Premium
A univariate versus a multivariate parameter distribution in a stochastic‐conceptual analysis of unsaturated flow
Author(s) -
Russo David,
Bresler Eshel
Publication year - 1982
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/wr018i003p00483
Subject(s) - hydraulic conductivity , scaling , surface runoff , infiltration (hvac) , groundwater recharge , mathematics , ponding , multivariate statistics , stochastic modelling , hydraulic head , flow (mathematics) , water flow , soil science , mechanics , hydrology (agriculture) , statistics , environmental science , thermodynamics , soil water , geotechnical engineering , geology , physics , groundwater , geometry , drainage , ecology , biology , aquifer
Solutions of stochastic‐conceptual flow problems obtained by utilizing a single stochastic parameter, the scaling factor α, are compared with solutions obtained by using multivariate parameter distribution. Two cases of one‐dimensional vertical flow are analyzed: (1) piston flow of solute under steady uniform surface water recharge and (2) transient water flow when uniform recharge rate is applied on the surface. Under steady vertical infiltration, expressing the variabilty of the hydraulic conductivity function K(θ ) by the single stochastic‐variate α gives essentially the same results as the case when the variability in K(θ ) is expressed by three stochastic variates. The agreement between the simulated average solute concentration profiles obtained by the two methods is improved as the recharge rate increases and a larger portion of the field is under ponding conditions. Similar results are found for the simulated potential runoff. For the case of the transient flow of water the variability in K(θ ), as well as the variability in the soil water retentivity function h(θ ), is expressed by a scaling factor α obtained from hydraulic conductivity data. Using α as the only stochastic variate resulted in mean water content (θ) profiles which are in a relatively good agreement with the mean θ profiles obtained when expressing the variability of K(θ ) and h(θ ) by five stochastic variates. The agreement between the distributions of the θ profiles obtained by the two methods, is rather poor. This agreement, as well as the agreement between the distributions of the depth of the wetting fronts, calculated by the two methods, are only slightly improved by using α as water content dependent stochastic variate. The results suggest that the scaling factor α, as a single stochastic variate, can be successfully used for stochastic analysis of water and solute flows under steady and transient infiltration conditions.