Você está na página 1de 27

Adjustment of Global Gridded

Precipitation for Orographic


Effects

Jennifer C. Adam 1
Elizabeth A. Clark 1
Dennis P. Lettenmaier 1
Eric F. Wood 2
1. Dept. of Civil and Environmental Eng., University of Washington
2. Dept. of Civil Eng., Princeton University
Motivation:
The Orographic Effect on Precipitation
precipitation divide
increased decreased
condensation precipitation:
and the rain
precipitation shadow
due to
orographic
lifting

* No global gridded precipitation datasets take into account


the effects of orographic lifting
Figure taken from http://jamaica.u.arizona.edu
PRISM (Daly et al. 1994, 2001, 2002)
(Parameter-elevation Regressions on
Independent Slopes Model)

2.5 minute
Topographic
facets
Regresses P
against
elevation on
each facet
0 100 200 300 400 500 mm/month
Basin Area/Station Location Distributions
100%
Africa
20%
100% Station Count
Asia Basin Area
Cumulative Percent

20%
100%
Australia
20%
100%
Europe
20%
100%
North America
20%
100%
South America
20%
0.5 1.0 1.5 2.0 2.5
Elevation, km
22 GPCC 1/81/8 PRISM 22 PRISM

15,000 Stream Flow


400 800 1200 1600
Precipitation, mm/year Simulations
10,000

5,000
0 m3/s
J F M A M J J A S O N D
Observed
with GPCC precipitation
From Nijssen et al. 2001 (J. Clim.) with PRISM precipitation
Objectives
Consistent framework to account for
orographic effects in 0.5 gridded gauge-
based precipitation estimates on a global
scale
Utilize existing gridded precipitation
product: Willmott & Matsuura (2001) with
correction for precipitation under-catch by
Adam & Lettenmaier (2003).
Mean annual correction for 1979-1999
Outline of Steps
Step1.
1 Definition of Correction Domain (regions of
complex topography, only)

Step2.
2 Average Correction Ratio for Gauged Basins:
the Budyko Method

Step3.
3 Fine-Scale Spatial Distribution of Correction
Ratios within Gauged Basins

Step4.
4 Fine-Scale Interpolation of Correction Ratios
to Un-Gauged Basins then aggregate to 0.5
Step 1 Select Correction Domain
1. Pre-selection according to slope:
- slopes calculated from 5-minute DEM
- aggregated to half-degree
2. Set Slope Threshold
- 6 m/km (the approximate slope above
which Willmott & Matsuura (2001)
differs by more than 10% from PRISM)
3. Final Domain smoothing then final
selection
0 5 10 15 20 25 30 35 40
Slope, m/km
Step 2 Basin-Average Correction
Ratios: the Budyko Method
1. Determine actual basin average
precipitation by solving 2 simultaneous
equations:
(1) Water Balance equation: P = E + Q
(2) Derivative of Budyko (1974) E/P vs.
PET/P curve
2. Calculate average correction ratio for each
basin:
Rave = Pactual Poriginal
Budyko (1974) Curve
1.2 Moisture
Limited
1.0
Energy
0.8
Limited
E/P

0.6
0.4
0.2
0.0
0 1 PET/P 2 3
Budyko (1974) S&V, gamma=1.5
S&V, gamma=1.0 S&V, gamma=0.5
Physical Limits

S&V: Sankarasubrumanian and Vogel (2002)


Uses an additional parameter: soil moisture storage capacity
Gauged Basins

357 mountainous basins chosen


Spatial Distribution of Correction
Step 3
Ratios Within Gauged Basins
1. Break correction domain into a 5-min grid of
correction bands gives degree of topographic
influence
2. Spatially distribute: use quadratic equation
r (band ) = A band + B band + C
2

B, C: from constraints, i.e. (Rave conserved


over basin and rband unity outside correction
domain)
A: from regressing Rave with PRISM
(developed using 101 basins in western North
America, tested over 5 basins in central Asia)
e.g. San Joaquin, CA

Correction
Bands

Elevation, m
1 2 3 4 5 6

r (band ) = A band 2 + B band + C Correction Ratio


Step 4 Interpolate to Un-Gauged Areas
1. Interpolation at 5-min resolution: uses a
linear distance weighting scheme
Only interpolates to cells with the same
correction band and the same slope type
(i.e. windward vs. leeward determined
from NCEP/NCAR reanalysis data)
2. Aggregate 5-min correction ratios to the
final resolution 0.5
3. Apply to original data via multiplication
Results

0.4 0.8 1.2 1.6 2.0


Correction Ratio
Increases by Continent

Continent Corr. Domain All Areas


Global 20.2 % 6.2 %
Africa 7.4 % 1.4 %
Eurasia 20.3 % 10.0 %
North America 34.4 % 9.7 %
South America 26.6 % 3.6%
Summary
Satisfies a need for gridded precipitation
data that account for orographic effects in a
globally-consistent framework
Uses combination of water balance and
Budyko (1974) curve to get magnitude,
PRISM used to help derive spatial variability
Final product: 1979-1999 0.5 climatology
that accounts for gauge under-catch and
orographic effects (global increases are 11.7%
and 6.2%, respectively, for a net of 17.9%)
Increases with Elevation
100%
Africa
100%
Percent Increase

Australia
100%

Eurasia
100%

North America
100%
South America
100%
Globe
0 1 2 3 4 5 6 km
Implied PRISM
Adjustments
(as compared to
Willmott&Matsuura 2001)
Net: 13.6%, 35.5%
Cv = 1.41

Adam et al.
Adjustments

Net: 16.1%, 41.6%


Cv = 0.42

Percent Increase
Further PRISM Comparisons

Adam et al. PRISM


Correction Bands

Least Most
Topographic 1 3 5
Topographic
Influence 2 4 6 Influence
Dominant Effective Wind
Direction
Slope Type

Upslope
Downslope
Cross-Wind
Equations
dS
1 = P E Q G P = E +Q
dt

E P -Q
2 = f {, } = f {, }
P P
PET
Where = = Aridity Index
P
b
Where = = Soil Moisture Storage Index
P
b max = max(E) + max(S)
max(S) = Soil Moisture Storage Capacity
P, P PET
max(E) =
PET, P > PET
Water Balance

In General:
dS
= P E Q G
dt
Long term mean over watershed:
P = E +Q
Q obtained from streamflow
measurements
Distributions with Elevation
20% Station Density
Africa Area
Percent per Elevation Increment

20%
Asia
50%
Australia
20%
Europe
20%
North America
20%
South America

0 0.5 1.0 1.5 2.0 2.5 km