Applying NASA Earth Observations to Mitigate the Impacts of Chilean Drought

By Chile Water Resources Team , posted on November 17th, 2013 in DEVELOP Fall 2013 VPS


Elevation zone and MODIS snow cover extent maps of the Limarí Basin, Coquimbo, Chile. Image Credit: Chile Water Resources Team, NASA DEVELOP National Program.

Team Location: Langley Research Center, Hampton, Virginia

Joshua Kelly, Project Lead (University of Rhode Island)
Bethany Burress (Christopher Newport University)
Jeffry Ely (Old Dominion University)
Ajoke Williams (Massachusetts Institute of Technology)
Amberle Keith (Idaho State University)

Kenton Ross, Ph.D. (NASA, DEVELOP National Science Adviser)
James Favors (Wise County, DEVELOP National Program)
Angelica Gutierrez (National Oceanic and Atmospheric Administration)

Past/Other Contributors:
Pedro Bejares (Embassy of Chile to the United States)
Joaquin Tagle (Embassy of Chile to the United States)
Ricardo Cabezas Cartes (Centro de Información de Recursos Naturales)

For the past four years, Chile has been experiencing a record drought that has driven broad-ranging stressors for many sectors in the country, most notably water resources and agriculture. This has been especially true in the northern regions of Chile, such as the LimarÍ River Basin, where desert-like conditions already exist and strains on limited water resources are more pronounced.  This basin is located in the Coquimbo region, or Region IV, of Chile and covers an area of 11,696 square kilometers.

A major source of available water is snowmelt from the adjacent Andes Mountains where snow can accumulate above 2,500 meters. Maximum river discharge is reached during the late spring and summer months (October-January) coinciding with the growing season. The Centro de Información de Recursos Naturales (CIREN) maintains and interprets images and data as they relate to agriculture and natural resources within Coquimbo. Currently, measurements of streamflow and available water from snowpack are limited to sparse monitoring stations in the mountains; therefore forecasts of available water for the growing season are unreliable.

This project used existing hydrologic and temperature data from the monitoring stations in combination with remote-sensing techniques to better determine potential water availability for the 2013 growing season (September to November). Global digital elevation data collected by the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor on board the Terra satellite was used to delineate the watershed boundary of the Limarí Basin. The daily snow cover extent within the Limarí Basin was determined and mapped by using data collected by the Moderate Resolution Imaging Spectroradiometer (MODIS) on board the Terra satellite. Snow cover data was used with other hydrologic data in the Snowmelt Runoff Model (SRM) created by the U.S. Department of Agriculture (USDA) to forecast the amount of daily Andean snowmelt discharge. These predictions will help the regional water group decide how to allocate water most efficiently in the community.

Return to the Fall 2013 VPS page.

facebooktwittergoogle_plusredditpinterestlinkedinmailby feather

Sam Weber
Sam Weber

That's some crazy elevation changes, sweet project! In your discharge equation, what does the recession coefficient represent?

Lauren Makely
Lauren Makely

It sounds like your partners will be very impressed with your progress! What kind of future work do they have in mind?


You all achieved great products for your partner.  Nice work.  I'd like to know more about the script you wrote to automate the collection and calculation of MODIS data. Is there more general info you could give?


@Sam Weber Thanks Sam! The recession coefficient describes the distribution of discharge over several days resulting from a single precipitation event.  


@Lauren Makely Thanks Lauren! We are planning on continuing to work with the Embassy and CIREN on expanding the capabilities of the model by incorporating forecasted temperature, precipitation, and snow cover data.  This would allow us to provide them with actual water availability estimates.



Jeff here, was on this project.

The process is currently in 3 steps. 

1: python scripts downloads all relevant MODIS data for an input year.

2: ArcMap model made with model builder batch processes the MODIS data to prepare it for matlab, outputing tif images which have been re-projected, upsampled and clipped to match the elevation zone image file.

3: Matlab script loads each image, starting with the first day for the input year and after ensuring the modis data and elevation zone tiff are properly aligned, performs the following basic logical check on every pixel. If a pixel is clouds today, but was snow yesterday, that pixel is corrected to snow. This operates on the assumption that clouds are simply passing over the snow and imposes reasonable continuity on the snow covered area profiles. This is done forward in time for each day, and statistics on snow covered area per elevation zone are produced and output to an excel file for easy input into WinSRM.

In the future, intentions are to integrate (and translate the Matlab part) this entire process into one python script.


@JeffEly - wow, that sounds really impressive!  Great work!  Is this project continuing? 


@anjward7 We are planning on continuing the project in the spring term. We are working on expanding the capabilities of the model by developing snow cover forecasting techniques and incorporating them into the SRM.  This would allow us to provide them with actual water availability estimates for the 2014 growing season.