Zimmerman, Richard: Old Dominion University (Project Lead)
Li, Jiang: Old Dominion University (Co-Investigator)
Schaeffer, Blake: EPA (Co-Investigator)
Project Funding:
2017 - 2020
NRA: 2016 NASA: Carbon Cycle Science
Funded by NASA
Abstract:
Vegetated coastal ecosystems cover only 0.2% of the ocean, yet they are disproportionally important in sequestering organic carbon relative to other ecosystems, and may account for 50% of all the carbon buried in marine sediments. Although less abundant than saltmarshes or mangrove ecosystems, seagrass meadows represent important sites of organic carbon accumulation that may account for 15% of that buried carbon. Besides burying carbon in anoxic sediments, seagrass meadows promote carbon storage in ocean waters by raising seawater alkalinity through carbonate dissolution and pyrite precipitation that is not captured in most Blue Carbon estimates. There is a large uncertainty in global Blue Carbon burial due to poorly constrained estimates of global seagrass coverage, regional disparities in situ data availability, and significant differences in the nature of seagrass ecosystem function derived from functional and morphological differences among species.
The proposed study will develop software tools for extracting seagrass distribution and biomass from appropriately resolved multispectral imagery that ultimately can be scaled up to provide global coverage when combined with information on submarine bathymetry and water quality. We will provide test case verification of our approach by quantifying abundances and mapping distributions of seagrasses in the Chesapeake Bay and Eastern Gulf of Mexico (EGOM), two regions of the coastal United States encompassing important seagrass resources that are currently threatened by changes in water quality, climate warming and sea level rise, and with which we have extensive local experience. We will exploit archived remote sensing imagery available from the USGS LandSat-8 and the Digital Globe (DG) WorldView-2, WorldView-3 satellites, and determine temporal changes in seagrass resources as permitted from the archived data. Remotely sensed estimates of above-ground biomass will be determined using algorithms developed by the PIs that have previously been successful in turbid coastal regions. A new genetic algorithm will be explored that can improve classification and threshold determination. The above ground biomass retrievals will be used to estimate potential Blue Carbon deposits consisting of below-ground seagrass biomass, buried leaf litter and allochthonous detrital carbon (terrestrial and marine) for each site. Remotely sensed seagrass distributions will be linked to a predictive bio-optical model we call GrassLight that will be used to explore the differential effects of water quality, ambient temperature and CO2 availability on seagrass abundance and distribution, and ultimate impacts on Blue Carbon reserves.
Advances in understanding the optics of shallow water environments, submerged vegetation canopies and seagrass physiology, combined with improved spatial resolution of radiometrically calibrated remote sensing platforms, now enable seagrass ecosystems to be monitored from orbiting platforms such as LandSat-8, WorldView-2, and WorldView-3 in addition to hyperspectral systems. Extensive meadows covering very large regions (e.g. Bahamas Banks) can even be quantified with MODIS. In anticipation of future NASA missions offering increased spatial and spectral resolution over current orbiting systems, we will develop procedures to quantify seagrass assets and monitor changes from existing archived data that can provide a bridge to new orbiting platforms as they become available. The algorithms and models developed here will quantify status for a key coastal marine habitat in the US and provide a pathway to utilize future higher resolution datasets providing increased spatial and spectral resolution to observe coastal environments (HyspIRI, and future high-resolution sensors) across the globe.
Publications:
Islam K.A., Pérez D., Hill V., Schaeffer B., Zimmerman R., Li J. (2018) Seagrass Detection in Coastal Water Through Deep Capsule Networks. In: Lai JH. et al. (eds) Pattern Recognition and Computer Vision. PRCV 2018. Lecture Notes in Computer Science, vol 11257. Springer, Cham.
Islam, K. A., Hill, V., Schaeffer, B., Zimmerman, R., Li, J. 2019. Semi-Supervised Adversarial Domain Adaptation for Seagrass Detection in Multispectral Images. 2019 IEEE International Conference on Data Mining (ICDM). DOI: 10.1109/ICDM.2019.00134
Islam, K. A., Hill, V., Schaeffer, B., Zimmerman, R., Li, J. 2020. Semi-supervised Adversarial Domain Adaptation for Seagrass Detection Using Multispectral Images in Coastal Areas. Data Science and Engineering. 5(2), 111-125. DOI: 10.1007/s41019-020-00126-0
Oguslu, E., Islam, K., Perez, D., Hill, V. J., Bissett, W. P., Zimmerman, R. C., Li, J. 2018. Detection of seagrass scars using sparse coding and morphological filter. Remote Sensing of Environment. 213, 92-103. DOI: 10.1016/j.rse.2018.05.009
Pérez D., Islam K., Hill V., Zimmerman R., Schaeffer B., Li J. (2018) DeepCoast: Quantifying Seagrass Distribution in Coastal Water Through Deep Capsule Networks. In: Lai JH. et al. (eds) Pattern Recognition and Computer Vision. PRCV 2018. Lecture Notes in Computer Science, vol 11257. Springer, Cham
Perez, D., Islam, K., Hill, V., Zimmerman, R., Schaeffer, B., Shen, Y., Li, J. 2020. Quantifying Seagrass Distribution in Coastal Water with Deep Learning Models. Remote Sensing. 12(10), 1581. DOI: 10.3390/rs12101581
Ul Hoque, M. R., Islam, K. A., Perez, D., Hill, V., Schaeffer, B., Zimmerman, R., Li, J. 2018. Seagrass Propeller Scar Detection using Deep Convolutional Neural Network. 2018 9th IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON). DOI: 10.1109/UEMCON.2018.8796636
More details may be found in the following project profile(s):
