The sensitivity of hydrologic processes across North America considering model structure and parametric uncertainty

2020 ◽  
Author(s):  
Juliane Mai ◽  
James Craig ◽  
Bryan Tolson ◽  
Richard Arsenault
Keyword(s):  
Sensitivity Analysis ◽  
North America ◽  
Rocky Mountains ◽  
Parametric Uncertainty ◽  
Weighted Sums ◽  
Model Structure ◽  
Snow Melt ◽  
Hydrologic Processes ◽  
South Eastern ◽  
Model Variability

<p>Information on the sensitivity of model parameters and model components such as processes are essential for model development, model improvement, and model calibration, amongst others.</p><p>In this work we apply the method of the extended Sobol’ Sensitivity analysis (xSSA) which not only considers parametric uncertainty but also fully incorporates structural uncertainties (Mai et al. (2019) WRR; under review). The results of such an analysis yield not only the traditional parameter sensitivities but also sensitivities of model process options (e.g., different snowmelt algorithms) and sensitivities of model processes (e.g., snowmelt, infiltration, baseflow). </p><p>The Raven hydrologic modelling framework (http://raven.uwaterloo.ca) allowing for flexible model structures is employed in this work. We used three options each for infiltration, quickflow, and snow melt as well as two options each for baseflow, and soil evaporation. Rather than considering 108 (3x3x3x2x2) discrete model setups, we used weighted sums of all process options yielding an infinite number of models tested.</p><p>The analysis is performed for 5797 basins across Canada (CANOPEX; Tarek et al. (2019) HESSD) and the US (USGS). The lumped basin setups use daily precipitation and  minimum/ maximum daily temperature. The sensitivity analysis is based on 20 years of daily streamflow simulations (1991-2010) after two years of spin-up (1989-1990). No observed streamflow is required for the analysis.</p><p>In total more than 450 million model runs were performed to determine sensitivities of parameters, process options and processes (51%, 35%, and 14% of model runs, respectively) across the almost 5800 basins. The computational demand was about 12 core years producing 23 TB of raw model outputs.</p><p>The analysis allows for unique, new insights into the importance of hydrologic processes and parameters (practically) independent of the model (structure) used. A few highlight results are: 1) Baseflow and other sub-surface processes are of low importance across North America- especially when time points of high flows are of interest. 2) Percolation, evaporation, and infiltration show very similar patterns with increased importance in South-eastern US and west of the Rocky Mountains. 3) Up to 30% of the overall model variability can be attributed to snow melt in regions that are snow dominated (Northern Canada and Rocky mountains). Potential melt shows a similar gradient as snow melt with sensitivities of above 60% in the Province of Quebec and the Rocky Mountains. 4) Direct runoff (quickflow) is the most sensitive of all hydrologic processes- especially in South-Eastern US it is responsible for more than 80% of the model variability.</p>

scholarly journals Simultaneously determining global sensitivities of model parameters and model structure

2020 ◽  
Vol 24 (12) ◽  
pp. 5835-5858
Author(s):  
Juliane Mai ◽  
James R. Craig ◽  
Bryan A. Tolson
Keyword(s):  
Sensitivity Analysis ◽  
Hydrological Modeling ◽  
Structural Model ◽  
Hydrologic Model ◽  
Model Parameters ◽  
Model Structure ◽  
Modeling Framework ◽  
Model Structure Uncertainty ◽  
Model Variability

Abstract. Model structure uncertainty is known to be one of the three main sources of hydrologic model uncertainty along with input and parameter uncertainty. Some recent hydrological modeling frameworks address model structure uncertainty by supporting multiple options for representing hydrological processes. It is, however, still unclear how best to analyze structural sensitivity using these frameworks. In this work, we apply the extended Sobol' sensitivity analysis (xSSA) method that operates on grouped parameters rather than individual parameters. The method can estimate not only traditional model parameter sensitivities but is also able to provide measures of the sensitivities of process options (e.g., linear vs. non-linear storage) and sensitivities of model processes (e.g., infiltration vs. baseflow) with respect to a model output. Key to the xSSA method's applicability to process option and process sensitivity is the novel introduction of process option weights in the Raven hydrological modeling framework. The method is applied to both artificial benchmark models and a watershed model built with the Raven framework. The results show that (1) the xSSA method provides sensitivity estimates consistent with those derived analytically for individual as well as grouped parameters linked to model structure. (2) The xSSA method with process weighting is computationally less expensive than the alternative aggregate sensitivity analysis approach performed for the exhaustive set of structural model configurations, with savings of 81.9 % for the benchmark model and 98.6 % for the watershed case study. (3) The xSSA method applied to the hydrologic case study analyzing simulated streamflow showed that model parameters adjusting forcing functions were responsible for 42.1 % of the overall model variability, while surface processes cause 38.5 % of the overall model variability in a mountainous catchment; such information may readily inform model calibration and uncertainty analysis. (4) The analysis of time-dependent process sensitivities regarding simulated streamflow is a helpful tool for understanding model internal dynamics over the course of the year.

scholarly journals Simultaneously Determining Global Sensitivities of Model Parameters and Model Structure

2020 ◽  
Author(s):  
Juliane Mai ◽  
James R. Craig ◽  
Bryan A. Tolson
Keyword(s):  
Sensitivity Analysis ◽  
Hydrological Modeling ◽  
Structural Model ◽  
Hydrologic Model ◽  
Model Parameters ◽  
Model Structure ◽  
Modeling Framework ◽  
Model Structure Uncertainty ◽  
Model Variability

Abstract. Model structure uncertainty is known to be one of the three main sources of hydrologic model uncertainty along with input and parameter uncertainty. Some recent hydrological modeling frameworks address model structure uncertainty by supporting multiple options for representing hydrological processes. It is, however, still unclear how best to analyze structural sensitivity using these frameworks. In this work, we apply an Extended Sobol' Sensitivity Analysis (xSSA) method that operates on grouped parameters rather than individual parameters. The method can estimate not only traditional model parameter sensitivities but is also able to provide measures of the sensitivities of process options (e.g., linear vs. non-linear storage) and sensitivities of model processes (e.g., infiltration vs. baseflow) with respect to a model output. Key to the xSSA method's applicability to process option and process sensitivity is the novel introduction of process option weights in the Raven hydrological modeling framework. The method is applied to both artificial benchmark models and a watershed model built with the Raven framework. The results show that: (1) The xSSA method provides sensitivity estimates consistent with those derived analytically for individual as well as grouped parameters linked to model structure. (2) The xSSA method with process weighting is computationally less expensive than the alternative aggregate sensitivity analysis approach performed for the exhaustive set of structural model configurations, with savings of 81.9 % for the benchmark model and 98.6 % for the watershed case study. (3) The xSSA method applied to the hydrologic case study analyzing simulated streamflow showed that model parameters adjusting forcing functions were responsible for 42.1 % of the overall model variability while surface processes cause 38.5 % of the overall model variability in a mountainous catchment; such information may readily inform model calibration. (4) The analysis of time dependent process sensitivities regarding simulated streamflow is a helpful tool to understand model internal dynamics over the course of the year.

Wildfires in the southern Canadian Rocky Mountains and their relationship to mid-tropospheric anomalies

1993 ◽  
Vol 23 (6) ◽  
pp. 1213-1222 ◽  
Author(s):  
E.A. Johnson ◽  
D.R. Wowchuk
Keyword(s):  
North America ◽  
Rocky Mountains ◽  
Large Scale ◽  
Fire Frequency ◽  
Fuel Moisture ◽  
Large Fire ◽  
Canadian Rocky Mountains ◽  
Area Burned ◽  
Upper Level ◽  
Moisture Conditions

In this paper we present evidence for a large-scale (synoptic-scale) meteorological mechanism controlling the fire frequency in the southern Canadian Rocky Mountains. This large-scale control may explain the similarity in average fire frequencies and timing of change in average fire frequencies for the southern Canadian Rocky Mountains. Over the last 86 years the size distribution of fires (annual area burned) in the southern Canadian Rockies was distinctly bimodal, with a separation between small- and large-fire years at approximately 10–25 ha annual area burned. During the last 35 years, large-fire years had significantly lower fuel moisture conditions and many mid-tropospheric surface-blocking events (high-pressure upper level ridges) during July and August (the period of greatest fire activity). Small-fire years in this period exhibited significantly higher fuel moisture conditions and fewer persistent mid-tropospheric surface-blocking events during July and August. Mid-tropospheric surface-blocking events during large-fire years were teleconnected (spatially and temporally correlated in 50 kPa heights) to upper level troughs in the North Pacific and eastern North America. This relationship takes the form of the positive mode of the Pacific North America pattern.

Climatic implications of latest Pleistocene and earliest Holocene mammalian sympatries in eastern Washington state, USA

2008 ◽  
Vol 70 (3) ◽  
pp. 426-432 ◽  
Author(s):  
R. Lee Lyman
Keyword(s):  
North America ◽  
Late Pleistocene ◽  
Rocky Mountains ◽  
Taxonomic Composition ◽  
Washington State ◽  
The West ◽  
Mammalian Faunas ◽  
Eastern Washington

AbstractFor more than fifty years it has been known that mammalian faunas of late-Pleistocene age are taxonomically unique and lack modern analogs. It has long been thought that nonanalog mammalian faunas are limited in North America to areas east of the Rocky Mountains and that late-Pleistocene mammalian faunas in the west were modern in taxonomic composition. A late-Pleistocene fauna from Marmes Rockshelter in southeastern Washington State has no modern analog and defines an area of maximum sympatry that indicates significantly cooler summers than are found in the area today. An earliest Holocene fauna from Marmes Rockshelter defines an area of maximum sympatry, including the site area, but contains a single tentatively identified taxon that may indicate slightly cooler than modern summers.

scholarly journals GCM Systematic Error Correction and Specification of the Seasonal Mean Pacific–North America Region Atmosphere from Global SSTs

1999 ◽  
Vol 12 (1) ◽  
pp. 273-288 ◽  
Author(s):  
Thomas M. Smith ◽  
Robert E. Livezey
Keyword(s):  
North America ◽  
General Circulation ◽  
Southern Oscillation ◽  
Circulation Model ◽  
Systematic Errors ◽  
Sea Surface ◽  
Average Output ◽  
Surface Temperatures ◽  
Model Variability ◽  
The U.S

Abstract Specifications of 1- and 3-month mean Pacific–North America region 700-hPa heights and U.S. surface temperatures and precipitation, from global sea surface temperatures (SSTs) and the ensemble average output of multiple runs of a general circulation model with the same SSTs prescribed, were explored with canonical correlation analysis. In addition to considerable specification skill, the authors found that 1) systematic errors in SST-forced model variability had substantial linear parts, 2) use of both predictor fields usually enhanced specification performance for the U.S. fields over that for just one of the predictor fields, and 3) skillful specification and model correction of the heights and temperatures were also possible for nonactive or transitional El Niño–Southern Oscillation situations.

Primates of the Lac Pelletier Lower Fauna (Eocene: Duchesnean), Saskatchewan

1990 ◽  
Vol 27 (4) ◽  
pp. 520-524 ◽  
Author(s):  
John E. Storer
Keyword(s):  
North America ◽  
Great Plains ◽  
Rocky Mountains

Four genera of primates are present in the early to mid-Duchesnean Lac Pelletier Lower Fauna. Phenacolemur leonardi sp.nov., Trogolemur sp., Omomys sp., and Macrotarsius cf. M. montanus make up the latest diverse primate assemblage known from North America and from the Great Plains. This primate assemblage is similar to the earliest Duchesnean assemblage from the Wood locality, Badwater Creek area of central Wyoming, and primate genera appear to have been widely distributed through the Great Plains and Rocky Mountains in the Uintan–Duchesnean.

Sign in / Sign up

Export Citation Format

Share Document