TY - JOUR
T1 - Semidistributed hydrological modeling
T2 - A "saturation path" perspective on TOPMODEL and VIC
AU - Kavetski, Dmitri
AU - Kuczera, George
AU - Franks, Stewart W.
PY - 2003/1/1
Y1 - 2003/1/1
N2 - The conceptual foundation, behavior, and implementation of semidistributed hydrological models such as TOPMODEL and VIC are reviewed from a "saturation path modeling" perspective, hereinafter referred to as SPM. A general family of SPM models is analyzed, based on the assumptions that (1) an event-invariant relationship between the saturated area and catchment storage exists (which may include hysteretic and stochastic components) and (2) ponding is negligible and quick flow is generated from precipitation on saturated areas. The VIC and TOPMODEL are then obtained from specific constitutive functions and numerical approximations. In general, SPM leads to compact equations in canonical ODE form, making the behavior of the model transparent and facilitating improvements of both the conceptual model itself and its numerical implementation. A comprehensive numerical analysis of SPM, TOPMODEL, and VIC is presented, providing insight into the timescale dependence of their parameters. It is demonstrated that uncontrolled numerical implementations, particularly traditional fixed-step explicit Euler methods, can introduce considerable artifacts into the behavior of conceptual models such as TOPMODEL, corrupting calibration and prediction.
AB - The conceptual foundation, behavior, and implementation of semidistributed hydrological models such as TOPMODEL and VIC are reviewed from a "saturation path modeling" perspective, hereinafter referred to as SPM. A general family of SPM models is analyzed, based on the assumptions that (1) an event-invariant relationship between the saturated area and catchment storage exists (which may include hysteretic and stochastic components) and (2) ponding is negligible and quick flow is generated from precipitation on saturated areas. The VIC and TOPMODEL are then obtained from specific constitutive functions and numerical approximations. In general, SPM leads to compact equations in canonical ODE form, making the behavior of the model transparent and facilitating improvements of both the conceptual model itself and its numerical implementation. A comprehensive numerical analysis of SPM, TOPMODEL, and VIC is presented, providing insight into the timescale dependence of their parameters. It is demonstrated that uncontrolled numerical implementations, particularly traditional fixed-step explicit Euler methods, can introduce considerable artifacts into the behavior of conceptual models such as TOPMODEL, corrupting calibration and prediction.
KW - Distribution function modeling
KW - Numerical models
KW - Parameter distributions
KW - SPM
KW - TOPMODEL
KW - VIC
UR - http://www.scopus.com/inward/record.url?scp=1542442449&partnerID=8YFLogxK
U2 - 10.1029/2003WR002122
DO - 10.1029/2003WR002122
M3 - Review article
AN - SCOPUS:1542442449
SN - 0043-1397
VL - 39
JO - Water Resources Research
JF - Water Resources Research
IS - 9
ER -