Comparisons of Simulation Results Using the NWS

1
HYDROLOGY LABORATORY, OHD/NWS/NOAA
Comparisons of Simulation Results Using the NWS
Hydrology Laboratory's Research Modeling
System (HL-RMS)
Ziya Zhang, Victor Koren, Seann Reed, Michael
Smith, and David Wang
Hydrology Laboratory Office of Hydrologic
Development National Weather Service/NOAA
2
What is HL-RMS?
A Research Modeling System being developed by
researchers in the Hydrology Laboratory, Office
of Hydrologic Development, NWS/NOAA. Modeling
framework for testing lumped, semi-distributed,
and fully-distributed hydrologic modeling
approaches.
?
?
3
HL-RMS Components
?
Pre-simulation Determines connectivity
matrix Derives SAC parameters Derives routing
parameters Simulation Simulates rainfall/runoff
for each grid (SAC, ...) Conducts hillslope and
channel routing Computes hydrographs
?
?
?
?
?
?
?
4
HL-RMS Capabilities
?
Ingests NEXRAD Stage III xmrg data Uses lumped or
distributed model parameters Uses lumped or
distributed precipitation Computes hydrographs at
any interior points Selection between lumped,
semi-distributed, and fully distributed modes
?
?
?
?
5
HL-RMS Capabilities
cont.
Output Arc/Info grids for selected
variables Flexible to modify gridded
parameters Modular design to test other
models Computational element NEXRAD 4km HRAP grid
?
?
?
?
6
Test Basin
Blue River Basin, OK
Area 1233 km2
7
Test Runs
Continuous Simulation from 1993 to 2000
?
Lumped parameters from manual calibration, lumped
forcing Distributed parameters, distributed
forcing Scaled distributed parameters,
distributed forcing
?
?
Scale factor was based on comparing a calibrated
value of a basin to an averaged value of gridded
values of that basin
8
Samples of Gridded Parameters
9
Test Results
Distributed Channel Runoff
Channel Runoff (cms)
November 7, 1996 2200
outlet
10
Test Results
Rainfall Surface Runoff
November 7, 1996 700 - 800
11
Test Results
Hydrographs _at_ Interior Points
A
B
C
Basin Outlet
12
Test Results
Hydrographs _at_ Interior Points
A
B
Basin Outlet
C
13
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1994
outlet
14
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1995
outlet
15
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1996
outlet
16
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1997
outlet
17
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1998
outlet
18
Test Results
Hydrograph Comparison
Total Precipitation (mm)
An Event in 1999
outlet
19
Observations
?
Simulations using HL-RMS in distributed mode
without any calibration effort yields better
results than calibrated lumped simulations at
most times Distributed modeling yields much
better results for events where rainfall has big
variation across a basin Distributed simulation
results are better when the SAC parameters are
scaled based on calibrated parameters
?
?
20
Observations
cont.
?
Rainfall distribution has the biggest effect on
differences between lumped and distributed
simulations Derived gridded rainfall-runoff and
routing parameters are reasonably good Use of
only gridded parameters yields higher peaks in
hydrographs Recession part of simulated
hydrographs is off the most compared to observed
discharges
?
?
?
21
Observations
cont.
?
HL-RMS is a workable and a flexible research tool
to study hydrologic processes, and to test the
sensitivity of model parameters and input forcing
on simulations