Project Mekong is a collaborative research effort aimed at enhancing floodplain and agricultural water management in the Mekong River Delta by using remote sensing and earth observing satellites. Partner institutions include NASA, The Mekong River Commission (MRC), The United States Geological Survey (USGS) Nebraska Water Sciences Center, The University of South Carolina, and Texas A&M University. Funding is provided by NASA's Applied Sciences Water Resources Program. Project Mekong is led by Dr. John D. Bolten and Joseph Spruce at NASA's Goddard Space Flight Center, Dr. Venkat Lakshmi at the University of South Carolina, and Dr. Raghavan Srinivasan at Texas A&M University.

Frequently Asked Questions

What am I looking at?

The default layer on the map shows surface water extent [RED] determined from MODIS satellite imagery captured during the the last 4 days. Classification and cloud filtering algorithms are applied to the imagery to determine surface water. For a full technical description, please contact Aakash Ahamed (aakash dot ahamed at nasa dot gov).

What data is used?

MODIS near real-time LANCE imagery is the primary data source used to determine near real-time surface water. Clouds are filtered out using the MOD35_L2 product. Historic MODIS imagery (M*D09Q1; M*D09A1) and the MODIS permanent water mask (MOD44W) are used to train surface water classifiers which are applied to the near real-time imagery.

How accurate is this?

MODIS surface water products have been validated using multiple high-resolution radar sensors for flood and non-flood conditions during the years 2011 and 2013. Over 7 million pixels were considered in the validation exercise. Validation involved constructing confusion matrices to determine user's and producer's accuracy. Overall accuracy of the product was 87%, and accuracy for any given image ranged from 79% - 98%. Please contact Aakash Ahamed (aakash dot ahamed at nasa dot gov) for more info.

What are the limitations?

  1. Cloud Cover The tropics (including this area), can be cloudy for many consecutive days, which obscures satellite view of the land surface. However, images are temporally composited to reduce cloud cover, and just one cloud-free overpass is all that is needed to make accurate determinations of surface water extent. We use the MOD35_L2 cloud product and the M*D09A1 cloud product to build cloud masks.
  2. Resolution Due to the 250m resolution of the MODIS sensor, water bodies smaller than this are unresolvable.
  3. Mixed Pixels, Cloud Shadows, and Rice Paddy The algorithm may misclassify mixed pixels (those which contain both water and land), and sediment rich water due to spectral similarities to land. Similarly, cloud shadows and rice paddy may be misclassified as surface water due to spectral similarities.

Services That Make This Work Possible

DISCLAIMER: These products are currently undergoing testing and development; NASA does not guarantee product accuracy