Parameter regionalization of a monthly water balance model for the conterminous United States Nov. 16, 2015 Andy Bock, USGS, Colorado Water Science Center Lauren Hay, USGS, National Research Program (NRP) Greg McCabe, USGS, NRP Steve Markstrom, USGS, NRP R.Dwight Atkinson, U.S. Environmental Protection Agency
http://www.hydrol-earth-syst-sci-discuss.net/12/10023/2015/hessd-12-10023-2015.pdf
Research Direction Determine water availability at sub-watershed resolution for the conterminous United States (CONUS) Needs to account for human and ecological needs Assessments for areas that lack reliable measured streamflow information
Research Direction Reference Quality Kiang and others, 2013
Research Direction Areas of poor model skill Nash-Sutcliffe Coefficient (NSE) Newman and others, 2015 Increasing Skill NSE measures skill in simulated streamflow in replicating natural streamflow
Research Direction Areas of poor model skill Nash-Sutcliffe Coefficient (NSE) Newman Bock and and others, others, 2015 2015 Increasing Skill NSE measures skill in simulated streamflow in replicating natural streamflow
Research Direction CONUS-wide water availability Development of methods to predict hydrologic response in areas that are ungaged or lack reliable measured streamflow (regionalization) o For hydrologic models, determine the appropriate amount of information to transfer to ungaged areas o How we can use this framework to add uncertainty to estimates of streamflow for these areas?
Sensitivity Analysis A framework for regionalization Utilize parameter sensitivity analysis (SA) Apply a SA method to across CONUS for a Monthly Water Balance Model (MWBM) Identify areas or regions of similar model behavior based on the spatial variability of parameter sensitivities Calibrate MWBM for these areas in a group-wise fashion Simulated Observed
Sensitivity Analysis A framework for regionalization Utilize parameter sensitivity analysis (SA) Apply a SA method to across CONUS for a Monthly Water Balance Model (MWBM) Identify areas or regions of similar model behavior based on the spatial variability of parameter sensitivities Calibrate MWBM for these areas in a group-wise fashion Site1 Site2 Site3
Monthly Water Balance Model McCabe and Markstrom, 2007 MWBM Parameters Range Default 1. Drofac Controls fraction of PPT that becomes runoff 0, 0.10 0.05 2. Rfactor Controls fraction of surplus that becomes runoff 0.10, 1.0 0.5 3. Tsnow Threshold below which all PPT is snow ( o C) -10.0, -2.0-4.0 4. Train Threshold above which all PPT is rain ( o C) 0.0, 10.0 7.0 5. Meltcoef Proportion of snowpack that becomes runoff 0.0, 1.0 0.47
Sensitivity Analysis Fourier Amplitude Sensitivity Test FAST (Cukier and others, 1973, 1975; Reusser and others, 2011) Quantifies the first-order partial variance (FOPV) of model output explained by each parameter Identify the influential/non-influential parameters
Monthly Water Balance Model Geospatial Fabric for Nat. Hydrologic Modeling 110,000 Hydrologic response units (HRU) Viger and Bock, 2014
Application of Sensitivity Analysis FAST 1000x Parameter Sets Simulation of MWBM for each parameter set at each HRU for the period of record 1950-2010 (Maurer and others, 2002) Measure parameter sensitivity for the Rainfall Runoff Ratio (R/P) (RR)
Application of Sensitivity Analysis
Calibration and Regionalization Geospatial Fabric Gage FAST calibration Gage
Calibration and Regionalization Model Calibration Mean Monthly Z-scores Scaled measured (1950-2010) and simulated mean monthly streamflow to Z-scores: Z 1,12 = (x 1,12 -µ)/σ Only parameters identified as sensitive in the FAST analysis were calibrated Parameters not considered sensitive were kept at default values
Calibration and Regionalization Model Calibration Mean Monthly Z-scores Scaled measured (1950-2010) and simulated mean monthly streamflow to Z-scores Diagnose model skill in different parts of the monthly flow regime
Nash-Sutcliffe Efficiency *GageNSE NSE for each gage from simulated streamflow produced from calibrated parameters for each individual streamgage *GroupNSE NSE for gage from simulated streamflow produced from group parameters
Mean Monthly Z-scores
Error by Gage *High month with highest mean monthly measured streamflow *Median 2 months with the median mean monthly measured streamflow * Low month with the lowest mean monthly measured streamflow
Error by Region
Results Spatial variability in parameter sensitivities gave some indication of dominant physical processes across CONUS Regionalization based upon SA provided adequate model representation of mean-flow and to a lesser extent low-flow portions of the mean monthly hydrograph adequately through most of the CONUS Model skill was poorer for high-flow portions of the mean monthly hydrograph, especially in regions where finer-scale controls, both temporally and spatially, exert more control over the dominant hydrologic processes
References Bock, A.R., L. E. Hay, G. J. McCabe, S. L. Markstrom, and R. D. Atkinson, 2015, Parameter regionalization of a monthly water balance model for the conterminous United States, Hydrol. Earth Syst. Sci. Discuss., 12, 10023-10066, doi:10.5194/hessd-12-10023-2015,. Cukier, R.I., C.M. Fortiun, K.E. Shuler, A.G. Petschek and J.H. Schaibly, 1973. Study of sensitivity of coupled reaction systems to uncertainties in rate coefficients 1, J. Chem. Phys., 59, 3873-3878. Cukier, R.I., J.H. Schaibly, and K.E. Shuler, 1975. Study of sensitivity of coupled reaction systems to uncertainties in rate coefficients 3, J. Chem. Phys., 63, 1140-1149,. Kiang, J.E., D.W. Stewart, S.A. Archfield, E.B. Osborne, and K. Eng., 2013. A National Streamflow Network Gap Analysis, U.S. Geological Survey, Scientific Investigative Reports 2013-5013, 94 pp. Maurer, E.P., A.W. Wood, J.C. Adam, D.P. Lettenmaier, and B. Nijssen, 2002. A Long-Term Hydrologically-Based Data Set of Land Surface Fluxes and States for the Conterminous United States, Journal of Climatology 15. McCabe, G.J., and S.L. Markstrom, 2007. A monthly water-balance model driven by a graphical user interface: U.S. Geological Survey Open File Report 2007-1088, 6 p.
References Reusser, D., W. Buytaert, and E. Zehe, 2011. Temporal dynamices of model parameter sensitivity for computationally expensive models with the Fourier amplitude sensitivity test, Water Res. Res., 47, W07551. Sankarasubramanian, A. and R.M. Vogel, 2003. Hydroclimatology of the continental United States,Geophys. Res. Lett., 30, 1 4, doi:10.1029/2002gl015937. Viger, R.J., and A. Bock, 2014. GIS Features of the Geospatial Fabric for National Hydrologic Modeling, US Geological Survey, doi: http://dx.doi.org/10.5066/f7542kmd.
http://www.hydrol-earth-syst-sci-discuss.net/12/10023/2015/hessd-12-10023-2015.pdf Steve Markstrom Wednesday, Session 59 o Towards Simplification of Hydrologic Modeling: Identification of Dominant Processes