FIGURE 6 Semi-variogram with strong spatial dependence during the period of August 2019 - February 2020
Table 4 shows the spatial dependence of water table data from November 2018 until July 2019 are classified as weak spatial dependence (>75%). A flat semi-variogram indicates weak dependence, with most range value is 0 m, and it shows no spatial correlation (Figure 5).
Moderate – strong spatial dependence is shown from water table data from August 2019 until February 2020, which indicates spatial correlation with the correlation range average within this period is 2258 m (Figure 6). From this point forward, the analysis will be separated into two groups data: weak spatial dependence (November 2018 – July 2019) and strong spatial dependence (August 2019 – February 2020).
Cross-validation for all data is done to see the accuracy of the Kriging map, based on the predicted and measured data (Table 5). For weak spatial dependence, the mean error is negative, which means that the predicted value is lower than the actual measurement. For strong spatial dependence, the mean error is positive, which means that the predicted value is higher than the actual measurement. But overall, all the mean error values are close to 0, which means the predictions are unbiased. Most of the root-mean-square standardized error values close to 1, which indicates the standard errors are normally distributed. RMSE value for weak spatial dependence data is higher than RMSE value for strong spatial dependence data. It demonstrates that weak spatial dependence has a higher variance compared to strong spatial dependence. Similar circumstances have occurred for average standard error, where the weak spatial dependence data has a higher value compared with the strong spatial dependence.
TABLE 5 Cross-validation results of Kriging map for the period of November 2018 - February 2020