Uncertainties on the GIS based potential natural vegetation simulation using Comprehensive and Sequential Classification System

Abstract

ABSTRACT The use of GIS-based ecological models is increasing and the accuracy of input datasets of these models is improving. Still, there is a significant gap in quantifying the uncertainty related to the input data and the accuracy of these models’outputs. This study quantified error in annual cumulative temperature derived from using daily mean temperature and monthly mean temperature, and the uncertainty and error propagated in the Comprehensive and Sequential Classification System (CSCS) model that predicts Potential Natural Vegetation (PNV) in China. The error in annual cumulative temperature derived from daily mean temperature and monthly mean temperature is particularly high in Northwest and Northern China. The deviations in annual cumulative temperature have different effects on each PNV including edge effects on the model’s predictability due to error propagation in the interpolation method and overlay analysis. Future research can focus on the assessment of model behavior with the uncertainty of data itself and different spatial analysis methods including the spatial resolution of datasets. There is a need to develop a unique algorithm that would enable a better assessment of attribute error and location error in the spatial modeling.