Using satellite imagery for stormwater pollution management with Bayesian networks

Area of Science:

• Environmental Science
• Remote Sensing
• Water Quality Management

Background:

• Urban stormwater runoff is a major source of pollution in Santa Monica Bay.
• Monitoring and modeling stormwater pollutant loading is challenging, often relying on land use data.
• Existing methods for estimating pollutant loading from land use are common.

Purpose of the Study:

• To investigate an alternative approach for estimating stormwater pollutant loadings directly from satellite imagery.
• To develop a method for classifying pollutant loads using satellite data.
• To identify areas with the greatest pollutant loading for targeted management.

Main Methods:

• A Bayesian network approach was used to classify a Landsat ETM(+) image.
• The study area was the Marina del Rey area within the Santa Monica Bay watershed.
• Pollutant loads for eight parameters were classified into six levels.

Main Results:

• Spatial estimates of pollutant loads were generated as thematic maps.
• The greatest pollutant loading areas were identified.
• The approach successfully estimated pollutant loads directly from satellite imagery.

Conclusions:

• The developed approach provides a viable alternative for estimating stormwater pollutant loadings, especially where land use data is unavailable.
• The findings can aid in developing best management strategies for stormwater pollution.
• This method supports the establishment of total maximum daily loads in watersheds.