Enhancing geostatistical precipitation estimations for the Santiago River basin, Mexico

Accurate precipitation estimation is crucial for understanding the hydrological cycle, its applications in basin-specific planning, and outliers event prediction. Multivariate geostatistics leverage correlated variables, such as terrain elevation and shoreline distance, to reduce estimation error uncertainty. However, the distinct characteristics of humid and dry seasons demand specific estimation approaches. Precise precipitation estimation poses a challenge in the vast and diverse Santiago River basin (SRB) along Mexico’s west coast. This study assessed precipitation estimates for dry and humid seasons using ordinary kriging and ordinary cokriging with altitude and shoreline distance as auxiliary variables. Evaluation of error metrics revealed superior results incorporating shoreline distance as a covariable in the wet month of July, especially after logarithmic transformation, yielding a 17% improvement in average standardized error compared to the univariate approach. Conversely, optimal results were achieved for the dry month (February) using ordinary kriging excluding outliers’ values, effectively reducing the average squared error.