National Aeronautics and
Space Administration
Monitoring Marsh Conditions in Coastal Alabama Using NASA Earth Observations to Support the
Alabama Coastal Foundation’s Restoration and Conservation Initiatives
MOBILE BAY
ECOLOGICAL
FORECASTING II
Darius Hixon (Project Lead)
Austin Clark
Tyler Lynn
Manoela Rosa
Monitoring Marsh Conditions in Coastal Alabama Using NASA Earth Observations to Support the
Alabama Coastal Foundation’s Restoration and Conservation Initiatives
MOBILE BAY ECOLOGICAL FORECASTING II
NASA Applied Sciences’ program that collaborates with decision makers to
conduct environmental research projects using NASA Earth observations.
DATA
DEVELOP
DECISION MAKERS
DEVELOP is a dual-capacity building program:
Partners & Participants
DEVELOP bridges the gap between NASA Earth Science and society, building capacity
in both its participants and end-user organizations to better prepare them to handle
the environmental challenges that face society.
NASA DEVELOP
What is DEVELOP?
Apply NASA Earth observations in Mobile and Baldwin Counties to investigate:
Marsh health trends
Marsh extent
Urban development
Objectives
Image Credit :Mobile Bay Eco Forecasting II Team
Community Concerns
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: National Oceanic and Atmospheric Administration
Pollution Trash Dumping Water Quality Turbidity
Marsh Extent Urbanization Ecological Diversity Economic Cost
The study area included Mobile and Baldwin Counties, located in Coastal Alabama
Study Period: Jan 1987-May 2016
Forecast models and maps through 2030
Study Area and Period
0 20 40 60 8010Kilometers ±AlabamaAlabama
Study AreaStudy Area
Image Source: Mobile Bay Eco Forecasting II Team
MARSH HEALTH FORECAST
MARSH EXTENT
URBANIZATION
NASA Satellites and Sensors Used
Landsat 5 - TM
AQUA - MODIS
Image Credit: NASA
TERRA - MODIS
Landsat 7 – ETM+Landsat 8 - OLI
Methodology: Marsh Extent Analysis
Landsat's 5, 7 and 8 images from 1987-2016, once every
5 years
Stack of 4 images each year (12 band composite)
Used segments to create training samples
Segmented Images
Calculated reflectance at the top of the atmosphere
Data
Image Credit: Mobile Bay Eco Forecasting II Team
Methodology: Marsh Extent Analysis
Produced Land Cover Map
Water
Woody Wetland
Urban
Non-Woody Wetland
Upland Herbaceous
Barren
Upland Forest
Analysis
Image Credit: Mobile Bay Eco Forecasting II Team
Masked out everything but the marshes
Calculated area of marshes for each year
Results: Marsh Extent
Marsh extent increasingly changed from 1987 to 2001
Between 2001 and 2006, extent has dropped significantly
Hurricanes Ivan (2004) and Katrina (2005)
Drought (2006 & 2007), Winter (2010)
Conservation efforts &/or natural regeneration21000
21500
22000
22500
23000
23500
24000
24500
1987 1992 1996 2001 2006 2011 2016
Ac
res
Years
Marsh Extent
Marsh Extent
Image Credit: Mobile Bay Eco Forecasting II Team
Methods and Software
Data
Methodology: Urban Development
Analysis
Image Credit: Mobile Bay Eco Forecasting II Team
Normalized Difference Impervious Surface Index (NDISI)
NOAA Impervious Surface Analysis Tool (ISAT)
Top of atmosphere reflectance
2015 Landsat 8
Images
NOAA Coastal
Change
Analysis
Program (C-
CAP) (2001-
2011)
National Land
Cover
Databases
(2001-2011)
Analysis
NDISI Formula: TIR-(VIS+NIR+SWIR)/3 TIR+(VIS+NIR-SWIR)/3
ISAT: C-CAP,
NLCD, and
HUC-12
watershed
shapefiles to
estimate risk,
percent
imperviousness,
and impervious
surface area
0 1 2 3 40.5Miles
±Impervious
Surfaces 2015
Value
High : 1.00
Low: 0.16
Impervious surfaces increased by over 24% in priority watersheds from 2001-2011
Addition of 2000+ acres of impervious surfaces 2001-2011
Average growth of +2.4 percent each year
Seven watersheds exceed 10% surface imperviousness including four priority watersheds
Additional eight watersheds fall into 5-10% warning category
Many of these watersheds contain marshes
Results: Urban Development
R² = 0.9965
0.00
2000.00
4000.00
6000.00
8000.00
10000.00
12000.00
14000.00
Ac
res
Impervious Surface Growth in Priority
Watersheds
2001 2006 2011 2016 (Estimated)
Impervious Surfaces
Trend Line
Years
Image Credit: Mobile Bay Eco Forecasting II Team
Image Credit: Mobile Bay Eco Forecasting II Team
2001 2006
Results: Urban Development
2011
Percent Imperviousness
Average
0.00 - 5.00
5.01 - 10.00
10.01 – 25.00
Datasets
Software and Manipulation
Analysis
Methodology: Marsh Health Trends
8-Day MODIS Composite NDVI Images from the USDA Forest Service ForWarn Dataset from 2000 to 2014
ArcGIS and model builder
Microsoft Excel
Maps and excel time series plots displaying the diversity of health trends in the study area
Image Credit: Mobile Bay Eco Forecasting II Team
NDVI was selected because it is calibrated for chlorophyll in the leaves of plants
This detects seasonality in plant types as well as significant change in biodiversity and leaf coverage
Works by comparing the amount of near infrared light reflected from each pixel to the amount of red light reflected
Methodology: Marsh Health Trends
Methodology: Marsh Health Trends
Marsh health trends were generated using the model pictured at left
This model created graphs in Microsoft Excel and maps to spatially analyze relative trends
Percent maximum was used to account for different plant species
Image Credit: Mobile Bay Eco Forecasting II Team
Mask PMAX Wetlands
1 3 4 7
Image Credit: Mobile Bay Eco Forecasting II Team
Results: Marsh Health
40
50
60
70
80
90
100
2000 2003 2006 2009 2012 2015
Pe
rce
nta
ge
of M
ax
Years
Palustrine (PST) vs. Estuarine (EST)
PST PMAX EST PMAX Linear (PST PMAX) Linear (EST PMAX)
These differences in vegetation can be seen in the health patterns
Trend lines were also drawn, indicating a downward trend in both palustrine and estuarine wetlands, with the palustrine health declining faster than the estuarine
Results: Marsh Health
Other interesting features can be determined from the trends, such as hurricanes, droughts, and harsh winters
Little Lagoon was hit directly by Hurricane Ivan in 2004 and then a drought two years later in 2006, significantly affecting the healthLittle
Lagoon
PMAX
Hurricane
Ivan
Image Credit: Mobile Bay Eco Forecasting II Team
50
60
70
80
90
100
2000 2003 2006 2009 2012 2015
Pe
rce
nta
ge
of M
ax
Years
Little Lagoon
PMAX Linear (PMAX)
Image Credit: Mobile Bay Eco Forecasting II Team
Results: Marsh Health
These trends can help to pinpoint areas in need of restoration following significant stressing events
Ivan
Conclusions
Impervious surface areas are increasing in priority watersheds
Marsh extent declined during study period but there is evidence of recovery
Overall wetland health trends are downward
Health trends seem to be most impacted by weather conditions and urbanization
Image Credit: Mobile Bay Eco Forecasting II Team
Spatial and temporal resolution
In situ monitoring
Ground truth validation
Atmospheric correction
Cloud Removal
Processing time
Image Credit: Mobile Bay Eco Forecasting II Team
Errors and Uncertainties
Acknowledgements
This material is based upon work supported by NASA through contract NNL11AA00B and cooperative agreement NNX14AB60A.
Bernard Eichold, M.D., Dr. PH, Mobile County Health Department
Kenton Ross, Ph. D., NASA Langley Research Center
Saranee Dutta, Previous contributor
Vishal Arya, Previous contributor
Jeanett Bosarge, Previous contributor
Courtney Kirkham, Previous contributor
Mark Berte, Alabama Coastal Foundation
Just Cebrian, Ph. D., Dauphin Island Sea Lab
Mobile Bay National Estuary Program
Eastern Forest Environmental Threat Assessment Center of the USDA Forest Service
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s)
and do not necessarily reflect the views of the National Aeronautics and Space Administration.
Image Credit: Mobile Bay Eco Forecasting II Team