The ECOPOTENTIAL Virtual Lab – I:

A Tool for Raster maps of NDVI, Water Turbidity and Flood Masks from Remote Sensing data

An interview with Javier Bustamante, EBD-CSIC, Spain

In this interview with Javier Bustamante (CSIC, Spain) we present here a useful tool for deriving raster maps of NDVI, water turbidity and flood masks from remote sensing data developed by the LAST-EBD remote sensing laboratory of the Spanish National Research Council.

The tool can be used through the ECOPOTENTIAL Virtual Laboratory (

Javier, you are the coordinator of the Doñana Biological Station group that takes part in the ECOPOTENTIAL project. Can you introduce yourself and your research group?

I am Javier Bustamante, a 56 years old ecologist working as researcher at Doñana Biological Station (EBD), a research institute of the Spanish National Research Council (CSIC) located in Seville and managing a scientific reserve at the Doñana National Park in the South West of Spain. I became interested in the use of GIS and remote sensing tools for ecology and conservation biology and in 2003 we created at our institute the Remote Sensing and GIS lab (LAST-EBD) where I am the Scientist in charge. I coordinate the work of CSIC in ECOPOTENTIAL project where 10 scientists are involved, but it has been Diego García, a 40 years old Geographer and Python/GIS programmer, who has taken the hard work of transforming our models into code lines in the ECOPOTENTIAL Virtual Laboratory VLAB.

Can you describe the tool that you developed?

LAST EBD scheme

Our service in the ECOPOTENTIAL Virtual Lab is called LAST-EBD Flood mask, Water Turbidity and NDVI (yes, the name is pretty descriptive).  It generates a series of thematic raster maps starting from a satellite image. Being specific, we use a scene of the Landsat series of satellites from the sensors TM, ETM+ or OLI from which we derive a vegetation index (NDVI), a water turbidity index, and flood mask. The process first generates a normalized image based on pseudo-invariant areas (PIAs) and, based on it, the indicated products. The normalization process is a crucial step to be able to apply the same algorithms to images from different satellite sensors and dates. In a last step, that we are still developing, the flood masks are used to compute the annual hydroperiod. 

Can you explain in more detail what is it meant for? 

Our model aims to improve the monitoring of certain environmental phenomena that can be studied with remote sensing. The vegetation index like NDVI, for example, is important for a whole series of environmental issues, from the primary production of marsh vegetation to the tracking of forest fires. The monitoring of presence of water in the marsh (which is the heart of Doñana National Park) and the water turbidity, are vital to understand the spatial distribution habitats for waterbirds in the marshland, and even to study issues related to climate change. The models have been developed to cope with a long time series of satellite data and is currently tuned to be applied in Doñana Natural Space. 

What are the outputs of the model?

When the calculation process is completely finished, a “scene_output.tar.gz” folder is created, containing three raster files (in GeoTIFF format) of the NDVI, the Flood mask and Water Turbidity mask for the input scene.

Who are the potential users of this service?

Potential users mainly are the managers and technicians of Doñana Natural Space, who need to know the surface of the flooded area, the primary production of the vegetation and the water quality. Nevertheless, our model can be applied to any other Landsat scene; you just need to change the data in the input data folder as these data are specific for the Doñana scene, with some pseudo invariant areas of the new scene and also a new reference scene to perform the normalization.


What should a potential user do to use your model? 

Our model runs with Landsat TM, ETM+ and OLI Sensors (Landsat 5, 7 & 8 satellites) for the scene: path 202 – row 34. In order to run it, you need to feed it with two inputs, both compressed in a tar.gz file: one folder called “scene”, with the compressed Landsat scene to be processed, and other folder called “data “with all the additional info needed to run the process. Second input (“data”) would remain invariable to all Landsat scenes to be processes, so it would be always called through this link:

Eventually, users can find more info about our process in GitHub (

Whom to contact to have further information?

Javier Bustamante ( or Diego García (