Assessing changes in land use and their effects on river flows in tropical catchments of Ecuador

Changes in land use and land cover (LULC) can have direct effects on streamflows at the catchment scale. However, studies of this type are still lacking in some tropical areas of the globe where LULC changes constitute the main antropic changes at watershed scale; for instance, deforestation, and agricultural modification among others. Therefore, for analysing impacts on streamflow, the first challenge is to achieve a good characterization of LULC. 

Here the objective is to evaluate whether these changes have occurred for two small basins on the South American coast in Ecuador (Junín and Paján) and, to analyse their possible effects on flows with special emphasis on minimum flows. Both catchments are characterized by savanna (Aw) climate with  seasonal rainfall patterns and a long period of minimum to zero precipitation. Changes in LULC are studied using two sources of remote sensing information: ESA-CCI-LC at spatial resolution of 300 m and Landsat TM at spatial resolution of 30 m, together with local authorities reports within the last decades. While ESA-CCI-LC directly provides LULC information , LULC was retrieved from surface reflectance after preprocessing (atmospheric and topographic corrections) from Landsat by  using a supervised algorithm. In a preliminary approach the observed changes are compared with variation in streamflow in the outlet of both catchment. 

Results show that for one of the catchments, Junín, ESA-CCI-LC does not reflect changes, while the analysis with Landsat TM shows a decrease of 11.7% of evergreen forest and an increase in agricultural activities. For the other basin, Paján, the first source indicates an increase in evergreen forest, while the second source indicates the opposite, a decrease of 7.4%. The contradictions in these sources highlight the importance of taking into account local knowledge as well as the appropriate selection of spatial resolution in the analysis. Finally, regarding the effects of the LULC changes in streamflow the initial approach was not sufficient to capture any direct effect and therefore, a deeper analysis based on specific features of the streamflow signal are planned as the next step.