28 / 2023-12-20 09:38:45

Change Detection of Land Use and Land Cover for Modeling the Management of Urama Wetland, Venezuela

Change detection,Land use and land cover,tropical wetland,wetland management

The purpose of this study was to analyze dynamics of bitemporal and spatial changes in land use (LU) and land cover (LC) in Urama river basin with a natural wetland as Urama River, Venezuela. The method involved: 1) Phase 1. Satellite image data collection. Seven multispectral images acquired by Landsat sensors were used. The criterion used to analyze LULC change detection was 31-year data collection period from 1986 to 2017, including 1991, 2000, 2008, 2015, 2016 and 2017. 2) Phase 2. Preliminary processing of satellite images. Preliminary processing of Landsat images consisted of performing absolute and relative geometric, radiometric, topographic and atmospheric corrections corresponding to each image. 3) Phase 3. Application of change detection methods. The selected methods were: 1) preclassification: A) algebraic methods (differentiation of reflectance images) and B) transformation method (differentiation of principal component images); 2) Post-classification method. Results of preclassification reflectance image differentiation method. Reflectance differentiation images for six periods (1986-2017, 1991-2017, 2000-2017, 2008-2017, 2015-2017, 2016-2017) showed the occurrence of positive, negative and unchanged values. Most of the Urama basin was covered by vegetation. The decrease in reflectance could be due to the change from vegetation to degraded soil. The increase in reflectance could be due to a change in agricultural use. Results of preclassification method based on differentiation of principal components (PCs) images were obtained by transforming reflectance variables into principal component of each Landsat image. Similarity in reflectance covariance coefficients in optical region bands and difference with respect to band 5 (near-infrared) covariance might arise because predominant landscape cover was vegetation. Results of post-classification method indicated that the proportions of areas associated with LULC classes from classified maps in Urama Basin from 1986 to 2017 showed that the order of prevalence of class occurrence was: a) vegetation (50-85%), b) agriculture (15-20%) , c) rural areas (5-10%), d) bare ground (5-10%), e) water bodies (5%), f) clouds (1-10%) and g) shadows (<1%) . By comparing results of this method with those of PCs method, it was shown that the pattern found in spectral band covariances in optical infrared region coincided with the spectral profile corresponding to vegetation or agricultural areas. Postclassification method allowed validation of classes in which changes or no changes were detected using preclassification methods. The combination of methods allowed higher levels of accuracy in LULC change predictions from the PCs method in scenarios where one class predominates, one of characteristics of tropical wetlands.