| Abstract |
: |
It is certainly necessary for a practical, applicative and accurate method to predict the missing data. IDW (Inverse Distance wighting) method is one of the methods and it is quite well known among practitioners. This method is used to predict the missing data from several measured data (multipoint interpolation). In predicting the missing data, it will be better to use the closest data, than using the far data. Therefore, it is necessary for accuracy improvement of the IDW method by bringing data into closer points to the concerned data points, namely the points located between two data and having perpendicular position to the addressed data or called as Perpendicular Line. To know the reliability of IDW Perpendicular method, so it is necessary to test its ability to result the value in the addressed point rather than by the measured rain data as the exact function. The exact function is a surface value (z) as the function from x and y coordinates. The testing has done by using an exact surface value using 3,4, and 5 data. The calculation results of MAD, MSE, MPE show that IDW-P method (1,43%) has much smaller error than the original IDW (10,18 %) while the determination coefficient for IDW-P method is 0.997 which is greater than the original IDW method where is 0.901. In the application process in curved area, IDW-P method has average error percentage of 1,35 %, while the original IDW method has 6,02 %. In the application process using surface data, the IDW-P method has average error percentage of 6,11 %, while the original IDW method has 7,44 %, so that have improvement 1,33%. Generally, the multipoint interpolation of IDW-P method is better than the original IDW method. |
