Geo-spatial model validation in the context of low-land conversion due to households emerging in the Colombo sub-urban

Abstract

This study examines how individual households have been emerging in the low lying areas during the period 2005-2012 in the core study area of Colombo Metropolitan Region ( CMR) and also the process through which they have gradually established themselves as either stable or unstable households. Mass manipulation of geo-spatial factors in innumerable land plots has inevitably led to increasing negative environmental effects in the region. Hence, an attempt is made to build a geo-spatial model that can be used as a guide and index to help understand how the unending process of individual households emerging in the CMR. The primary focus of the study is comparison of main model data with validation model; hence, validation procedure of this study has been explained and compared with the main model. The typical individual household plot has been chosen as the unit of analysis. Information from 408 households was collected from the core study area to build the main model and perform validation and then the data was tested with a spatial logistic regression model. The main model indicated an accuracy of about 92.2% together with high significance levels for 8 variables out of the total 19 variables namely Household income (HI), Ground water surface level (GWS), Public participatory practice (PPP),Permanent plants growing in plot (PPG), Rain water remaining in the plot (RWR), Skilled jobs (SKJ), Technical skills and adaptation (TSA) and Low lying related plants availability (LLP). Predicted probability value of each housing plot mapped with GIS can be seen with the spatial distribution displayed clearly. Accuracy of the model validation process is 85.09% that indicates compatibly well. Based on predicted probability value of each land plot, both models have been run together with field data from the geo-spatial information system