Rain Gauge Network Optimization in Brunei-Muara District Using Historical Data

Rainfall data is imperative for agricultural productivity and managing adverse rainfall events. To capture an accurate representation of rainfall data, an optimal rain gauge network is required. Analysis of historical rainfall data from the Brunei-Muara district is performed to provide insights on how to improve the existing rain gauge network. To achieve this, geospatial factors affecting rainfall similarity between locations are identified. These geospatial factors provide the basis for quantifying the quality of placement of rain gauges in the existing network as well as the identification of potential locations for additional rain gauge nodes. Kriging error is explored to identify potential regions to add additional rain gauge nodes. Spatial interpolation methods in the form of ordinary kriging and inverse distance weighting are utilized to identify redundant gauges and evaluate the placement of rain gauge nodes. The mean absolute error of the interpolated values allows for a metric to assess the performance of a rain gauge network. Results indicate a positive correlation between pairwise geodesic distance and rainfall similarity. By combining kriging error uncertainty and MAE, a more comprehensive assessment of the rain gauge network is possible by accounting for local prediction uncertainty and global prediction accuracy.