Skip to Main content Skip to Navigation
New interface
Theses

Potentialités des données de télédétection optique et radar libres d’accès pour l’évaluation et le suivi des écosystèmes forestiers tropicaux : étude de cas au Togo, en République Démocratique du Congo, en Guyane française et en République Dominicaine

Abstract : This study aims to explore the potential of optical remote sensing and free access radar data for the assessment and monitoring of tropical, dry or wet forest ecosystems. Different test sites located in these tropical forest ecosystems have been selected. These are: the Virunga National Park in the Democratic Republic of Congo (DRC), the Oti-Keran-Mandourie Biosphere Reserve (OKM) and Togodo Wildlife Reserve (RFT) in Togo, the area around the bridge linking the city of Saint-Georges de l'Oyapock and the plain of the Kourou coast in French Guiana and the province of Monté Cristi in the Dominican Republic. Different data were used in this study: for radar images, Sentinel-1 time series, Alos-2 mosaics and, for optical images, Sentinel-2 and Landsat-8. Exogenous data such as GPS points, digital terrain models and reference maps. The methodology approach used consists of pretreatment on optical and radar images. Specific approaches, varying by study site, included: photo detailed interpretation, supervised SVM classification, forest inventory and application of allometric equations, a wavelet decomposition detection approach, a detection approach automatic changes by thresholding and the characterization of these changes. The main results are:PNVI site: land cover maps and forest, non-forest binary maps of 1987, 1997, 2007 and 2017 are produced on the PNVI. Over the 30-year period using the binary maps between 1987 and 2017 the average annual rate of deforestation is 1.07%. This high deforestation rate shows the increasing pressure on forest resources in the Virunga landscape. OKM and RFT site: a classification carried out on a combination of optical and radar images gives slightly better performances than classifications carried out on optical and radar images considered separately. The land cover maps from these classifications were used as a basis for estimating carbon stocks through forest resource assessment. At the Saint-Georges de l'Oyapock site, temporal analysis using wavelet decompositions revealed three main types of changes due to anthropogenic deforestation, seasonal changes and agricultural changes. On the site of the Monté Cristi province in the Dominican Republic, the joint analysis of radar and optical images made it possible to propose a cartography comprising 18 field-controlled land cover classes with an overall accuracy of more than 90%. Historical forest monitoring shows a decline in forest cover. At the same time, we observe a regression of the surface of mangroves between 2015 and 2018.This study has highlighted the immense potential of optical and radar remote sensing data in the characterization, mapping and monitoring of land use layers in tropical ecosystems in different regions of the world and according to seasonal conditions. While each type of remote sensing data has these discriminatory qualities and capabilities, this study has shown that the joint and combined use of two types of data significantly increases the characterization and discrimination of land-use classes and thus increases the chances of reliability of the actions to be carried out
Complete list of metadata

Cited literature [200 references]  Display  Hide  Download

https://theses.hal.science/tel-02384384
Contributor : ABES STAR :  Contact
Submitted on : Thursday, November 28, 2019 - 12:36:07 PM
Last modification on : Wednesday, September 28, 2022 - 5:52:31 AM
Long-term archiving on: : Saturday, February 29, 2020 - 3:22:20 PM

File

TH2018PESC2181.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02384384, version 1

Collections

Citation

Anoumou Kemavo. Potentialités des données de télédétection optique et radar libres d’accès pour l’évaluation et le suivi des écosystèmes forestiers tropicaux : étude de cas au Togo, en République Démocratique du Congo, en Guyane française et en République Dominicaine. Sciences de l'information et de la communication. Université Paris-Est, 2018. Français. ⟨NNT : 2018PESC2181⟩. ⟨tel-02384384⟩

Share

Metrics

Record views

455

Files downloads

70