Hydrologic consistency as a basis for assessing complexity of monthly water balance models for the continental United States

Guillermo F. Martinez, Hoshin Vijai Gupta

Research output: Contribution to journalArticle

29 Citations (Scopus)

Abstract

Methods to select parsimonious and hydrologically consistent model structures are useful for evaluating dominance of hydrologic processes and representativeness of data. While information criteria (appropriately constrained to obey underlying statistical assumptions) can provide a basis for evaluating appropriate model complexity, it is not sufficient to rely upon the principle of maximum likelihood (ML) alone. We suggest that one must also call upon a "principle of hydrologic consistency," meaning that selected ML structures and parameter estimates must be constrained (as well as possible) to reproduce desired hydrological characteristics of the processes under investigation. This argument is demonstrated in the context of evaluating the suitability of candidate model structures for lumped water balance modeling across the continental United States, using data from 307 snow-free catchments. The models are constrained to satisfy several tests of hydrologic consistency, a flow space transformation is used to ensure better consistency with underlying statistical assumptions, and information criteria are used to evaluate model complexity relative to the data. The results clearly demonstrate that the principle of consistency provides a sensible basis for guiding selection of model structures and indicate strong spatial persistence of certain model structures across the continental United States. Further work to untangle reasons for model structure predominance can help to relate conceptual model structures to physical characteristics of the catchments, facilitating the task of prediction in ungaged basins.

Original languageEnglish (US)
Article numberW12540
JournalWater Resources Research
Volume47
Issue number12
DOIs
StatePublished - 2011

Fingerprint

water budget
catchment
persistence
snow
prediction
basin
modeling

ASJC Scopus subject areas

  • Water Science and Technology

Cite this

Hydrologic consistency as a basis for assessing complexity of monthly water balance models for the continental United States. / Martinez, Guillermo F.; Gupta, Hoshin Vijai.

In: Water Resources Research, Vol. 47, No. 12, W12540, 2011.

Research output: Contribution to journalArticle

@article{895c817039a14d548041074c3c46d894,
title = "Hydrologic consistency as a basis for assessing complexity of monthly water balance models for the continental United States",
abstract = "Methods to select parsimonious and hydrologically consistent model structures are useful for evaluating dominance of hydrologic processes and representativeness of data. While information criteria (appropriately constrained to obey underlying statistical assumptions) can provide a basis for evaluating appropriate model complexity, it is not sufficient to rely upon the principle of maximum likelihood (ML) alone. We suggest that one must also call upon a {"}principle of hydrologic consistency,{"} meaning that selected ML structures and parameter estimates must be constrained (as well as possible) to reproduce desired hydrological characteristics of the processes under investigation. This argument is demonstrated in the context of evaluating the suitability of candidate model structures for lumped water balance modeling across the continental United States, using data from 307 snow-free catchments. The models are constrained to satisfy several tests of hydrologic consistency, a flow space transformation is used to ensure better consistency with underlying statistical assumptions, and information criteria are used to evaluate model complexity relative to the data. The results clearly demonstrate that the principle of consistency provides a sensible basis for guiding selection of model structures and indicate strong spatial persistence of certain model structures across the continental United States. Further work to untangle reasons for model structure predominance can help to relate conceptual model structures to physical characteristics of the catchments, facilitating the task of prediction in ungaged basins.",
author = "Martinez, {Guillermo F.} and Gupta, {Hoshin Vijai}",
year = "2011",
doi = "10.1029/2011WR011229",
language = "English (US)",
volume = "47",
journal = "Water Resources Research",
issn = "0043-1397",
publisher = "American Geophysical Union",
number = "12",

}

TY - JOUR

T1 - Hydrologic consistency as a basis for assessing complexity of monthly water balance models for the continental United States

AU - Martinez, Guillermo F.

AU - Gupta, Hoshin Vijai

PY - 2011

Y1 - 2011

N2 - Methods to select parsimonious and hydrologically consistent model structures are useful for evaluating dominance of hydrologic processes and representativeness of data. While information criteria (appropriately constrained to obey underlying statistical assumptions) can provide a basis for evaluating appropriate model complexity, it is not sufficient to rely upon the principle of maximum likelihood (ML) alone. We suggest that one must also call upon a "principle of hydrologic consistency," meaning that selected ML structures and parameter estimates must be constrained (as well as possible) to reproduce desired hydrological characteristics of the processes under investigation. This argument is demonstrated in the context of evaluating the suitability of candidate model structures for lumped water balance modeling across the continental United States, using data from 307 snow-free catchments. The models are constrained to satisfy several tests of hydrologic consistency, a flow space transformation is used to ensure better consistency with underlying statistical assumptions, and information criteria are used to evaluate model complexity relative to the data. The results clearly demonstrate that the principle of consistency provides a sensible basis for guiding selection of model structures and indicate strong spatial persistence of certain model structures across the continental United States. Further work to untangle reasons for model structure predominance can help to relate conceptual model structures to physical characteristics of the catchments, facilitating the task of prediction in ungaged basins.

AB - Methods to select parsimonious and hydrologically consistent model structures are useful for evaluating dominance of hydrologic processes and representativeness of data. While information criteria (appropriately constrained to obey underlying statistical assumptions) can provide a basis for evaluating appropriate model complexity, it is not sufficient to rely upon the principle of maximum likelihood (ML) alone. We suggest that one must also call upon a "principle of hydrologic consistency," meaning that selected ML structures and parameter estimates must be constrained (as well as possible) to reproduce desired hydrological characteristics of the processes under investigation. This argument is demonstrated in the context of evaluating the suitability of candidate model structures for lumped water balance modeling across the continental United States, using data from 307 snow-free catchments. The models are constrained to satisfy several tests of hydrologic consistency, a flow space transformation is used to ensure better consistency with underlying statistical assumptions, and information criteria are used to evaluate model complexity relative to the data. The results clearly demonstrate that the principle of consistency provides a sensible basis for guiding selection of model structures and indicate strong spatial persistence of certain model structures across the continental United States. Further work to untangle reasons for model structure predominance can help to relate conceptual model structures to physical characteristics of the catchments, facilitating the task of prediction in ungaged basins.

UR - http://www.scopus.com/inward/record.url?scp=84855488459&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84855488459&partnerID=8YFLogxK

U2 - 10.1029/2011WR011229

DO - 10.1029/2011WR011229

M3 - Article

AN - SCOPUS:84855488459

VL - 47

JO - Water Resources Research

JF - Water Resources Research

SN - 0043-1397

IS - 12

M1 - W12540

ER -