Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Suède → NDVI time series analysis for desert locust outbreak detection and quantification analysis of its impact on vegetation productivity of Sahel

Lund University (2020)

NDVI time series analysis for desert locust outbreak detection and quantification analysis of its impact on vegetation productivity of Sahel

Oikonomopoulos, Evangelos

Titre : NDVI time series analysis for desert locust outbreak detection and quantification analysis of its impact on vegetation productivity of Sahel

Auteur : Oikonomopoulos, Evangelos

Université de soutenance : Lund University

Grade : Master’s Degree (Two Years) 2020

Résumé
It has been shown that the semi-arid environment of the Sahelian belt plays an important role in the global carbon uptake as the fluctuation of its primary productivity can be determinant for the global carbon uptake. The insect Schistocerca gregaria commonly known as the desert locust is a disturbance factor that can affect the vegetation productivity for consecutive seasons. In this study the detection of the desert locust outbreaks during the 2003-2005 upsurge was attempted through a Time Series Analysis of seasonal NDVI (small integral) coupled with detailed field observations. In cases of severely infested pixels the seasonal greenness could appear to be lower during the suffering season of 2004 compared to the 14-year interannual average of the pixel examined, indicating that the pixel is infested. At the same time Sahel’s gross primary productivity (GPP) loss due to the locust outbreak was specified by comparing modeled gross primary productivity estimations between the outbreak and the non-outbreak years. The negative impact on primary productivity was also analyzed for the various stages of locust infestations. The process was based on a z-score statistical analysis while the index of Rain Use Efficiency was introduced in order to optimize the method. The outcome of the study showed that the decrease of seasonal NDVI in the infested pixels during the year of the upsurge is not statistically significant and consequently a relationship between greenness and locust infestations cannot be established for the development of a detection tool. Secondly, a significant negative impact on vegetation productivity due to locust outbreaks cannot be confirmed by analyzing the negative divergence from its interannual average. Furthermore, the claim that NDVI and GPP would indicate exceptionally low productivity performance during the year of the upsurge is disproved by the fact that also in years in which the studied pixels were not infested, those measures demonstrate similarly low behavior. The addition of the RUE index was not proven successful as it was shown that the RUE was highly regulated by the fraction’s denominator fluctuation, precipitation. Finally, it has been shown that the negative impact on productivity and seasonal greenness slightly variates between the different stages of infestations. The inadequacy to establish relationship between the field observations and the greenness can be attributed to various factors such as the spatial resolution, the heterogenous nature of vegetation and the field observation selection criteria and treatment.

Mots Clés  : Physical Geography and Ecosystem Analysis, Primary Productivity, Desert Locust, NDVI, Time Series Analysis, Outbreak Detection, Food Security, Sahel, Insect Disturbances, GEM

Présentation

Version intégrale

Page publiée le 1er janvier 2022