Spatial dispersion analysis of spring flood peak discharge 1% probability of the Lithuanian rivers

Abstract

The aim of the research is to determine whether the geostatistical modeling is suitable to analyze the spatial dispersion of the selected hydrological characteristic as well as to find out the level of accuracy of predicting the values of unexplored Lithuanian rivers using this method. The research analyzed 55 rivers and used values of spring flood peak discharge 1% probability (A1%) collected in 74 water measurement stations (WMS) located on these rivers. The best results of interpolation of the spring flood peak discharge 1% probability were obtained using Gaussian variogram model and Ordinary Kriging method. With this method, values of the spring flood peak discharge 1% probability for the unexplored rivers can be predicted with an error of 11.5%. The analysis of modeled values of the spring flood peak discharge 1% probability revealed that the most significant errors (>48%) were obtained with seventeen basins and 11 of them have area of up to 200 km2 (values were twice as high as the analyzed ones). the greatest errors were obtained with 17. Moreover, it was noticed that significant errors are obtained when basins with highly different areas occur next to each other in the data set. Grouping of the analyzed water measurement stations by basin area into three groups was not effective, because errors remained high (15%, 4%, 7%) and counters occurred in the spatial dispersion maps. Following the reduction of the geographic parameter values with the basin area of 200 km2 and after the geostatistical analysis of these values, it can be argued that significant errors remained with basins that cover small area; therefore, it is necessary to perform a more detailed analysis of basins where greater errors of the hydrological characteristic have been obtained.