论文部分内容阅读
In hydrology, Digital Elevation Model (DEM) plays an essential role, from understanding the hydrological characteristics of a watershed to setting up a hydrological model.Access to high resolution DEM, however, is costly and often constraint by data sharing policy.Although the Shuttle Radar Topography Mission (SRTM) is a publicly accessible Digital Elevation Model (DEM) at no cost, its vertical accuraciesat forested and highly urbanized area are known to be rather poor.These low accuracies are due to the 5.6cm wavelength used by SRTM that does not penetrate vegetation well, and the averaging errors of the inherited coarse resolution especially at areas where scattered elevation variation is known, e.g.urban area with various height of buildings and ground/road.This paper considers both forest and urban area of Singapore as a proof of concept of an approach to improve the SRTM dataset.The approach makes full use of (1) the introduction of multi-spectral imagery (Landsat 8), amended into SRTM data; (2) the Artificial Neural Network (ANN) to perform non-linear pattern recognition; and (3) a high accuracy DEM (1m vertical accuracy) as the reference to train the ANN.The study shows a series of significant improvements of the SRTM when assessed with reference DEM, such as the RMSE reduction of 52# to 68# and the increased visibility of the elevation patterns.The approach has also been applied to other cities, e.g.Surabaya and Bandung (Indonesia), to demonstrate its robustness.Such promising approach implies significant reduction of cost in obtaining reliable topography required for varieties of hydrological applications, in contrast with available commercial DEM.This is because the amount of extensive survey and alternatively high accuracy remote sensing imagery required to derive such DEM is reduced to the minimum; in this case only a portion of representative reference DEM to train the Artificial Neural Network is required.Such affordable DEM is useful for many budget strapped developing regions, as an essential input to develop flood related disaster management support system.The high accuracy DEM allows policy makers to better assess the vulnerability of coastal cities with rising sea level.