|
||||||||||||
|
Further Data Analysis using the Simple Linear Least Squares Regression | |||||||||||
As the SRTM DEM for world-wide coverage is at a maximum resolution of only 3 arc-secs (approx. 90m), some resampling was required to increase the accuracy and resolution of the DEM. |
|
Yi : Dependent variable or regressand to be explained in the regression model Click here for more information on Simple Least Squares Regression. Click here to find out more about R Square.
Before carrying out our analysis of sea level change on the Singapore’s coastline, we considered the error distribution of our resampled CGIAR DEM. However, for the elevation data, from 0 - 15 m (in terms of Spot Height) the errors are distinctly one-directional (view error distribution chart). Hence, we calibrated DEM by adding the mean error from DEM cell values which fulfil the same criteria as the sample points. The calibrated data showed distinct improvements in correlation, representation and error rate. HERE IS THE ERROR COMPARISON AND REGRESSION ANALYSIS OF THE CALIBRATED DATA.
|
||||||||||
|
||||||||||||
Testing the Accuracy of our Resampled DEM | ||||||||||||
Results show that: (i) resampled data consistently performed better than original data; and, (ii) CGIAR (Resampled) DEM had the highest correlation coefficient of .988 (p<0.0005)
|