Development of an Ensemble
Gridded Hydrometeorological
Forcing Dataset over the
Contiguous United States
Andrew J. Newman1, Martyn P. Clark1, Jason Craig1, Bart Nijssen2,
Andrew Wood1, Ethan Gutmann1, Naoki Mizukami1, Levi Brekke3, and
Jeff R. Arnold4
National Center for Atmospheric Research, Boulder CO, USA
2 University of Washington, Seattle WA, USA
3U.S. Department of Interior, Bureau of Reclamation, Denver CO, USA
4 U. S. Army Corps of Engineers, Institute for Water Resources, Seattle
WA, USA
1
Outline
•
•
•
•
•
•
Motivation
Methodology
Input station data
Validation
Examples
Summary
Motivation
Methodology
Input Data
Validation
Examples
Summary
Opportunities for hydrologic prediction
hydrological predictability
meteorological predictability
Meteorological predictability:
How well can we forecast the
weather and climate?
Hydrological Prediction: How
well can we estimate the
amount of water stored?
Accuracy in precipitation
estimates
Fidelity of hydro model
simulations
Effectiveness of hydrologic data
assimilation methods
Opportunity: Characterize
hydrologic and historical
forcing uncertainty
Water Cycle (from NASA)
Motivation
Methodology
Input Data
Validation
Examples
Summary
Hydrologic model uncertainties
• Characterize uncertainty in hydrologic model
simulation
▫ Uncertainty in historical model inputs
 Probabilistic quantitative precipitation estimation

▫ Uncertainty in model parameter choice and model
structure
▫ …leading to estimate of uncertainty in model states and
fluxes …and uncertainty in initial conditions
• Reduce uncertainty in modeled hydrologic states
▫ Probabilistic snow estimation from stations and satellites
▫ Ensemble snow data assimilation methods
Motivation
Methodology
Input Data
Ensemble Generation
Validation
Examples
Summary
Example over the Colorado Headwaters
Step 1:
Estimate probability of
precipitation (PoP) via
logistic regression, amount
and uncertainty at each
grid cell (locally weighted
regression & residuals)
observations
Clark & Slater (2006), Newman et al. (2014, in prep)
Motivation
Methodology
Input Data
Validation
Examples
Summary
Ensemble Generation
Example over the Colorado Headwaters
Step 2:
Synthesize ensembles
from PoP, amount & error
using spatially correlated
random fields
observations
Clark & Slater (2006), Newman et al. (2014, in prep)
Motivation
Methodology
Input Data
Validation
Examples
Summary
Development of Serially Complete Station
Dataset
• Daily Global Historical Climate Network (GHCN) and
SNOTEL observations for 1980-2012
• USA, Canada, Mexico
• Following Eischeid et al. (2000), processed stations with >10
years of data and nearby stations that have 10 years of overlap
with target station(s)
• Then, for each station:
• Fill 1-2 day temperature gaps with interpolation
• Fill all precipitation and > 2 day temperature with CDF
matching using highest available rank correlation station
Motivation
Methodology
Input Data
Validation
Examples
Summary
Development of Serially Complete Station
Dataset
• 12,000+ stations with serially complete data
• Precipitation, temperature or both
Motivation
Methodology
Input Data
Validation
Examples
Summary
Validation: Discrimination & Reliability
• Discrimination (a-d, red & black lines)
• Want to maximize separation of event, non-event PDFs
• Reliability (e-h, blue & black lines)
• Want to be close to 1-1 line (predicted = observed
probabilities)
Motivation
Methodology
Input Data
Validation
Examples
Summary
Validation: Comparisons to other datasets
• Maurer
Maurer et al.
(2002)
•
NLDAS-2
Xia et al. (2012)
•
Daymet Ver. 2
Thornton et al.
(2014)
• NLDAS – Maurer
• Daymet - Maurer
Motivation
Methodology
Input Data
Validation
Examples
Summary
Validation: Comparisons to other datasets
• Maurer et al. (2002)
• Interpolation between
observations increases
precipitation occurrence
• Speckled pattern
•
Grid points with obs
• Ensemble:
• SCRFs applied to PoP
field more realistic PoP
• Reduces PoP
• Data differences
responsible for PoP
increases
Motivation
Methodology
Input Data
Validation
Examples
Summary
Validation: Comparisons to other datasets
• Minimal differences in mean
temperature
• Except in intermountain
west
• Maurer uses constant
elevation lapse rate (-6.5
K km-1)
• Ensemble lapse rate
derived from station data
(including Snotel)
• Maurer shown to be too
cold in higher elevations
(e.g. Mizukami et al. 2014)
• Ensemble warmer than
Maurer in high terrain
Motivation
Methodology
Input Data
Example Output
• Central US Flood of 1993
• June 1993 total precipitation
Validation
Examples
Summary
Motivation
Methodology
Input Data
Validation
Examples
Summary
Example Application
• Snowmelt dominated basin in Colorado Rockies
• Example water year daily temperature (a)
• Snow water equivalent accumulation (b)
• Simple temperature index model (optimized for Daymet
(green))
Motivation
Methodology
Input Data
Validation
Examples
Summary
Summary
• Developed first of its kind ensemble hydrometeorolgical dataset
• Required development of serially complete station dataset
• Both will be available for download at:
http://ral.ucar.edu/projects/hap/flowpredict/subpages/pqpe.php
• Methodology follows Clark and Slater (2006) with modifications
• Ensemble has:
• Realistic probability of precipitation (PoP)
• Estimate of forcing uncertainty
• Better observation error estimates for data assimilation
• Propagation of forcing uncertainty into model states
• Useful for state uncertainty estimation
• Ensemble calibrations will allow for estimates of parameter
uncertainty (impact of noise in parameter estimation)