2024 CSDMS meeting-077
From CSDMS
Revision as of 12:15, 18 March 2024 by Cresposito (talk | contribs) (Created page with "{{CSDMS meeting personal information template-2024 |CSDMS meeting first name=Christopher |CSDMS meeting last name=Esposito |CSDMS meeting institute=The Water Institute |CSDMS meeting city=New Orleans |CSDMS meeting country=United States |CSDMS meeting state=Louisiana |CSDMS meeting email address=cesposito@thewaterinstitute.org }} {{CSDMS meeting select clinics1 2024 |CSDMS_meeting_select_clinics1_2024=5) Coastal evolution analysis and inundation modeling with GRASS GIS }...")
(if you haven't already)
Log in (or create account for non-CSDMS members)
Forgot username? Search or email:CSDMSweb@colorado.edu
Browse abstracts
Advancing Capacity for Marsh Modeling Retrospective Analysis: Synthetic Historical DEMs for Model Initialization and Validation
Christopher Esposito,
The Water Institute New Orleans Louisiana, United States. cesposito@thewaterinstitute.org
Harris Bienn, The Water Institute Baton Rouge Louisiana, United States. hbienn@thewaterinstitute.org
Maricel Beltran-Burgos, The Water Institute New Orleans Louisiana, United States. mbeltranburgos@thewaterinstitute.org
Renee Collini, The Water Institute NEW ORLEANS Louisiana, United States. rcollini@thewaterinstitute.org
High quality Digital Elevation Models (DEMs) do not exist in coastal wetlands prior to the widespread use of aerial LiDAR beginning in the early 2000's. This makes it difficult to develop models that capture the historical evolution of specific coastal marshes, creating a challenge in communications between the modeling community and wetland managers who seek to understand model outputs in the context of their experience, observations, history of management decisions, and perception of risk.
The project team is working with managers at four coastal wetlands to advance a method that will fill this data gap using historical remotely sensed imagery, historical in-situ observations, and machine learning. The team will compile Landsat imagery collected within one year of an existing high quality DEM. The suites of Landsat imagery will be processed to produce maps showing inundation frequency based on the Normalized Difference Water Index (NDWI), and these will be used as training data for a deep learning image segmentation model that relates inundation frequency with wetland elevation. The segmentation model will then be validated with observational data and applied to the period before DEMs are widely available but during which Landsat sensors are consistent with today’s standards (i.e. 1984 to the present).
