Modeling Uncertainty of Digital Elevation Models SRTM and ASTER and Their Impacts on Landform Classification in Garm-Chay Basin

Digital Elevation Models (DEMs) is one of the main geographical data models which form the basis of the different spatial analysis. DEM is known as fundamental data in for many modeling tasks. Nowadays, the results validation of GIS spatial analyzes, has become a major challenge in the world of GIS .The quality of a DEMis dependent upon a number of interrelated factors, including the methods of data acquisition, the nature of the input data, and the methods employed in generating the DEMs.Analysis of uncertainty in different fields, due to data quality and related issues such as error, uncertainty models, errors propagation, errors elimination and uncertainties in the data, are felt more than any other times. Of all these factors, data acquisition is the most critical one. Previous studies on DEM data acquisition have focused either on examination of generation method(s), or on case studies of accuracy testing. These studies are not adequate, however, for the purpose of understanding uncertainty (an indicator used to approximate the discrepancy between geographic data and the geographic reality that these data intend to represent) associated with DEM data and the propagation of this uncertainty through GIS based analyses. The development of strategies for identifying, quantifying, tracking, reducing, visualizing, and reporting uncertainty in DEM data are called for by the GIS community.
In order to apply uncertainty analysis on DEMs this study aimed to evaluate the error rate and uncertainty of elevation data obtained from SRTM and ASTER satellites. The objectives of this study are: (1) to understand the sources and reasons for uncertainty in DEMs produced by cartographic digitizing; (2) to develop methods for quantifying the uncertainty of DEMs using distributional measures and (3) to measure the uncertainty associated with DEMs and minimize the chances of error by manse of optimizing models. Quantifying uncertainty in DEMs requires comparison of the original elevations (e.g. elevations read from topographic maps) with the elevations in a DEM surface. Such a comparison results in height differences (or residuals) at the tested points to analysis the pattern of deviation between two sets of elevation data, conventional ways are to yield statistical expressions of the accuracy, such as the root mean square error, standard deviation, and mean. In fact, all statistical measures that are effective for describing a frequency distribution, including central tendency and dispersion measures, may be used, as long as various assumptions for specific methods are satisfied. Our research methodology includes several steps. The first step causing the statistical indices ME, STD and RMSE, the error rate of DTMs ​​ for obtaining the chances of error in ach model. It has to be mentioned that the main attraction of the RMSE lies in its easy computation and straightforward concept. However, this index is essentially a single global measure of deviations, thus incapable of accounting for spatial variation of errors over the interpolated surface. In order to obtain more accurate results, then uncertainty of data errors was also simulated by Monte Carlo method and error propagation pattern was extracted by interpolation of results.
The results of this step show that, the DEM derived from pair stereo ASTER despite having better spatial resolution, included more errors and practically lacking the details of DTM 30 meters. Finally, removing the error propagation pattern from DEMs, the secondary DEM was produced. By recalculating indicators describing the error and comparing these values with the initial values, the results indicate that, both DEMs show more accuracy after eliminating the error propagation pattern. TPI Index was used to determine the location of basin topography and the basin is divided into 6 classes and error rate in each class was calculated before and after the simulation. The results showed that, the error rates in all classes before and after the simulation in both DEMs were reduced. In terms of uncertainty analysis methods for DEMs, results of our research indicated that the RMSE methods alone is not sufficient for quantifying DEM uncertainty, because this measure rarely addresses the issue of distributional accuracy. To fully understand and quantify the DEM uncertainty, spatial accuracy measures, such accuracy surfaces, indices for spatial autocorrelation, and variograms, should be used results also indicated that Monet Carlo simulation is indeed sufficient methods for simulation error in DEMs. Results of this research are great of important for uncertainty analysis in domain of Geosciences and can be used for improving the accuracy of modeling in variety of applications.