|
Surface approximation of Point Data using different Interpolation Techniques – A GIS approach
4.0 Results and Discussions:
Overlaying functionality of GIS was used for comparison of results obtained from Spline, IDW and TIN techniques. 4.1 IDW Vs Spline IDW estimates grid cell values by assigning the values of sample data points in the vicinity of the cell. The closer a point is to the center of cell being estimated, the more influence or weight it has in the averaging process. On the other hand, Spline estimates grid cell values by fitting a minimum curvature surface to the sample data. The results from the two interpolation techniques are different even for the study area, which is very small. Considering the example of surface creation for GW depth for post monsoon 2000, both methods produce good estimates and classify the area into four predefined classes (<2m, 2-3m, 3-5m and >5m). However, percentage occurrence of each class in the study area was found to be different. GW classes <2m, 2-3m, 3-5m and >5m were found to be 8.4%, 22%, 60.8% and 8.8% respectively using IDW whereas they were 9.5%, 25.5%, 52% and 13% respectively by spline method (Fig-4). Results from two techniques suggest that neither estimation process could be generalized for a particular application. Thus, the choice of proper interpolation technique is highly area specific. Again spline suits to those areas where there are little variations in the data within a short horizontal distance. IDW on the other hand suits for such kind of data, in which the value has a local influence that diminishes with distance. Since there are little variations in the GW value in the entire area and with the prior knowledge of the physiography of the area, which is nearly level to gently sloping, the results obtained from spline interpolation is found to be more closer to the expected value. Therefore spline interpolation technique was found more appropriate for our application.  Figure 4 Class view GW depth variation from Spline and IDW Interpolation Techniques 4.2 Grid Vs TIN Since TIN is a triangulation method it excludes the area outside the point location extent whereas grid assumes entire rectangular area displayed. It means there is no extrapolation in case of TIN. Therefore to use TIN, in addition to well-distributed sampling locations in the area, appropriate number of points should also lie on the periphery or even outside the study area. Grid model is simple and processes on theme tend to be more efficient than TIN, which is two-step process, first generation of TIN and second conversion of TIN into grid. Results from TIN are nearly similar to that of spline interpolation as compared to that of IDW in the study area. Moreover the rigid mesh structure of grid does not adapt to the variability of terrain (losing information between mesh points), source data may not be captured and reflected properly in resulting analysis like interpolation. However, if there is slight variation in the terrain information grid (spline) analysis is best-suited interpolation technique. The resolution of TIN can vary, that is, they can be more detailed in the areas where the surface is more complex and less detailed in the areas where the surface is simpler. The coordinates of the source data are maintained as part of the triangulation so subsequent analysis like interpolation results in no information loss. Linear feature like roads and streams can be represented by enforcing them in the model as triangle edges. Grids on the other hand being used if source data’s positioned accuracy is not very high or there is no need of using linear feature like roads and streams exactly. On the other hand, if the source data is very accurate and one would like to maintain the accuracy or to represent linear features TIN can be used. Since the study area has very little variation in topography and very high positional accuracy is not required in the GW study the surface approximation based on grid analysis was carried out for further processing for pre and post seasons from 1995-2000. The resulting surfaces from grid were used for weighted analysis giving overall situation of the area in a span of six years. 4.3 Weighted Analysis Since the project deals with vast amount of data to be processed, a model was designed to use results from various analysis to get a cumulative result that facilitates the dynamism within the change at the same point of time. Surfaces created for Depth to GW and other GW quality parameters such as EC, pH and SAR in GW in both pre and post monsoon periods from 1995 to 2000 were used in model. The idea was to get a scenario that represents the overall situation of the area in context of above parameters in a span of six years. GIS have not only made it possible to process, analyze and combine such type of data but they have also made easy to organize and integrate spatial data into larger system to accomplish various correlation studies. Surfaces created for Depth to GW in each year were used as input theme for the model. The input data were reclassified which replaces the input grid values with another set of values depending upon the ranges fixed for ground water heights. In this study, four ranges of GW heights were set which are < 2m, 2-3m, 3-5m and >5m. When the grid values are reclassified by range (four ranges here) each new value in output grid is assigned a range of numbers based on the values in input grid. The reclassification step was done for all the six years data in order to get homogeneity in the input values. A weighted overlay function was performed over the reclassified values to create a single output that combines all six years data into a single value depending upon the weightage assigned to each theme. This is based on the fact that usually the theme or factors in the input values are not equally important. Thus one has to prioritize the themes depending upon their importance by assigning weightage to them. The weightage function overlay lets all those issues into considerations. It again reclassifies the values in the input grid theme onto a common evaluation scale of suitability. The input grids are weighted by importance and added to produce an output grid. In this study, all the six-year data have given equal importance (weightage). This implies that 16.5 percent of influence for each year data over the derived output to make the total influence for all the six year themes equal to 100 percent. The resulting cell values are added to produce the output grid, which represent the overall trend of Depth to GW from 1995-2000. |