ISOTROPIC SEMIVARIOGRAM MODELING USING MAXIMUM LIKELIHOOD METHOD AND COKRIGING INTERPOLATION (CASE STUDY: PEATLAND GROUND WATER LEVEL DATA IN KALIMANTAN ISLAND)
Indonesia is a country with the second largest peatland area in the world, reaching 13.43 million hectares. Kalimantan is one of the three major islands in Indonesia which has a peatland area of 4.5 million hectares. Peat is soil resulting from the fertilization of organic matter through the product...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/73000 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Indonesia is a country with the second largest peatland area in the world, reaching 13.43 million hectares. Kalimantan is one of the three major islands in Indonesia which has a peatland area of 4.5 million hectares. Peat is soil resulting from the fertilization of organic matter through the production of tropical rainforest biomass. Peat data that represents groundwater level, soil moisture, and rainfall is useful in analyzing the physical properties of peat soils such as water content, bulk density, and bearing capacity. The peat data can be analyzed spatially using geostatistics. The isotropic semivariogram is one of the statistics that can be used in modeling spatial relationships between locations that are affected by a certain distance. This Final Project research aims to determine the distribution of experimental semivariograms and the best isotropic semivariogram models for peat data on Kalimantan and to apply these models to the cokriging interpolation of the primary variable Ground Water Level (TMA) and secondary variable Rainfall (CH). The experimental semivariogram calculation uses the Matheron and Cressie-Hawkins approach which will then be modeled with the exponential, Gaussian, and Spherical theoretical semivariogram models. The parameter estimation method used is Maximum Likelihood. The distribution obtained based on the simulation results is the Log-normal for both variables using the Matheron and Cressie-Hawkins approaches and the cross semivariogram. The best model for each variable with both approaches is the Spherical with MSE 0.38108 for TMA Matheron, 0.77837 for TMA Cressie-Hawkins, 0.71836 for CH Matheron, and 1.07150 for CH Cressie-Hawkins. The model is used to estimate the location of unobserved points using cokriging, which produces values close to the actual data in the model equation using the Cressie-Hawkins approach. |
|---|