Study region: This study refers to the Hydrological Open Air Laboratory (HOAL) watershed, located in South-West Austria. Study focus: The spatial variability of saturated hydraulic conductivity, Ks, is sometimes synthetized by geometric mean, K̃s, and coefficient of variation, while areal-average infiltration models rely upon the arithmetic mean, K¯sl, associated to log-transformed Ks, and relative coefficient of variation, CVal. Robust estimation of K¯sl and CVal, as well as of K̃s and associated coefficient of variation, CVg,would require a large number of Ks observations. The determination of the minimum number of Ks measurements, n*, for obtaining sufficiently accurate values of each aforementioned quantity over an area is an open issue addressed here. A statistical approach has been applied to Ks datasets on three grassy plots for an uncertainty analysis based on the non-parametric bootstrap method with replacement. The uncertainty of each quantity has been derived for different observation numbers and areas. New hydrological insights for the region: Considering different sub-regions in the largest plot the uncertainty is almost invariant with increasing the sub-region area beyond a threshold. Furthermore, for a given n*, the uncertainty of K¯sl and CVal is much smaller than that of K̃s and CVg. Our approach defines a methodology for determining over an area the n* associated to a fixed uncertainty level in the joint estimation of the selected quantities. Guidelines for investigations over different plots are also proposed.
A statistical approach for the assessment of the saturated hydraulic conductivity applied to an Austrian region
Flammini A.
;Morbidelli R.;Corradini C.;Dari J.;Saltalippi C.;
2023
Abstract
Study region: This study refers to the Hydrological Open Air Laboratory (HOAL) watershed, located in South-West Austria. Study focus: The spatial variability of saturated hydraulic conductivity, Ks, is sometimes synthetized by geometric mean, K̃s, and coefficient of variation, while areal-average infiltration models rely upon the arithmetic mean, K¯sl, associated to log-transformed Ks, and relative coefficient of variation, CVal. Robust estimation of K¯sl and CVal, as well as of K̃s and associated coefficient of variation, CVg,would require a large number of Ks observations. The determination of the minimum number of Ks measurements, n*, for obtaining sufficiently accurate values of each aforementioned quantity over an area is an open issue addressed here. A statistical approach has been applied to Ks datasets on three grassy plots for an uncertainty analysis based on the non-parametric bootstrap method with replacement. The uncertainty of each quantity has been derived for different observation numbers and areas. New hydrological insights for the region: Considering different sub-regions in the largest plot the uncertainty is almost invariant with increasing the sub-region area beyond a threshold. Furthermore, for a given n*, the uncertainty of K¯sl and CVal is much smaller than that of K̃s and CVg. Our approach defines a methodology for determining over an area the n* associated to a fixed uncertainty level in the joint estimation of the selected quantities. Guidelines for investigations over different plots are also proposed.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.