Dataset: Data from an analysis of habit change in Port Fourchon, LA from 2002, 2014, and 2022

Data not availableVersion 1 (2024-10-28)Dataset Type:Unknown

Principal Investigator: James Nelson (University of Georgia)

Student: Herbert Leavitt (University of Georgia)

Student: Alexander Thomas (University of Georgia)

BCO-DMO Data Manager: Amber D. York (Woods Hole Oceanographic Institution)


Project: CAREER: Integrating Seascapes and Energy Flow: learning and teaching about energy, biodiversity, and ecosystem function on the frontlines of climate change (Louisiana E-scapes)


Abstract

This data set contains the analysis of habit change in Port Fourchon, LA from 2002, 2014, and 2022. The analysis uses LandSat9 data to determine the change in black mangrove cover across the three time points.

XXX

Views

XX

Downloads

X

Citations

Location description:   All data for this analysis were collected near Port Fourchon, Louisiana, USA (29.10 °N, 90.19 °W). The marshes around the port are microtidal, with a mean tidal range of ~0.37 m. The site sits at the precise edge of black mangrove expansion into saltmarsh habitats and although some land loss in the areas has occurred, mangroves in the area have been expanding since the 1990s (Osland et al., 2013).

 

Satellite Imagery

     Satellite imagery data products used for analysis in this paper were obtained solely from the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS). Multispectral satellite imagery was chosen based on the best available data for the given year of observation. Image analysis for the year 2022 used Landsat-9 satellite imagery, analysis for 2014 utilized Landsat-8 satellite imagery, and analysis for 2002 utilized Landsat-7 satellite imagery. Imagery from each satellite displays a spatial resolution of 30 m for each spectral band. All image layers were clipped from their original sizes to encompass the greater Port Fourchon, LA marsh, spanning an area of 49,427 hectares.

     Of the eight spectral layers acquired from the U.S Geological Survey, six were isolated and used for segmentation and analysis. The six spectral layers imported into eCognition for this project corresponded to the mean pixel values for Red, Green, Blue, Near Infrared (NIR), and two bands of Short-Wave Infrared (SWIR). A Normalized Difference Vegetation Index (NDVI) was calculated using the color bands Red and NIR. Habitat classification was completed using eCognition Developer (Version 10.4, Trimble Germany GmbH, Munich, Germany) through a combination of manual rule set class assignments and a Nearest Neighbor classification algorithm (NN). The NN classification algorithm categorizes image objects based on their respective spectral values. These values of interest are manually input into a training set along with other parameters of interest defined by the user. The features influencing the NN algorithm for this project include mean spectral values for: Red, Green, Blue, NIR, and NDVI.  

      Habitat delineation began with separating the class “water” from the mosaic. A multiresolution segmentation algorithm using a scale parameter of 30, a shape parameter of 0.3, and a compactness parameter of 0.75 (using the bands Red, Green, and Blue) followed by a spectral difference algorithm with a parameter of 50 (using all layers) segmented and merged the majority of water habitat into a super polygon. The remaining areas of water were delineated through a combination of algorithmic rule set assignments and NN classification, ensuring all water bodies were accurately classified. The second class to be classified included all man-made structures within Port Fourchon. Man-made structures included all areas of human development including port structures, roads, debris found within the marsh, as well as regions of sand accumulation from restoration efforts. Man-made structure also included the beach area lining the entirety of the southern end of the port as this region has seen extensive human restoration throughout the years (CPRA Master Plan 4th edition 2023). Isolation of man-made structures included the use of a multiresolution algorithm with a scale parameter of 5, a shape parameter of 0.3 and a compactness parameter of 0.75 (using the bands: Red, Green, Blue, NIR, and NDVI). Rule set class assignments utilizing the red band were especially useful in isolating these areas of human development and needed little manual correction. The vegetated areas of Spartina alterniflora and Avicennia germinans were subsequently classified using the same combination of algorithmic rule set assignments and NN classification. NDVI was used as a major parameter for vegetated habitat classification as it is particularly useful for distinguishing between Spartina alterniflora marshland and Avicennia germinans habitat (Chandra et. al 2010) due to the mangroves increased mean NDVI value. All classes were subsequently merged into their respective polygons and exported as shapefiles for area analysis in R statistical software.


Related Datasets

No Related Datasets

Related Publications

No Related Publications