Diagnosis and Evaluation of Mangrove Areas in Venezuela

  • Identification of potential mangrove areas using ASTER satellite imagery, in marine and coastal areas.
  • 119 areas were studied using various digital processing techniques: False color composition,
    Normalized Difference Vegetation Index (NDVI), Principal Component Analysis (PCA), Tasselled
    Cap Transformation and Unsupervised Classification
Project Description

GIS Objetives:

  • The project required to generate a potential map of the mangrove areas in Venezuela, as a base of reference for diagnosis and evaluation of the mangroves.


  • Image Acquisition: Since the extension of the study area cover the entire Caribbean and Atlantic coast of Venezuela, including Lake Maracaibo, 50 Aster satellite images were used.
  • Preprocessing: original HDF image format was transformed to Geotiff, using Hegtools (HDF-EOS to GIS Format Conversion Tool). Using Idrisi (with a macro) all image were transform from UTM, datum WGS84 to UTM datum Provisional South American 1956. Our cartographic reference was 1:100.000 scale map form the IGVSB (Venezuela Geographic Institute) in UTM projection, datum Provisional South American 1956.
    Definition of Study Areas (subimages): Using available literature and information from other projects, 119 windows were identified with the presence of mangroves in the coast of Venezuela. 119 subimages were created.
  • Processing of satellite images: Due to the different features present in the 119 areas of mangroves, a single type of processing to identify the mangroves was not possible. The following processing were done to each of the subimages (using Idrisi with macros):
    • False color composition (RGB: 321 and 342).
    • NDVI (Normalized Difference Vegetation Index).
    • PCA (Principal component analysis).
    • Tasselled Cap Transformation.
    • For each of the previous processing results, Unsupervised Classification was done with each of the following combinations: bands 1 to 9, bands 2,3 and 4, NDVI result (single band), PCA results (PCA1, PCA2, PCA3), Tasselled Cap result (brightness, greenness, wetness).
  • Analysis: This analysis was divided into two stages:
    • Visual identification of potential sites: based on existing maps of the IGVSB, the literature review and studies by other institutions, the images were visually analyzed through the use of false color compositions to identify possible areas of mangroves.
    • Identification of mangrove formations with unsupervised classification: with the results of the different classifications applied to each subimage and selected the classification in which one or more categories correspond to the potential areas identified visually in false color compositions.
  • Field validation: To validate the results of the analysis a field validation is required. Large areas of mangroves reported in the literature are easily identifiable in the images. But there is a level of uncertainty both in determining their limits when they are confused with other types of vegetation formation, or in the case of small formations or sites have not been reported in the literature. A detailed field protocol was generated to capture the data, which specifies the preparation before leaving for the field, the activities to be undertaken in the field sampling points and daily backups of data. With this protocol ensures that fieldwork generates useful information for validation of the interpretation and classification of images. Also for each subimage checkpoints were defined. The field validation was done by another team.
Results (examples):
  • No single type of processing has been positive for the identification of all the mangroves formations. This is due to the heterogeneity of the types of coastal ecosystems where mangroves are present in the Venezuelan coast.
  • Depending on the types of vegetation around mangrove formations, they can help or hinder identification through visual interpretation or classification.
  • Vegetation types with the same spectral response of mangroves may occur. It was therefore necessary to remove from the classified images, some areas with the category of mangroves, which by its spatial context shows that do not correspond to mangroves.
  • The differences between the total mangrove areas reported in the literature and the results of this study could be due to different causes such as:
    • Overestimation of mangrove areas, due to the work scale used in others studies.
    • Real change (decrease) in the area of mangroves between the date of that reported in the literature and satellite images.
    • Underestimation of the mangrove areas, due to errors in identification and classification of satellite images.

Project employer:
Employer: ecoSIG – IVIC
Responsibility: Remote Sensing Leader and Technician
Work team: Rodrigo Lazo
Status: Completed March 2009