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

Accueil du site → Doctorat → Chine → 2006 → Study on Retrieval of Reed LAI and Scaling Effect in Occurrence Area of Oriental Migratory Locust Based on Remote Sensing Technology

Nanjing Normal University (2006)

Study on Retrieval of Reed LAI and Scaling Effect in Occurrence Area of Oriental Migratory Locust Based on Remote Sensing Technology

Chen Jian

Titre : Study on Retrieval of Reed LAI and Scaling Effect in Occurrence Area of Oriental Migratory Locust Based on Remote Sensing Technology

Auteur : Chen Jian

Grade : Doctoral Dissertation 2006

Université : Nanjing Normal University

Résumé partiel
The aim of this paper is to build up LAI retrieval model which suited for reed canopy and then research the relations between reed LAI and oriental migratory locust The radiation transfer model-FCR model and BP neural network model are built up in this paper with the aid of remote sensing technology and in situ data in study area Nandagang marsh located in Huanghua city of Hebei province, and then the LAI retrival methods are studied. The scaling effect and the spatial heterogeneity are also discussed using semivariogram in the paper.On the basis of LAI obtained from remote sensing data, the oriental migratory locust occurrence are monitored through reed LAI in the end.LAI retrival model based on TM and MODIS data are built up respectively. Whether based on TM or on MODIS data, the effect of LAI retrival through neural network model is better than that of regression model. The background spectral difference can be diminished because of FCR model.The paper draws out the following conclusions through the above studies : the relations of reed LAI and remote sensing can not express through linearity.Heterogeneity are exist in MODIS pixels which have 1km spatial resolution, so little relations exist in reed and MODIS data. The LAI precision obtained from low resolution remote sensing data can improve using neural network model. It was found that the nonlinearity of the algorithm for NDVI brings only little error to the scaling transformation of LAI, however, the spatial heterogeneity of LAI is the main renson giving rise to the scaling effect. The study indicates that in the study area the valid spatial scale of reeds is about 360m and the spatial self-correlation will disappear if exceeding this spatial distance. The site where locust occurring can be distinguished by the variation of NDVI, but that is easy to make misjudge and unsuitable for monitoring the extent of the locust occurrence. On the basis of retrieving LAI exactly, it was found that the range and extent of locust occurrence can be find out more exactly according to the variation between two or more LAI remote sensing data.

Mots clés : leaf area index (LAI) oriental migratory locust reed marsh remote sensing scaling effect

Présentation (Dissertationtopic)

Page publiée le 7 mai 2013, mise à jour le 24 avril 2018