Coastal Habitats and Conversion to Pond Aquaculture: Myanmar, Thailand, Cambodia and Vietnam

Clark Labs, in partnership with the Gordon and Betty Moore Foundation, has mapped an inventory of pond aquaculture and coastal habitats in Myanmar, Thailand, Cambodia and Vietnam for the 2013-2015 period. The data were developed in support of the Foundation’s Marine Conservation Initiative. The database includes a mapping of present day vulnerability for further conversion to pond aquaculture in areas of mangrove, cropland, stabilized sand and sea salt production. Vulnerability expresses the suitability of the land for conversion without considering the rate of change. Also included is a mapping of the risk of conversion for these same four land covers by 2050 using provincial level rates of change from 1988-1990 to 2013-2015 projected forward to 2050 using compound growth. Note: vulnerability and risk maps are not available for Myanmar at this time, only the land cover map.

The land cover map is based on Landsat 8 imagery from 2013-2015. The vulnerability maps were developed using a multi-layer perceptron neural network and supplementary data including Landsat TM imagery from 1988-1990, SRTM (Shuttle RADAR Topography Mission) elevation, proximity to coast and proximity to pond aquaculture in 1988-1990. The risk maps were developed by prior-probability adjustment of the vulnerability maps.

Explore the data using an interactive web map. In addition, download the data using the links below. The land cover is at a 15m resolution while the vulnerability and risk layers are at a 30m resolution. The data are copyright © 2018 by Clark Labs, Clark University. You are free to use these for non-commercial purposes only. All other applications require permission.

All files are in raster format and distributed as compressed GeoTIFF files. Each zipped archive contains text files describing the spatial referencing system and the legend or data values. For information on the procedures used and accuracy/skill of the mapping, please see:

    J. Ronald Eastman, Stefano C. Crema, Florencia Sangermano, Sean Cunningham, Xiao Xiao, Zhuoyue Zhou, Po Hu, Caroline Johnson, Bernadette Arakwiye and Nelson Crone (forthcoming) A Baseline Mapping of Aquaculture and Coastal Habitats in Thailand, Cambodia and Vietnam.

    Sean Cunningham, J. Ronald Eastman and Stefano C. Crema (forthcoming) Mapping Vulnerability and Risk of Land Conversion for Pond Aquaculture in Thailand, Cambodia and Vietnam


Land Cover 2013 – 2015

Land Cover Classification based on Landsat 8 images from 2013 – 2015 for the coastal zones of Myanmar, Thailand, Cambodia and Vietnam. The coastal zone is defined as:

  • a 10 km buffer from the coastline on the ocean except for islands >200 km from the mainland
  • a 10 km buffer on land as well except if <5 m elevation and <60 km from the coast

Land cover classes include:


Pond Aquaculture, Cage/Raft Aquaculture, Macroalgae Aquaculture, Integrated Mangrove‐Shrimp and Other Aquaculture

Marine Habitats

Seagrass, Macroalgae, Undetermined Aquatic Vegetation, Coral, Rock Substratum, Sand Substratum, Mud Substratum and Undetermined Substratum

Wetland Habitats

Mangrove, Forested Wetland, Marsh, Flat and Open Water

Dryland Habitats

Forest, Shrubland and Grassland

Barren Habitats

Sand/Unconsolidated, Stabilized Sand and Rock Outcrop

Anthropogenic Habitats

Sea Salt Production, Built Up, Cropland and Quarry/Mine

Missing Data

Obscured by Clouds/Smoke and Obscured by Turbidity

Vulnerability of Land Conversion to Pond Aquaculture

The following layers were developed by relating historical transitions to pond aquaculture to covariates such as elevation, proximity to the coast, proximity to existing pond aquaculture and situational information (such as land cover and soil type) as encapsulated in Landsat TM reflectance data from 1988-1990 for Thailand, Cambodia and Vietnam. Historical transitions were established by comparing the 2013-2015 land cover to a classification of Landsat TM data from 1988‐1990 for mangroves, cropland, stabilized sand and sea salt production – the primary classes from which pond aquaculture is developed. The empirical modeling procedure used was the Multi-Layer Perceptron (MLP) Neural Network in the TerrSet software system (Clark Labs). The resulting maps express the vulnerability to transition from each of these four classes to pond aquaculture on a 0-1 scale. Technically, the numbers represent the probability of transition assuming an equal prior probability of transition or persistence. A simpler understanding is that it represents a measure of the vulnerability to transition irrespective of how prevalent the rate of transition is.

Vulnerability Cropland to Pond

Vulnerability Mangrove to Pond

Vulnerability Sea Salt Production to Pond

Vulnerability Stabilized Sand to Pond

Risk of Land Conversion to Pond Aquaculture by 2050

These layers express the probability of conversion to pond aquaculture by 2050 for Thailand, Cambodia and Vietnam. They are derived from the vulnerability maps (above) by means of prior correction that recognizes the expected quantity of transition given the historical rate of change. The procedure is based on the basic Markovian assumption of a constant rate of change and the rate of change determined from 1988-1990 to 2013-2015 (a 25 year span).

Risk Cropland to Pond

Risk Mangrove to Pond

Risk Sea Salt Production to Pond

Risk Stabilized Sand to Pond

Creative Commons License
The work on this page licensed under a Creative Commons Attribution 4.0 International License.