Geographia Technica, Vol 21, Issue 2, 2026, pp. 105-118

SPATIAL INTERPOLATION OF RAINFALL AT LAKE TOBA REGION WITH ORDINARY AND EXTERNAL DRIFT KRIGING

Nicholas ABISHA , Sri NURDIATI , Endar Hasafah NUGRAHANI , Mohamad Khoirun NAJIB , Syukri Arif RAFHIDA

DOI: 10.21163/GT_2026.213.07

ABSTRACT: This study aims to interpolate rainfall in the Lake Toba region using ordinary kriging (OK) and external drift kriging (EDK) with elevation and distance-to-lake as drift variables. Rainfall data from 34 stations (2007-2017) were evaluated across three scenarios: (1) without imputation, (2) sparse data using only complete stations, and (3) all stations using k-nearest-neighbors imputation. Leave-one-out cross-validation revealed that in dense-data scenarios (1 and 3), OK and EDK produced similar accuracy, indicating that dense local observations override the need for auxiliary variables. However, in the sparse-data scenario (2), EDK with distance-to-lake significantly outperformed both OK and EDK with elevation. Geographically, this highlights that standard assumptions of orographic lift fail in the caldera due to complex rain shadows over Samosir Island. Instead, distance-to-lake serves as a reliable proxy for the localized lake-land breeze circulations that drive the basin's microclimate. Ultimately, this study demonstrates that spatial interpolation in complex, data-sparse regions must be grounded in variables that accurately reflect physical hydrometeorological realities.


Keywords: Rainfall, Kriging, Lake Toba, Elevation, Distance-to-Lake.

Full article here