NASA DEVELOP: Louisiana Ecological Forecasting

By Ross Reahard

The Concern

Southeast Louisiana, including St. Bernard Parish (Figure 1), is very susceptible to tropical cyclones and high water caused by seasonal flooding events. Continued degradation of coastal forests and wetlands causes St. Bernard to lose more of its natural protective barriers, which in turn, exposes it to greater water inundation risk.  In 2005, Hurricanes Katrina and Rita severely damaged many ecosystems that served as natural barriers, which provided protection against the human and non-human populations of St. Bernard Parish.  In order to protect the communities within St. Bernard Parish, restore coastal biodiversity, and provide measures to counteract coastal land loss, it is imperative that these vital coastal ecosystems be rebuilt. Volunteer groups, non-governmental organizations (NGOs) and government entities often work separately and independently of each other and use different sets of information to choose the best planting sites for restoring coastal forests.

Loss of coastal forests has not only increased the vulnerability of human populations to storm impacts but has also, through the conversion of swamp and marshland to open water, decreased the amount of biodiversity in the parish. Cypress-tupelo swamps, as well as marshlands, provide vital habitat to several threatened or endangered species such as the brown pelican (Pelecanus occidentalis), bald eagle (Haliaeetus leucocephalus), whooping crane (Grus americana), and the Louisiana black bear (Ursus americanus luteolus).

In the early 1960s, when the Mississippi River Gulf Outlet (MRGO) canal was dredged as a connection from New Orleans to the Gulf of Mexico via Breton Sound, it resulted in a dramatic increase in salinity, which led to the destruction of thousands of acres of wetlands east of New Orleans. The MRGO canal was intended to help the area prosper economically, but instead it has only added to the numerous issues that plague this region. The construction of the MRGO has both directly and indirectly led to major damaging effects to its surrounding ecosystems, including land loss, erosion, increases in salinity, and habitat shifts and degradation.  The U.S. Army Corps of Engineers estimated a loss of 1,500 acres of cypress swamps, as well as damage to 10,300 acres of brackish marsh, 4,200 acres of saline marsh, and 3,400 acres of fresh/intermediate marsh [1], as a result of the construction of MRGO.  The resultant damage and loss that has occurred has, in part, been due to the fact that, in certain areas, the MRGO has expanded beyond its designed width of 650 feet, to over 3,000 feet, primarily through erosion of the banks. This has enabled saltwater to further permeate inland. By the time the MRGO was completed in 1965, the saltwater intrusion, on average, increased the salinity from 3.5 ppt (1959 to 1961) to 12 ppt, which resulted in killing and/or damaging over 600 hectares of swamp forest.

The Science

For this project, using NASA Earth Observing Systems (EOS), Natural Resources Conservation Service (NRCS) soil surveys, ancillary road and canal data, along with ground truthing, a comprehensive geographic information system (GIS) to help identify suitable planting sites for baldcypress and several other wetland tree species in St. Bernard Parish was created.

The team began by identifying the location of suitable soil types for planting in St. Bernard Parish. Soil types were classified and rated on a scale of 0 to 4, where 4 was the most suitable soil rating for the particular tree species of interest (Figure 2).  Using the detailed descriptions of local soil types published in a report by the United States Department of Agriculture (USDA) [2], a soil suitability index was created for each of the 14 trees defined by the report as native to the parish.  The most suitable soil types for each species were indexed as “4 – Well Suited.” The descriptive soil unit section of the report also contains information about which soil types are saline and which are brackish. These soil types, as well as those labeled as Urban and Dumps, were given a suitability index of “0 – Not Suited” for all tree species. Then, the intermediate ratings of “1 – Poorly Suited,” “2 – Fairly Suited” and “3 – Moderately Suited” were established by comparing the characteristics (water table depth, salinity, texture, etc.) of the preferred soil types for each tree species. From these data, researchers created a chart (Table 1) to visually display soil suitability for each tree species. 

Elevation data, in the form of a pre-processed Digital Elevation Model (DEM) at 1/9 arc-second (approximately 3 meters) spatial resolution, were obtained from the USGS National Elevation Dataset (NED) through the National Map Seamless Server (NMSS).  Accurate, high resolution elevation data were integral to the project because elevation is a key factor in determining suitable planting sites for baldcypress; this is because the tree grows best at elevations of 2.5 to 6 feet, or about 0.5 to 2 meters.  Then, for ease of analysis and to create an easy to interpret final product for the end-user, the DEM was recoded (Figure 3) to show elevation values using half-meter increments.

TIGER roads shapefiles, a standard GIS product from the Census Bureau, which contain transportation infrastructure data, were downloaded from the U.S. Census Bureau. The TIGER roads shapefile was quality checked, updated and edited to show highways, streets and private access roads with differing symbology. Access to transportation infrastructure is an important consideration for the end-users who are responsible for selecting potential planting sites since large amounts of trees and equipment must often be moved from a staging area to the final planting site. Parish boundary and water body shapefiles were downloaded from LSU Atlas Statewide GIS server.  The team also created a point shapefile, showing the location of pumping stations in the study area. Pumping stations drive freshwater, largely from urban runoff, into areas outside of the levee protection. While in the field, the DEVELOP team observed relatively healthy stands of baldcypress in close proximity to the pumping stations. This observation suggests that these locations create suitable habitats for the trees, most likely because of the discharge of freshwater into brackish or more saline environments; additionally, the roots of baldcypress seedlings require constant inundation with freshwater for proper development.

For the years 2010 and 2011, Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) imagery was obtained from the NASA Jet Propulsion Laboratory’s (JPL) AVIRIS 2006-2010 Flight Data Locator.  The processing procedure for the hyperspectral AVIRIS data cube predominantly followed methods outlined by Spruce et al. [3] [4]. To create a land cover classification map, the data were first spatially and spectrally subset to a working extent using 15 of the 224 available channels that effectively discriminate vegetation cover types (Table 2). The 30 classes were analyzed by overlaying the classified image on the original 3.5m true color composite and recoded into 4 groups: water, marsh, forest/scrub, and urban (Figure 4).

AVIRIS data from a single low altitude swath were also processed to calculate a Red Edge Normalized Difference Vegetation Index (NDVI). This analysis was conducted to highlight the robust capabilities of AVIRIS, as well as demonstrate to end-users one potential method of monitoring restored vegetation health. AVIRIS data from May 5, 2010 were atmospherically corrected and calibrated from radiance to reflectance using the QUick Atmospheric Correction (QUAC) method in ENVI Classic.  The Red Edge NDVI, a narrowband greenness vegetation index modified to focus on reflectance values along the red edge, was calculated through the Vegetation Index Calculator in ENVI.  This index uses greenness values as a proxy for vegetation health and vigor and is thus contingent upon the time of year that the data are collected. However, if used responsibly, this index can yield valuable information on stand health at very small spatial scales. Then, after the planting process is complete, vegetation indices, such as NDVI, can easily be calculated by end-users to monitor spatio-temporal trends in the health of restored forests.

In order to extend the analysis of coastal forests into the pre-space-borne era of earth observations, historic aerial photography was obtained from USGS’s Earth Explorer. The historic images were used to estimate the spatial extent of forested land pre-MRGO (Figure 5). Imagery collection dates include April and May of 1952 and February of 1956. Following acquisition, the historic imagery was cropped, georeferenced and mosaiced together through the use of ArcMap and ERDAS Imagine software. Features such as image texture and shadowing were used to help delineate forested from non-forested land.  

The Benefit

The Louisiana Ecological Forecasting team partnered with several local non-profit groups including the Wetlands Tree Foundation, St. Bernard Wetlands Foundation, and The Meraux Foundation. These groups are directly involved in potting and raising baldcypress seedlings and other native tree species and work extensively with volunteer groups to conduct plantings. Often, local knowledge is the main deciding factor when choosing where to plant.  Parish politics and land ownership issues sometimes prevent access to suitable sites. The DEVELOP team was able to provide our partners with unbiased scientific information (Figure 6) on the best-suited locations for baldcypress tree planting, and the ultimate expected outcome is that the planning process, as well as the in-the-field work, can be streamlined.

The DEVELOP team also partnered with the St. Bernard Parish Planning Commission’s Coastal Advisory Board. The Parish Planning Commission is very interested in including the DEVELOP team’s results in its upcoming master plan. A team member is also scheduled to present the team’s findings at the Planning Commission’s next meeting in October 2012.

Benefits of this work should be seen almost immediately. Seedlings will begin to go into the ground in early spring, and the maps created by DEVELOP will guide where the trees are planted. Not only will there be an immediate impact, but the results will continue to help guide coastal reforestation efforts into the foreseeable future. In addition, the basic methodology we developed can be easily tailored to fit a variety of other coastal restoration applications, which further demonstrates the usefulness and role NASA EOS can have in rebuilding coastal Louisiana.

Results from this project have already been presented to project partners at Stennis Space Center and were featured in the summer DEVELOP close-out presentation given at NASA Headquarters in Washington, D.C. The project was awarded first place in a virtual poster session sponsored by Esri and hosted on Institute of Electrical and Electronic Engineers’ (IEEE) Earthzine website. The team has also submitted this project’s abstract for possible presentation at the American Geophysical Union’s Annual Fall Meeting in San Francisco, as well as the Bays and Bayous Symposium in Biloxi, Mississippi.

Figure 1. St. Bernard Parish is located to the southeast of New Orleans, Louisiana.  The parish has a total area of 1,330 square miles, including 865 square miles of water and 465 square miles of land, which is the largest percentage of water per total area of all parishes in Louisiana.  The natural elevation of St. Bernard Parish varies from 6 feet below sea level to about 12 feet above sea level. Marshes and swamps which have been drained or impounded by levees are typically below sea level, and the natural levees and ridges along the Mississippi River and its many former distributaries are among the highest features in St. Bernard Parish.

Figure 2. Soil suitability map or Baldcypress (Taxodium distichum)

Figure 3. 3m DEM displaying non-optimal baldcypress elevations in gray scale (white for highest and black for lowest elevations). Colored areas are best suited for baldcypress, occurring at elevations of 0.5 to 2 meters above mean sea level

Table 1: Tree species and ratings for soil types

  • Baldcypress – Taxodium distichum
  • Water Tupelo – Nyssa aquatic
  • Black Willow – Salix nigra
  • Water Hickory – Carya aquatica
  • Overcup Oak – Quercus lyrata
  • Green Ash – Fraxinus pennsylvanica
  • Sugarberry – Celtis laevigata
  • Eastern Cottonwood – Populus deltoides
  • Nutall Oak – Q. nuttallii
  • Water Oak – Q. nigra
  • Pecan – Carya illinoinensis
  • American Sycamore – Plantanus occidentalis
  • Willow Oak – Q. phellos
  • Sweetgum – Liquidambar syraciflua

Table 2. Table from Spruce et al. 2002 showing the channel subset and associated AVIRIS bands, wavelengths, and reflectance regions used for the land cover classification

Figure 4. Land cover classification created from AVIRIS imagery showing location of forested and relatively open land. The zoomed in area shows the intersection of MRGO and Bayou la Loutre

Figure 5. Approximate extent of forested land pre-MRGO. Background imagery is a combination of 1952 and 1956 aerial photos acquired from USGS.

Figure 6. Soil ratings for suitable elevations located in relatively open areas, since forested and developed/urban areas are not suitable for planting seedlings.

[1] Caffey, R.H. and B. Leblanc.  2002.  “Closing the Mississippi River Gulf Outlet: Environmental and Economic Considerations.”  Interpretive Topic Series on Coastal Wetland Restoration in Louisiana, Coastal Wetland Planning, Protection, and Restoration Act, National Sea Grant Library, no. LSU-G-02-004: 1-4.
[2] Trahan, Larry J., Jeanette J. Bradely, Lyfon Morris, Donald R. McDaniel, and Richard Nolde.  1989.  "Soil Survery of St. Benard Parish, Louisiana." Report, USDA and Soil Conservation Service.
[3] Spruce, J.P. 2001. “Low-altitude AVIRIS data for mapping land cover in Yellowstone National Park: Use of ISODATA Clustering Techniques.” Proceedings of the Tenth JPL Airborne Earth Science Workshop, February 27-March 2, Jet Propulsion Laboratory, Pasadena, California, pp. 387-396.
[4] Spruce, J.P., E.G. Otvos, M.J. Giardino. 2002. “Low-altitude AVIRIS data for mapping landform types on West Ship Island, Mississippi.” Accessed online August 3, 2012: 

The author would like to graciously thank the LA Eco team members from the summer term: Maria Arguelles, University of Miami; Michael Ewing, University of Southern Mississippi; Chelsey Kelly, University of Southern Mississippi; and Emma Strong, University of Southern Mississippi, as well as our advisor and mentors
Dr. Russell Lambert, CSC and Joe Spruce, CSC. We would also like to thank the DEVELOP National Program Office and our many partners and collaborators. Without their help and support this project would not have been possible.

Editor’s Note: The DEVELOP National Program is a capacity building internship sponsored by NASA’s Applied Sciences Program that provides interns the opportunity to learn about NASA Earth Science and the practical applications of Earth observations.

Published Thursday, November 1st, 2012

Written by Ross Reahard