1. Runoff Simulations in Region12 (or almost the State of Texas)
Bryan Hong, Ph.D.

2. Outline of the Major Tasks
Preparing atmospheric forcing input for NOAH land surface model
1. Obtaining model predicted or observed data
- Next Generation Rain data (NEXRAD) for precipitation and North American Land Data Assimilation System (NLDAS) for others.
- Research period : from 2004 to 2007 (4 years)
2. Data processing for model input
- Spatial down-scaling
- Converting data format
Producing more accurate runoff simulations for Cedric’s routing model (RAPID)
1. Analyzing various simulations based on different runoff schemes
2. Analyzing runoff changes based on vegetation or soil parameter changes

3. Meteorological Variables
List of variables for model input
NLDAS from NCEP Eta Data Assimilation System (EDAS) output fields
via
Air Temperature at 2 m (K)
Specific Humidity at 2m (kg/kg)
Surface Pressure (Pa)
U Wind Component (m/s)
V Wind Component (m/s)
Downward Short Wave Radiation (W/m2)
Downward Long Wave Radiation (W/m2)
Total Precipitation (mm/s)
from Cosgrove, 2002 at GAPP meeting
Projection type
Geographic lat/long
from Lambert-conformal of EDAS
Spatial Resolution
0.125 degree (~ 15km)
interpolated from 40km base
Temporal Resolution
Hourly
interpolated from 3 hourly base

4. Meteorological Variables
NEXRAD rainfall
Precipitation NCEP/EMC
4km Gridded (GRIB)
Stage IV Data
via
From
Projection type
Polar Stereo Graphic
Spatial Resolution
~ 4km
Temporal Resolution
Hourly

5. Study Domain (Region 12) Map Projection Albers Spatial Resolution
Number of grid boxes
228 X 228
Total area
About 1M km2

6. Lost grids around coast lines
Data Processing
Missing NEXRAD data
Complemented by Stage II data
Lost grids around coast lines
due to the coarse resolution of NLDAS
Filled with the values of nearest neighbor grids
Lost grids around coast lines
NLDAS coverage
Study domain coverage
Lost grids

7. (Dickinson et al., 1998; Yang and Niu, 2003)
Runoff Model Experiments
Based on Niu et al., 2009
Dynamic Vegetation
SM factor (β)
for stomatal Res.
Runoff Schemes
EXP1
Dynamic Leaf Model
(Dickinson et al., 1998; Yang and Niu, 2003)
NOAH Type
SIMGM
EXP2
SIMTOP
EXP3
Schaake 96
EXP4
BATS
EXP5
CLM Type
Oleson et al., 2004
EXP6
EXP7
EXP8
EXP9
BATS Type
Dickinson et al., 1993
EXP10
EXP11
EXP12
TOPMODEL + groundwater scheme
(Niu et al., 2007)
Biosphere-Atmosphere Transfer scheme
(Dickinson et al., 1993)
Original NOAH surface runoff scheme
(schaake, 1996)
TOPMODEL +
equilibrium water table
(Niu et al., 2005)

8. Runoff Simulation
Comparison to stream flow observations in Guadalupe and San Antonio River Basins
SYMGM Group
BATS Group
Guadalupe River Basin
OBS_G
EXP1_G
EXP5_G
EXP9_G
EXP4_G
EXP8_G
EXP12_G
Mean
80.98
50.48
50.63
47.18
71.56
73.68
67.85
STDV
138.89
130.05
133.73
124.69
263.58
272.57
255.82
MIN
5.66
4.88
4.01
4.52
1.43
1.22
1.23
MAX
San Antonio River Basin
OBS_S
EXP1_S
EXP5_S
EXP9_S
EXP4_S
EXP8_S
EXP12_S
Mean
37.54
31.20
31.07
29.44
43.60
44.60
41.52
STDV
63.96
100.78
102.56
97.30
183.95
188.18
178.38
MIN
3.09
2.85
2.46
2.645
0.66
0.55
0.13
MAX
639.96
Lower mean but closer STDV
Closer mean but too high STDV

9. NOAH β + SYMGM CLM β + SYMGM BATS β + SYMGM Guadalupe River Basin
Discharge (m3/s)
2004
2005
2006
2007
CLM β + SYMGM
Discharge (m3/s)
2004
2005
2006
2007
BATS β + SYMGM
Discharge (m3/s)
2004
2005
2006
2007

10. NOAH β + BATS CLM β + BATS BATS β + BATS Guadalupe River Basin
Discharge (m3/s)
2004
2005
2006
2007
CLM β + BATS
Discharge (m3/s)
2004
2005
2006
2007
BATS β + BATS
Discharge (m3/s)
2004
2005
2006
2007

11. NOAH β + SYMGM CLM β + SYMGM BATS β + SYMGM San Antonio River Basin
Discharge (m3/s)
2004
2005
2006
2007
CLM β + SYMGM
Discharge (m3/s)
2004
2005
2006
2007
BATS β + SYMGM
Discharge (m3/s)
2004
2005
2006
2007

12. NOAH β + BATS CLM β + BATS BATS β + BATS San Antonio River Basin
Discharge (m3/s)
2004
2005
2006
2007
CLM β + BATS
Discharge (m3/s)
2004
2005
2006
2007
BATS β + BATS
Discharge (m3/s)
2004
2005
2006
2007

13. Other Experiments Land Cover Effect ? MAJOR LAND USE
Kim, 2008 at NASA IDS meeting
MAJOR LAND USE
For two river basins
Guadalupe
Built-up land (4.0)
Dry Crop/Pasture (51.8)
Crop/Grass Mosaic (34.5)
Mixed Forest (3.8)
San Antonio
Grassian (3.8)
Changing vegetation properties (roughness length and stomatal resistance) do not increase runoff much.

14. Other Experiments Soil Texture Effect ?
The land surface model uses only one soil texture either bottom or top soil
Runoff results does not show much difference between the use of soil texture maps.

15. Guadalupe River Basin (m3/s) San Antonio River Basin (m3/s)
Runoff vs. Rainfall
Choice of Rainfall Input?
Comparison of Runoff simulation to Rainfall inputs
Guadalupe River Basin (m3/s)
San Antonio River Basin (m3/s)
OBS_G
EXP1_G
EXP8_G
NLDAS
NEXRAD
OBS_S
EXP1_S
EXP8_S
Mean
80.98
50.48
73.68
47.46
51.27
31.64
32.35
STDV
138.89
130.05
272.57
118.74
143.18
83.76
101.51
MIN
5.66
4.88
1.22
3.09
2.854
0.5494
MAX
3458.5
639.96
891.75
OBS : Stream Flow Measurements for River Basins (m3/s)
EXP1 : Experiment with the ground runoff scheme and NOAH β (Niu et al., 2009)
EXP8 : Experiment with the BATS runoff scheme and CLM β (Niu et al., 2009)
NLDAS : 20% of NLDAS precipitation
NEXRAD : 20% of NEXRAD precipitation
cf. All experiments shown above have been conducted with the NEXRAD rain data.
Assumed Runoff ≈ 20% of rainfall

16. Subsurface Runoff Analysis
What Happen in Runoff?
A simple analysis for subsurface runoff and comparison to OBS.
Assumptions
During drying periods (no precipitation periods), subsurface runoff is the only source for stream flow. Standard deviations of subsurface runoff are generally low, indicating that water is steadily supplied from ground water.
When surface runoff is almost zero (less than 1.0 m3/s in this analysis), the observations also represent stream flows only from subsurface runoff.
OBS
EXP1
EXP5
EXP9
EXP4
EXP8
EXP12
Guadalupe
Mean
73.33
20.37
19.97
18.47
8.47
8.35
7.29
STDV
129.73
16.00
16.05
15.42
7.86
7.75
7.15
San Antonio
32.40
11.72
11.33
10.64
4.35
4.25
3.72
55.81
11.24
11.17
11.23
5.59
5.51
5.57
Total analyzed dates for dry seasons : of total 1461 dates for Guadalupe river basin
for San Antonio river basin

17. Summary
Experiments indicates that choice of β scheme is not much effective to runoff variation.
If the SYMGM runoff scheme is the best choice as shown in the previous research , we need more water for runoff simulation.
For the two river basin areas, changing vegetation and soil parameters is not substantially effective to increase runoff amount.
NEXRAD rain is a better choice to more water input than NLDAS rain.
The land surface model underestimates subsurface water in those two river basin areas.

18. Future Works PLEASE, GIVE ME WATER!! THANKS
Need runoff output validation with stream flow observations for entire Region12
Comparison the routing simulations with different runoff outputs from SYMGM and BATS runoff schemes
Comparison runoff simulations between predicted vegetation from the dynamic leaf model and NASA remote sensing data such as NDVI and LAI from MODIS
Experiments with various saturated hydraulic conductivities to obtain more subsurface runoff
PLEASE, GIVE ME WATER!! THANKS

19. San Antonio River Basin
Guadalupe River Basin
Discharge (m3/s)
2004
2005
2006
2007
San Antonio River Basin
Discharge (m3/s)
2004
2005
2006
2007
Cf. NEXRAD : 20% of NEXRAD precipitation