Performance across WorldView-2 and RapidEye for reproducible seagrass mapping

Impact of Atmospheric Correction on Classification and Quantification of Seagrass Density from WorldView-2 Imagery

Mapping the seagrass distribution and density in the underwater landscape can improve global Blue Carbon estimates. However, atmospheric absorption and scattering introduce errors in space-based sensors' retrieval of sea surface reflectance, affecting seagrass presence, density, and above-ground carbon (AGC seagrass ) estimates. This study assessed atmospheric correction's impact on mapping seagrass using WorldView-2 satellite imagery from Saint Joseph Bay, Saint George Sound, and Keaton Beach in Florida, USA. Coincident in situ measurements of water-leaving radiance (L W ), optical properties, and seagrass leaf area index (LAI) were collected. Seagrass classification and the retrieval of LAI were compared after empirical line height (ELH) and dark-object subtraction (DOS) methods were used for atmospheric correction. DOS left residual brightness in the blue and green bands but had minimal impact on the seagrass classification accuracy. However, the brighter reflectance values reduced LAI retrievals by up to 50% compared to ELH-corrected images and ground-based observations. This study offers a potential correction for LAI underestimation due to incomplete atmospheric correction, enhancing the retrieval of seagrass density and above-ground Blue Carbon from WorldView-2 imagery without in situ observations for accurate atmospheric interference correction.

Providing a framework for seagrass mapping in United States coastal ecosystems using high spatial resolution satellite imagery

Seagrasses have been widely recognized for their ecosystem services, but traditional seagrass monitoring approaches emphasizing ground and aerial observations are costly, time-consuming, and lack standardization across datasets. This study leveraged satellite imagery from Maxar's WorldView-2 and WorldView-3 high spatial resolution, commercial satellite platforms to provide a consistent classification approach for monitoring seagrass at eleven study areas across the continental United States, representing geographically, ecologically, and climatically diverse regions. A single satellite image was selected at each of the eleven study areas to correspond temporally to reference data representing seagrass coverage and was classified into four general classes: land, seagrass, no seagrass, and no data. Satellite-derived seagrass coverage was then compared to reference data using either balanced agreement, the Mann-Whitney U test, or the Kruskal-Wallis test, depending on the format of the reference data used for comparison. Balanced agreement ranged from 58% to 86%, with better agreement between reference- and satellite-indicated seagrass absence (specificity ranged from 88% to 100%) than between reference- and satellite-indicated seagrass presence (sensitivity ranged from 17% to 73%). Results of the Mann-Whitney U and Kruskal-Wallis tests demonstrated that satellite-indicated seagrass percentage cover had moderate to large correlations with reference-indicated seagrass percentage cover, indicative of moderate to strong agreement between datasets. Satellite classification performed best in areas of dense, continuous seagrass compared to areas of sparse, discontinuous seagrass and provided a suitable spatial representation of seagrass distribution within each study area. This study demonstrates that the same methods can be applied across scenes spanning varying seagrass bioregions, atmospheric conditions, and optical water types, which is a significant step toward developing a consistent, operational approach for mapping seagrass coverage at the national and global scales. Accompanying this manuscript are instructional videos describing the processing workflow, including data acquisition, data processing, and satellite image classification. These instructional videos may serve as a management tool to complement field- and aerial-based mapping efforts for monitoring seagrass ecosystems.

Temporal Stability of Seagrass Extent, Leaf Area, and Carbon Storage in St. Joseph Bay, Florida: a Semi-automated Remote Sensing Analysis

Seagrasses are globally recognized for their contribution to blue carbon sequestration. However, accurate quantification of their carbon storage capacity remains uncertain due, in part, to an incomplete inventory of global seagrass extent and assessment of its temporal variability. Furthermore, seagrasses are undergoing significant decline globally, which highlights the urgent need to develop change detection techniques applicable to both the scale of loss and the spatial complexity of coastal environments. This study applied a deep learning algorithmto a 30-year time series of Landsat 5 through 8 imagery to quantify seagrass extent, leaf area index (LAI), and belowground organic carbon (BGC) in St. Joseph Bay, Florida, between 1990 and 2020. Consistent with previous field-based observations regarding stability of seagrass extent throughout St. Joseph Bay, there was no temporal trend in seagrass extent (23 ± 3 km2, τ = 0.09, p = 0.59, n = 31), LAI (1.6 ± 0.2, τ = -0.13, p = 0.42, n = 31), or BGC (165 ± 19 g C m-2, τ = - 0.01, p = 0.1, n = 31) over the 30-year study period. There were, however, six brief declines in seagrass extent between the years 2004 and 2019 following tropical cyclones, from which seagrasses recovered rapidly. Fine-scale interannual variability in seagrass extent, LAI, and BGC was unrelated to sea surface temperature or to climate variability associated with the El Niño-Southern Oscillation or the North Atlantic Oscillation. Although our temporal assessment showed that seagrass and its belowground carbon were stable in St. Joseph Bay from 1990 to 2020, forecasts suggest that environmental and climate pressures are ongoing, which highlights the importance of the method and time series presented here as a valuable tool to quantify decadal-scale variability in seagrass dynamics. Perhaps more importantly, our results can serve as a baseline against which we can monitor future change in seagrass communities and their blue carbon.

Species level mapping of a seagrass bed using an unmanned aerial vehicle and deep learning technique

Background

Seagrass beds are essential habitats in coastal ecosystems, providing valuable ecosystem services, but are threatened by various climate change and human activities. Seagrass monitoring by remote sensing have been conducted over past decades using satellite and aerial images, which have low resolution to analyze changes in the composition of different seagrass species in the meadows. Recently, unmanned aerial vehicles (UAVs) have allowed us to obtain much higher resolution images, which is promising in observing fine-scale changes in seagrass species composition. Furthermore, image processing techniques based on deep learning can be applied to the discrimination of seagrass species that were difficult based only on color variation. In this study, we conducted mapping of a multispecific seagrass bed in Saroma-ko Lagoon, Hokkaido, Japan, and compared the accuracy of the three discrimination methods of seagrass bed areas and species composition, i.e., pixel-based classification, object-based classification, and the application of deep neural network.

Methods

We set five benthic classes, two seagrass species (Zostera marina and Z. japonica), brown and green macroalgae, and no vegetation for creating a benthic cover map. High-resolution images by UAV photography enabled us to produce a map at fine scales (<1 cm resolution).

Results

The application of a deep neural network successfully classified the two seagrass species. The accuracy of seagrass bed classification was the highest (82%) when the deep neural network was applied.

Conclusion

Our results highlighted that a combination of UAV mapping and deep learning could help monitor the spatial extent of seagrass beds and classify their species composition at very fine scales.

Simulated response of St. Joseph Bay, Florida, seagrass meadows and their belowground carbon to anthropogenic and climate impacts

Seagrass meadows are degraded globally and continue to decline in areal extent due to human pressures and climate change. This study used the bio-optical model GrassLight to explore the impact of climate change and anthropogenic stressors on seagrass extent, leaf area index (LAI) and belowground organic carbon (BGC) in St. Joseph Bay, Florida, using water quality data and remotely-sensed sea surface temperature (SST) from 2002 to 2020. Model predictions were compared with satellite-derived measurements of seagrass extent and shoot density from the Landsat images for the same period. The GrassLight-derived area of potential seagrass habitat ranged from 36.2 km² to 39.2 km², averaging 38.0 ± 0.8 km² compared to an observed seagrass extent of 23.0 ± 3.0 km² derived from Landsat (range = 17.9-27.4 km²). GrassLight predicted a mean seagrass LAI of 2.7 m² leaf m^-2 seabed, compared to a mean LAI of 1.9 m² m^-2 estimated from Landsat, indicating that seagrass density in St. Joseph Bay may have been below its light-limited ecological potential. Climate and anthropogenic change simulations using GrassLight predicted the impact of changes in temperature, pH, chlorophyll a, chromophoric dissolved organic matter and turbidity on seagrass meadows. Simulations predicted a 2-8% decline in seagrass extent with rising temperatures that was offset by a 3-11% expansion in seagrass extent in response to ocean acidification when compared to present conditions. Simulations of water quality impacts showed that a doubling of turbidity would reduce seagrass extent by 18% and total leaf area by 21%. Combining climate and water quality scenarios showed that ocean acidification may increase seagrass productivity to offset the negative effects of both thermal stress and declining water quality on the seagrasses growing in St. Joseph Bay. This research highlights the importance of considering multiple limiting factors in understanding the effects of environmental change on seagrass ecosystems.

A New Orbiting Deployable System for Small Satellite Observations for Ecology and Earth Observation

In this paper, we present several study cases focused on marine, oceanographic, and atmospheric environments, which would greatly benefit from the use of a deployable system for small satellite observations. As opposed to the large standard ones, small satellites have become an effective and affordable alternative access to space, owing to their lower costs, innovative design and technology, and higher revisiting times, when launched in a constellation configuration. One of the biggest challenges is created by the small satellite instrumentation working in the visible (VIS), infrared (IR), and microwave (MW) spectral ranges, for which the resolution of the acquired data depends on the physical dimension of the telescope and the antenna collecting the signal. In this respect, a deployable payload, fitting the limited size and mass imposed by the small satellite architecture, once unfolded in space, can reach performances similar to those of larger satellites. In this study, we show how ecology and Earth Observations can benefit from data acquired by small satellites, and how they can be further improved thanks to deployable payloads. We focus on DORA—Deployable Optics for Remote sensing Applications—in the VIS to TIR spectral range, and on a planned application in the MW spectral range, and we carry out a radiometric analysis to verify its performances for Earth Observation studies.

Comparing Sentinel-2 and WorldView-3 Imagery for Coastal Bottom Habitat Mapping in Atlantic Canada

Satellite remote sensing is a valuable tool to map and monitor the distribution of marine macrophytes such as seagrass and seaweeds that perform many ecological functions and services in coastal habitats. Various satellites have been used to map the distribution of these coastal bottom habitat-forming species, with each sensor providing unique benefits. In this study, we first explored optimal methods to create bottom habitat maps using WorldView-3 satellite imagery. We secondly compared the WorldView-3 bottom habitat maps to previously produced Sentinel-2 maps in a temperate, optically complex environment in Nova Scotia, Canada to identify the top performing classification and the advantages and disadvantages of each sensor. Sentinel-2 provides a global, freely accessible dataset where four bands are available at a 10-m spatial resolution in the visible and near infrared spectrum. Conversely, WorldView-3 is a commercial satellite where eight bands are available at a 2-m spatial resolution in the visible and near infrared spectrum, but data catalogs are costly and limited in scope. Our optimal WorldView-3 workflow processed images from digital numbers to habitat classification maps, and included a semiautomatic stripe correction. Our comparison of bottom habitat maps explored the impact of improved WorldView-3 spatial resolution in isolation, and the combined advantage of both WorldView’s increased spatial and spectral resolution relative to Sentinel-2. We further explored the effect of tidal height on classification success, and relative changes in water clarity between images collected at different dates. As expected, both sensors are suitable for bottom habitat mapping. The value of WorldView-3 came from both its increased spatial and spectral resolution, particularly for fragmented vegetation, and the value of Sentinel-2 imagery comes from its global dataset that readily allows for large scale habitat mapping. Given the variation in scale, cost and resolution of the two sensors, we provide recommendations on their use for mapping and monitoring marine macrophyte habitat in Atlantic Canada, with potential applications to other coastal areas of the world.

Vertical artifacts in high-resolution WorldView-2 and WorldView-3 satellite imagery of aquatic systems

Satellite image artefacts are features that appear in an image but not in the original imaged object and can negatively impact the interpretation of satellite data. Vertical artefacts are linear features oriented in the along-track direction of an image system and can present as either banding or striping; banding are features with a consistent width, and striping are features with inconsistent widths. This study used high-resolution data from DigitalGlobe's (now Maxar) WorldView-3 satellite collected at Lake Okeechobee, Florida (FL), on 30 August 2017. This study investigated the impact of vertical artefacts on both at-sensor radiance and a spectral index for an aquatic target as WorldView-3 was primarily designed as a land sensor. At-sensor radiance measured by six of WorldView-3's eight spectral bands exhibited banding, more specifically referred to as non-uniformity, at a width corresponding to the multispectral detector sub-arrays that comprise the WorldView-3 focal plane. At-sensor radiance measured by the remaining two spectral bands, red and near-infrared (NIR) #1, exhibited striping. Striping in these spectral bands can be attributed to their time delay integration (TDI) settings at the time of image acquisition, which were optimized for land. The impact of vertical striping on a spectral index leveraging the red, red edge, and NIR spectral bands-referred to here as the NIR maximum chlorophyll index (MCI_NIR)-was investigated. Temporally similar imagery from the European Space Agency's Sentinel-3 and Sentinel-2 satellites were used as baseline references of expected chlorophyll values across Lake Okeechobee as neither Sentinel-3 nor Sentinel-2 imagery showed striping. Striping was highly prominent in the MCI_NIR product generated using WorldView-3 imagery, as noise in the at-sensor radiance exceeded any signal of chlorophyll in the image. Adjusting the image acquisition parameters for future tasking of WorldView-3 or the functionally similar WorldView-2 satellite may alleviate these artefacts. To test this, an additional WorldView-3 image was acquired at Lake Okeechobee, FL, on 26 May 2021 in which the TDI settings and scan line rate were adjusted to improve the signal-to-noise ratio. While some evidence of non-uniformity remained, striping was no longer noticeable in the MCI_NIR product. Future image tasking over aquatic targets should employ these updated image acquisition parameters. Since the red and NIR #1 spectral bands are critical for inland and coastal water applications, archived images not collected using these updated settings may be limited in their potential for analysis of aquatic variables that require these two spectral bands to derive.

A Simple Cloud-Native Spectral Transformation Method to Disentangle Optically Shallow and Deep Waters in Sentinel-2 Images

This study presents a novel method to identify optically deep water using purely spectral approaches. Optically deep waters, where the seabed is too deep for a bottom reflectance signal to be returned, is uninformative for seabed mapping. Furthermore, owing to the attenuation of light in the water column, submerged vegetation at deeper depths is easily confused with optically deep waters, thereby inducing misclassifications that reduce the accuracy of these seabed maps. While bathymetry data could mask out deeper areas, they are not always available or of sufficient spatial resolution for use. Without bathymetry data and based on the coastal aerosol blue green (1-2-3) bands of the Sentinel-2 imagery, this study investigates the use of band ratios and a false colour HSV transformation of both L1C and L2A images to separate optically deep and shallow waters across varying water quality over four tropical and temperate submerged sites: Tanzania, the Bahamas, the Caspian Sea (Kazakhstan) and the Wadden Sea (Denmark and Germany). Two supervised thresholds based on annotated reference data and an unsupervised Otsu threshold were applied. The band ratio group usually featured the best overall accuracies (OA), F1 scores and Matthews correlation coefficients, although the individual band combination might not perform consistently across different sites. Meanwhile, the saturation and hue band yielded close to best performance for the L1C and L2A images, featuring OA of up to 0.93 and 0.98, respectively, and a more consistent behaviour than the individual band ratios. Nonetheless, all these spectral methods are still susceptible to sunglint, the Sentinel-2 parallax effect, turbidity and water colour. Both supervised approaches performed similarly and were superior to the unsupervised Otsu’s method—the supervised methods featuring OA were usually over 0.70, while the unsupervised OA were usually under 0.80. In the absence of bathymetry data, this method could effectively remove optically deep water pixels in Sentinel-2 imagery and reduce the issue of dark pixel misclassification, thereby improving the benthic mapping of optically shallow waters and their seascapes.