Open AccessEvaluando el desempeño de índices espectrales para identificar humedales alto andinos
Author(s) -
J. Aponte-Saravia,
Jesús Efrén Ospina-Noreña
Publication year - 2019
Publication title -
revista de teledetección
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.271
H-Index - 9
eISSN - 1988-8740
pISSN - 1133-0953
DOI - 10.4995/raet.2019.10580
Subject(s) - vegetation (pathology) , wetland , precipitation , environmental science , vegetation index , index (typography) , geography , vegetation types , physical geography , hydrology (agriculture) , normalized difference vegetation index , forestry , ecology , habitat , geology , leaf area index , meteorology , biology , computer science , medicine , geotechnical engineering , pathology , world wide web
High Andean wetlands are habitats critical to life forms that have adapted to these extreme high mountain ecosystems, and for living beings that inhabit the lower parts of the basin; they are spaces that contain high diversity of flora and fauna characteristic of these places and are strongly associated with the water component. There lies the importance of identifying and monitoring ecosystems, using easy applicable methods and allowing results every two weeks approximately, they are inexpensive and highly reliable. Methods of monitoring in short periods, they are economically profitable and provide reliable information, they correspond to the evaluations by satellite images, specifically applying the methods of spectral indices. Thereby, the objective of the research was to evaluate the performance of six indices, considered to be the most used to identify high Andean wetlands (humidity index at surface level, normalized difference water index, normalized difference vegetation index, enhanced vegetation index, index of vegetation to the surface and tasseled CAP vegetation), in periods of low precipitation, using imagery Landsat 8 OLI. Comparing the performance of those indexes in the identification of wetlands through cross-validation and bootstrap statistical learning, the index that showed better performance was tasseled CAP vegetation, revealing the lowest value of the average of the mean square error of iterations between the test failure rate and training. The index tasseled CAP vegetation, shows greater reliability to identify and evaluate high Andean wetlands.