zong-liang yang, guo-yue niu and robert e. dickinson* the university of texas at austin
DESCRIPTION
Modeling Runoff in CLM. Zong-Liang Yang, Guo-Yue Niu and Robert E. Dickinson* The University of Texas at Austin * Georgia Institute of Technology. Prepared for Hydrology Project, CCSM Land Model Working Group Meeting March 27, 2006 www.geo.utexas.edu/climate. - PowerPoint PPT PresentationTRANSCRIPT
Zong-Liang Yang, Guo-Yue Niu and Robert E. Dickinson*
The University of Texas at Austin
* Georgia Institute of Technology
Modeling Runoff in CLMModeling Runoff in CLM
Prepared for Hydrology Project, CCSM Land Model Working Group MeetingMarch 27, 2006
www.geo.utexas.edu/climate
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
1. Introduction
2. Design of a Simple TOPMODEL-Based runoff Scheme (SIMTOP)
3. Validate SIMTOP against GRACE ΔS
4. Development of a Simple Groundwater Model
5. Assess Model against GRACE ΔS
6. Conclusions
Outline
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Design of the Simple TOPMODEL-Based Runoff Scheme (SIMTOP)
Surface Runoff : Rs = P Fmax e – C f zwt
p = precipitation
zwt = the depth to water table
f = the runoff decay parameter which determines recession curve
Fmax and C = topographic parameters
Subsurface Runoff : Rsb= Rsb,maxe –f zwt
Rsb,max = the maximum subsurface runoff when the grid-mean water table is zero. It should be related to lateral hydraulic conductivity of an aquifer and local slopes. Rsb,max=1.0x10-4 mm/s through sensitivity experiments.
SIMTOP parameters:
Two calibration parameters Rsb,max and f Two topographic parameters Fmax and C
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Relationship Between the Saturated Area and Water Table Depth
Map of saturated areas showing expansion during a single rainstorm. [Dunne and Leopold, 1978]
zwt
fsat
fsat
fsat = Fmax(λ) e –C f zwt
λ – topographic wetness index derived from DEM
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Topographic Wetness Index: λ = ln(a/tanβ) = ln(a) – ln(S)
DEM –Digital Elevation Model
ln(a) – contribution area
ln(S) – local slope
The higher topographic wetness index, the wetter the pixel
1˚ x 1˚
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
1˚
1˚
TOPMODEL (Beven and Kirkby, 1979; Sivapalan et al., 1987) :
zi – zm = – (λi – λm) / f
where zi and λi are water table depth and topographic index at a pixel; while zm and λm are their grid-cell (catchment) mean values.
The Saturated Fraction the Grid-Cell:
Fsat = Prob { zi < 0 } or Prob { λi > λm – fzm } zi, λi
λ
PD
F
0.1
0.2
λm
1.0
0.5
CD
F
λλm
Fmax
zm λm
Lowlandhighland
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
A 1 ˚x 1˚ arid-cell in the Amazon River basin
Both Gamma and exponential functions fit for λi > λm
Fmax = 0.45; C = 0.6
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
A 1 ˚x 1˚ arid-cell in Northern Rocky Mountain
Gamma function fails, while exponential function works.
Fmax = 0.30; C = 0.5
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Woods and Sivapalan (2003) Fmax=0.35; C = 0.51 to 1.10
C ~ 0.6
Exponential function fits very well in well-developed catchments.
The larger the catchment, the better the fitting.
Justification of Surface Runoff Formulation and Derivation of Topographic parameters
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Global Fmax
a: Discrete Distribution
b: Gamma Function
c: Error of Gamma (b -- a)
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Justification of Subsurface Runoff Formulation
TOPMODEL (Beven and Kirkby, 1979)
Rsb= Rsb,max e fS
where S is the deficit of the subsurface water storage
Sivapalan et al., (1987) and Stieglitz et al. (1997 )
Rsb= K0/f e –λ e –f zwt
It needs very large K0, which is justified by soil surface macropore (1000 times lager than in LSM);
Chen and Kumar (2001):
Rsb = α K0/f e –λ e –f zwt (where αK0 is the lateral K)
Difficulties in determining α globally; λ needs very high resolution DEM (30 m or finer) to determine slopes.
Niu et al. (2005):
Rsb = Rsb,max e –f zwt (Rsb,max= 1.0x10-4 mm/s)
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
The Exponential Relationship between Streaflow and Water Table Depth
Groundwater level is highly correlated with streamflow in a strong nonlinear manner and
explains 2/3 of the streamflow (Yeh and Eltahir, 2005)
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Diagnostic Water Table Depth from Soil Moisture Profile
Water profile under
gravity
Soil water profile when gravity equals
to capillary force
Ψi – zi
Ψsat – zwt
GravityGravity
Capillary
Chen and Kumar (2001)
Koster et al. (2000);
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validation Against GRACE Terrestrial Water Storage Change Data
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validation Against GRACE Terrestrial Water Storage Change Data
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validation Against GRACE Terrestrial Water Storage Change Data
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validation Against GRACE Terrestrial Water Storage Change Data
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Prognostic Water Table: A Simple Groundwater Model
bot
botbota zz
zzKQ
)(
Water storage in an unconfined Aquifer:
Recharge Rate:
)1(bot
bota zzK
Gravitational Drainage
sba RQ
dt
dW ya SWz /
Upward Flow under capillary forces
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
fzsbsb eRR max,
Groundwater Discharge (Baseflow or Subsurface Runoff)
Properties of the Aquifer
1. Hydraulic Conductivity:
2. Specific Yield:
SIMTOP (Niu et al., 2005)
)(,
botzzfbotsatsat eKK
2.0yS
Model Design: A Simple Groundwater Model (SIMGM)
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validate the Model against the Valdai Data
The model reproduces SWE, ET, runoff, and water table depth.
The water table depth has two peaks and two valleys in one annual cycle
WTD is very sensitive to soil permeability (sand percentage, frozen soil), Runoff, and ET parameters.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validate the Model against GRDC Runoff
Good agreements between the modeled runoff and GRDC Runoff;
Runoff ~ exp(- f *wtd)
The modeled water table depth ranges from 2.5m in wet regions to 30m in arid regions.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Regional Averaged Runoff
Agreement between the modeled runoff and GRDC runoff in most regions except for mid-latitudes;
Surface runoff accounts for about 20%;
Groundwater discharge accounts for about 80% of the total runoff.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validate the Model Against GRACE ΔS Anomaly
The modeled ΔS anomaly agrees very well with GRACE data in river basins where ΔS is not affected by frozen soil;
Groundwater ΔS anomaly accounts for about 60-80% of the total ΔS anomaly;
The model capture the inter-annual and inter-basin variability of the ΔS anomaly .
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Validate the model Against GRACE WTD Anomaly
GRACE ΔS / 0.2
The modeled water table depth agrees very well with GRACE data in terms of inter-annual and inter-basin variability in river basins where ΔS is not affected by frozen soil;
The uncertainty in GRACE data is mainly the attenuation effects induced by smoothing.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
P – E, Groundwater Recharge, and Discharge
Phase lags in P – E, groundwater recharge and discharge;
Negative recharge in dry seasons when P – E is negative
Variations of P – E, groundwater recharge, discharge are consistent with the groundwater storage anomalies and WTD in terms of the inter-annual and inter-basin variability.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
The Impacts of Groundwater Model on SM and ET
It has a large impact on bottom-layer soil moisture, most obviously in cold regions (30% globally);
It has a smaller impacts on surface-layer soil moisture (5% globally);
The impacts on ET are mostly in arid-to-wet transition zones, i.e., the “hot spots” (20% in sensitive zones).
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Soil Moisture Profiles in Selected Regions
It has a large impacts on the soil moisture profile in most regions;
It has a relative small impacts in arid regions because the WTD is very deep and thus the capillary forces are weak.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Transpiration vs. Soil Surface Evaporation
Groundwater has a negligible impacts on transpiration, although it greatly increases deep soil moisture;
It enhanced the ground-surface evaporation in dry seasons corresponding to the increases in the surface-layer soil moisture.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions
Conclusions
1. We developed a simple groundwater model (SIMGM) for use in GCMs by representing the recharge and discharge processes in an unconfined aquifer, which is added as a single integration element.
2. It is first validated against the observed water table depth in a small cold-region watershed. It captures not only the summer valley also the winter valley of the observed water table.
3. On the global scale, it reproduces the GRDC runoff; Groundwater discharge accounts for about 80% of the total runoff.
4. The modeled ΔS anomaly agrees very well with GRACE data in terms of inter-annual and inter-basin variability in most river basins.
5. Groundwater ΔS anomaly accounts for about 60-80% of the total ΔS anomaly; The modeled water table depth agrees very well with that converted from GRACE. The groundwater storage and WTD anomalies are mainly controlled by P – E, or climate.
6. It produces a much wetter soil globally except for arid regions; It produces about 4 – 20% more annual ET mainly through the enhanced ground surface evaporation instead of transpiration in humid-arid regions.
Introduction | SIMTOP | Validation | SIMGM | Validation | Conclusions