDAS). The improvement is based on combining the classical AC approach of multi-spectral capabilities to correct for the atmospheric path radiance, extended to hyperspectral, with a gas correction algorithm to compensate for absorbing gases in the atmosphere, including water vapor. The SeaDAS-hyperspectral version is capable of operationally processing the AC of any hyperspectral airborne or spaceborne sensor. The new algorithm development was evaluated and assessed using the Hyperspectral Imager for Coastal Ocean (HICO) scenes collected at the Marine Optical BuoY (MOBY) site, and other SeaWiFS Bio-optical Archive and Storage System (SeaBASS) and AERosol Robotic NETwork - Ocean Color (AERONET-OC) coastal sites. A hyperspectral vicarious calibration was applied to HICO, showing the validity and consistency of HICO's ocean color products. The hyperspectral AC capability is currently available in SeaDAS to the scientific community at https://oceancolor.gsfc.nasa.gov/.