Estimating Continental-Scale Water Balance through Remote Sensing
Huilin Gao1, Dennis P. Lettenmaier1
Craig Ferguson2, Eric F. Wood2
1 Dept. of Civil and Environmental Engineering, University of Washington2Dept. of Civil and Environmental Engineering, Princeton University
2008 Fall AGU meetingU N I V E R S I T Y O F
WASHINGTONPRINCETONUNIVERSITY
Motivation
1.Importance for understanding water budget at continental scale2.Limitations of observations and modeling3.Advantages of remote sensing4.Challenges of remote sensing
∆S = P –R– ET
Research questions: how closely can the water budget be estimated solely using remote sensing data? What are the major error sources? What is the role of reservoir in water storage change?
1
Research StrategyR (observed) ?=? P – ∆S – ET (remote sensing)
Research Domain – Continental U.S.
Precipitation ET ΔS Runoff
Remote sensing TRMM 3B42-RT MODIS by Princeton
GRACE by CSR; GFZ; JPL
By difference
Observed/Modeled
Gridded gauge data
*VIC output *VIC output Observed runoff
High quality precipitation from gridded gauge measurements - help evaluate P
Variable Infiltration Capacity (VIC) model outputs using good forcings
- help evaluate ΔS and ET
2
1. Arkansas-Red 5. East Coast 9. Lower Mississippi 13. Rio Grande2. California 6. Great Lakes 10. Upper Mississippi 3. Colorado 7. Great Basin 11. Missouri4. Columbia 8. Gulf 12. Ohio
Major River Basins within the U.S.
Study period: 2003 ~ 2006Grid resolution: 0.5 deg; Temporal resolution: hourly, daily, monthly
123
4
5
6
7
89
10
11
12
13
3
Methodology
Tair (inst)(AIRS)
Net Longwave
Albedo(MODIS)
Downward Solar(GOES)
Net Shortwave
Net Radiation
Precipitation(TRMM)
Rainfall
Snow
ET (inst)(MODIS)
ET
ΔS(GRACE)
Snowmelt
Runoff Obs. Runoff
Tair > 0
Tair < 0
EF
Calibration, interpolation
Tair(hourly)
Grid
ded
gaug
e da
ta
Model output
Model output
4
Seasonal Precipitation
• TRMM real time product has significant errors in some basins
• Precipitation from remote sensing needs to be corrected for orographic effect
5
Seasonal Evapotranspiration
• It is difficult to validate remotely sensed ET at the continental scale
• Remotely sensed and modeled ET are seasonally consistent
6
Seasonal Storage Change
• GRACE products from different data centers are similar
• GRACE products over the west coast suffer from “signal leakage”
Range offset
VIC Max-Min ΔS (mm)
GR
AC
E M
ax-M
in Δ
S (
mm
)
California
Columbia
7
ΔSΔSΔS
Remote Sensing Capability by Basin
Good
Good
Good
Ark
ansa
sC
alifo
rnia
Col
orad
oC
olum
bia
Eas
t Coa
stG
reat
Lak
esG
reat
Bas
in
Low
er M
issi
Upp
er M
issi
Mis
sour
i
Gul
f
Ohi
oR
io G
rand
e
Cor
rela
tion
Coe
ff.
MA
E(m
m/m
o)R
ange
Off
set(
mm
)
Precip ET ΔS
8
Good
Good Good
PrecipETΔS
• TRMM real time precipitation has the largest error among the three
• ET has the best seasonal representation, but it is biased over some basins
• GRACE water storage change is biased low over the west coast
Remote Sensing Capability over All Basins
9
Seasonal Runoff
It is difficult to close water budget by solely using remote sensing data
10
damsmajor rivers
(Graf, 2006)Large Dams (storage > 1.2 km3) in the United States
Reservoir Impacts on Water Storage Change
11
15mm
5mm
GRACE
(http://www.legos.obs-mip.fr/en/soa/hydrologie/hydroweb/)
Remote Sensing of Reservoir Storage
12
Summary
• Accuracy towards closing the water budget at the continental scale from remote sensing heavily depends on precipitation quality;
• GRACE water storage change tends to be biased low over the west coast;
• Remotely sensed ET over the 13 basins is consistent with VIC output;
• Reservoir storage is a significant component for understanding terrestrial water storage.
13
Thanks!!!Questions?