Fontsize: [-] Text [+]

Processor to determin landcover changes developed

These tools are meant to support change detection analysis for e.g. landcover classifications from remote sensing data. They are based on the comparison of a pair of discrete raster images as e.g. coded landcover classifications from different years. They provide statistical and map information allowing the quantitative, qualitative and spatial evaluation and interpretation of changes that occured between the two input dates.


 


The processes have been implemented as Web Processing Services (WPS). The OGC Web Processing Service (WPS) is designed to standardize the way that GIS calculations are made available to the Internet, thus the executor of processes does not require any special software for data processing but simply executing a HTTP request whereby the data inputs can be web-accessible URLs or embedded in the request.


 


The developed change detection tools allow for the calculation of statistics describing overall change rate, class specific confusion matrix of changes rates and areas or class specific histograms for 2 discrete raster images.


 


The subsequent figure illustrates the class specific histograms for Landsat based landcover classifications over the Mekong Delta in 1989 and 2002.


 


Histogram


 


Furthermore, processes have been implemented that calculate images of occured changes. On the one hand the result can represent all changes that occured in the area as shown in the next figure. Whereby the white colour represents areas which have not changed and the coloured regions represent those areas where changes occured. The colour is representing the respective class the change was targeting to. E.g. red colour (class 31) is indicating a strong urban grow between 1989 and 2002 in Can Tho city. The bright green (class 20) indicates that lots of areas have been converted to tree plantations mostly fruit trees.


 


All Changes


 


The confusion matrix allready helps to understand which class changed to a new class and what is the effected area and percentage. In order to understand where these effected regions are situated another process claculates class specific images showing the changes from one specific class to new classes. The followong image e.g. shows the changes that occured for areas that have been class 21 (rice, herbaceous) in 1989. Obviously some chenges occur due to seasonalty reasons in rice cropping. The most dominant chancges however are the conversion to tree cover (light green) and the urbanisation (red).


 


21

Copyright © 2014 German Aerospace Center (DLR). All rights reserved.