Space-time modeling in rock layers through resistivity data (case study: Resistivity geoelectric data in Pontianak city)

West Kalimantan has the largest peat-land in Kalimantan sland after Central Kalimantan, which is 29.99%. Peat soils in West Kalimantan classified as low-maturity soils; therefore, it is still highly weathered. These could affect the infrastructure sector, especially in planting concrete piles for a building. The planted concrete stakes must reach a layer of soil/rock could be done by drilling, measuring log data, density data, or resistivity data. The data obtained requires considerable time, weather, cost, and energy factors. In addition, reaching a certain depth will take more time, effort, and cost as well. One way to make these factors more efficient is to perform predictive modeling for unmeasured/observable depths. In this article, stochastic modeling is carried out to predict the depth of soil/rock layers based on resistivity data in the West Kalimantan region and interpreted in the description of rock layers. The results obtained that the depth of the peat layer in Pontianak city is up to a depth of 4-5 meters measured from above sea level. This is number obtained from data estimation using the generalized space-time autoregressive (GSTAR) model with kernel Gaussian function. The random variable is represented by the resistivity value, while the time parameter index used is the depth of the soil layer. The results of the root mean square error (RMSE) obtained for locations 1, 2, and 3, respectively, are 12.87, 6.48, and 5.34. This RMSE value is relatively small based on the in-sample data used, it means the model is performing well. With the least-squares method, the parameter estimation is significant and stationary. However, in the residual test, an error is obtained which still has a time correlation.