recent progress and opportunities in macroscale hydrological modeling
DESCRIPTION
RECENT PROGRESS AND OPPORTUNITIES IN MACROSCALE HYDROLOGICAL MODELING. Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington presentation for Hydrological Sciences Review Royal Netherlands Academy of Arts and Sciences May 21, 2003. - PowerPoint PPT PresentationTRANSCRIPT
RECENT PROGRESS AND OPPORTUNITIES IN MACROSCALE
HYDROLOGICAL MODELING Dennis P. Lettenmaier
Department of Civil and Environmental Engineering
University of Washington
presentation for
Hydrological Sciences ReviewRoyal Netherlands Academy of Arts and Sciences
May 21, 2003
Outline of this talk
1) Macroscale modeling approach a) Strategy
b) Testing and evaluationc) Implementation
2) Examples a) Derived data sets b) S/I streamflow forecasting c) Hydrologic effects of climate change
3) Weak links and research opportunities
1. Macroscale modeling approach
a) Strategy
Traditional “bottom up” hydrologic modeling approach (subbasin by subbasin)
Flood of record
• Principal calibration locations were the Skykomish at Gold Bar and the Snoqualmie at Carnation
Snoqualmie River at Carnation, WA
Macroscale modeling approach (“top down”)
1 Northwest 5 Rio Grande 10 Upper Mississippi2 California 6 Missouri 11 Lower Mississippi3 Great Basin 7 Arkansas-Red 12 Ohio4 Colorado 8 Gulf 13 East Coast
9 Great Lakes
1. Macroscale modeling approach b) Testing and evaluation
Investigation of forest canopy effects on snow accumulation and melt
Measurement of Canopy Processes via two 25 m2 weighing lysimeters (shown here) and additional lysimeters in an adjacent clear-cut.
Direct measurement of snow interception
0
50
100
150
200
250
300
350
11/1/96 12/1/96 1/1/97 2/1/97 3/1/97 4/1/97 5/1/97
SW
E (
mm
)ObservedPredicted
Below-canopy
Shelterwood
Tmin = 0.4 C Zo shelterwood = 7 mmTmax = 0.5 C Zo below-canopy = 20 cm
Albedo based onexponential decaywith age; fitted tospot observationsof albedo
Calibration of an energy balance model of canopy effects on snow accumulation and melt to the weighing lysimeter data. (Model was tested against two additional years of data)
Summer 1994 - Mean Diurnal Cycle
Point Evaluation of a Surface Hydrology Model for BOREAS
Flu
x (W
/m2)
-100
100
300 Rnet
-50
50
150
250
H
0
60
120LE
0 3 6 9 12 15 18 21 24
SSA Mature Black Spruce
Rnet
H
LE
0 3 6 9 12 15 18 21 24
SSA Mature Jack Pine
Rnet
H
LE
0 3 6 9 12 15 18 21 24
Local time (hours)
NSA Mature Black Spruce
Observed Fluxes
Simulated Fluxes
Rnet Net Radiation
H Sensible Heat Flux
LE Latent Heat Flux
Range in Snow Cover ExtentObserved and Simulated
Eurasia North America
J F M A M J J A S O N D JMonth
Observed Simulated
0
4
8
12
16
20
sno
w c
ove
r ex
ten
t (1
06 km
2 )
J F M A M J J A S O N D JMonth
0
2
4
6
8
10
June 18th-July 20th, 1997
UPPER LAYER SOIL MOISTURE
0.40
0.10
0.20
0.30
SO
IL M
OIS
TU
RE
(%
)
XX
X
X
XX
X
XX
X
XX
XX X
X
TOPLATS regionalESTAR distributedTOPLATS distributed
11:00 CST JULY 12 1997
ESTAR TOPLATS
50
10
ESTAR TOPLATS
10
50
11:00 CST JUNE 20, 1997
Illinois soil moisture comparison
Mean Normalized Observed and Simulated Soil MoistureCentral Eurasia, 1980-1985
20°E 30°E 40°E 50°E 60°E 70°E 80°E 90°E 100°E 110°E 120°E 130°E 140°E
40°N 40°N
50°N 50°N
60°N 60°N
A
BC
D
E
F
G
H
0
100
200 S
oil M
oist
ure
(mm
)A
J F MA M J J A S O N D J
Nor
mal
ized
B
J F MA M J J A S O N D J
C
J F MA M J J A S O N D J
D
J F MA M J J A S O N D J
0
100
200
Soi
l Moi
stur
e (m
m)
E
J F MA M J J A S O N D J
Nor
mal
ized
F
J F MA M J J A S O N D J
G
J F MA M J J A S O N D J
H
J F MA M J J A S O N D J
Observed Simulated
Cold Season Parameterization -- Frozen Soils
Key
Observed
Simulated
5-100 cm layer
0-5 cm layer
1. Macroscale modeling approach c) Implementation
-60
-30
0
30
60
90
-60
-30
0
30
60
90
-150 -120 -90 -60 -30 0 30 60 90 120 150
Calibration at Global Scales
Selected 26 river basins, which were divided into calibration (red)
and validation (green).
•Only a limited number of model parameters were identified as calibration parameters. The remaining model parameters were determined independently and were not modified.
•Calibration parameters:•Infiltration capacity shape parameter bi
•Depth of second soil layer•Saturated hydraulic conductivity•Exponent for unsaturated hydraulic
conductivity.
•Calibration was performed for nine out of 26 river basins.
•Simulated from 1980-1993 and compared to observed discharge.
Seasonal Evapotranspiration (1980-1993)Uncalibrated (base case) simulation
Global Mean Annual Runoff Ratio (1980-1993)Uncalibrated (base case) simulation
2. Examples a) Derived data sets
LDAS Long-Term Retrospective Data Set, 1950-2000
Ed Maurer
Dennis LettenmaierUniversity of Washington
Department of Civil and
Environmental Engineering
Motivation
Baseline forcing data – for water and energy balance studies (e.g., GEWEX WEBS).
Derived “Pseudo-Observations” – for variables not widely measured (e.g., soil moisture) – analogous to reanalysis.
Climate variability and change - characterizing variability and change in variables not directly observed.
Implementation Strategy
VIC model implemented for 15 sub-regions, with consistent forcings.
Surface forcing data:
Daily precipitation; maximum and minimum temperatures (from gauge measurements)
Radiation, humidity parameterized from Tmax and Tmin
Wind (from NCEP/NCAR reanalysis)
Soil parameters: derived from Penn State State STATSGO in the U.S., FAO global soil map elsewhere.
Vegetation coverage from the University of Maryland 1-km Global Land Cover product (derived from AVHRR)
100šW
100šW
98šW
98šW
96šW
96šW
40šN 40šN
42šN 42šN
Missouri River Basin
Precipitation Station Locations
Temperature and Precipitation Data
Within the U.S.:•Precipitation adjusted for time-of-observation
•Precipitation re-scaled to match PRISM mean for 1961-90 (especially important in western U.S.
Precipitation and Temperature from gauge observations gridded to 1/8o
Avg. Station density:Area Km2/station
U.S. 700-1000
Canada 2500
Mexico 6000
Validation with Observed Runoff
Hydrographs of routed runoff show good correspondence with observed and naturalized flows.
Comparisons with Illinois Soil Moisture
19 observing stations are compared to the 17 1/8º modeled grid cells that contain the observation points.
Persistence
Moisture Level
Moisture Flux
Variability
Evaluation of Energy Forcings
Comparison with 4 SURFRAD Sites
• 3-minute observations aggregated to 3-hour
• Average Diurnal Cycle is for June, July, August 1996-99
• Peak underestimated 3-15% at each site (avg. 10% for all sites)
• Daily average within 10%, (avg. 2%)
Seasonal Soil Moisture Variation
•Shown is seasonal variation of soil moisture.•Top plot is scaled by the total soil pore volume.•Bottom plot is scaled by its dynamic range for 50-years.
Soil Moisture - Active Range
50-Year Soil Moisture Range Scaled by Annual Precipitation
Scale indicates level of hydrologic interaction of soil
column
Soil Moisture - Persistence
Persistence of soil moisture anomalies, based on the full 50+ year timeseries at each grid cell.
Persistence is generally seen where soil moisture interaction is high.
Variables
Data Availability
Monthly Average VariablesCurrently downloadable from www.hydro.washington.edu
Each file contains monthly data for 1950-2000
Avg. File Size (compressed netCDF): 120 MB
3-Hourly VariablesArchived at San Diego Supercomputer Center (link from www.hydro.washington.edu)
Each file contains 3-hourly data for one year
Avg. File Size (compressed netCDF): 200-450 MB
Daily VariablesArchived at San Diego Supercomputer center (link from www.hydro.washington.edu)
Each file contains daily average data for one year
Avg. File Size (compressed netCDF): 20-100 MB
Example Application 1: Hydrologic predictability over the Missouri River basin
Methods for Determining Runoff Predictability
• Indices Characterizing Sources of Predictability:
SOI – An index identifying ENSO phaseAO – An index of phase of the Arctic OscillationSM – Soil moistureSWE – Snow water equivalent
• Varying Lead Times between Initial Conditions (IC) and Forecast Runoff
• Only Use Indices in Persistence Mode
ForecastSeason
DJF
Initialization Dates for DJF Forecast
Dec 1Dec 1 Mar 1 Jun 1 Sep 1
Lead-0Lead-4 Lead-3 Lead -2 Lead 1
D J F M A M J J A S O N
Climate
Land
Methods 2
• Multiple linear regression used between IC and runoff
• Variance explained (r2) indicates level of predictability
• Variables introduced in order of how well indices represent current knowledge of state:
1. SOI/AO2. SWE3. SM
• Incremental predictability
r2SOI/AO
r2SWE
Runoff
SOI/AO SWE
Methods 3Test for Significant Predictability (r2) in 2 steps
Local Significance:
•Tested at each grid cell
•Accounts for temporal autocorrelation
•95% confidence level estimated
Field Significance (Livezey and Chen, 1983):
•Tests area showing local significance over entire basin
•Accounts for limited sample size, spatial correlation in both predictors and predictand
•95% confidence for field significance
Total Runoff Predictability
1.5
4.5
7.5
10.5
13.5
Lead,months
• Uses all 4 indices to predict runoff
• “X” no field significance
• Field significance is domain-wide measure
Predictability due to Climate Signals
• Predictors currently available
• Moderate levels of r2
• Greater influence in winter, in area and lead time
• Difficulty in long-lead persistence prediction with climate signals
Predictability due to Soil Moisture
•Widespread predictability at 0 lead (1½ month)
•Winter Runoff: little predictability where runoff is high
•Summer Runoff: limited predictability to 3 seasons
Example Application 2: Hydrologic predictability over the North American
Monsoon (NAMS) region
Exploratory Work on Teleconnections between SST and Soil Moisture
Sea surface temperature: Extended Reconstruction of Global Sea Surface Temperature data set based on COADS data. (1847-1997) developed by T.M. Smith and R.W. Reynolds, NCDC. The original data resolution is 2ºlongitude, 2 º latitude. It was interpolated into 0.5 º resolution (The ocean domain is chosen according to the Bin Yu and J.M. Wallace’s paper, 2000, J. Climate, 13, 2794-2800)
Soil Moisture: VIC retrospective land surface dataset (1950-1997). The original data with 1/8 degree resolution is aggregated into 0.5 º resolution.
Study Domain and Datasets
Maximum Positive Correlation Coefficient
• SST has significant positive correlation coefficient with soil moisture in most areas even with lead time more than 9 months.
• Southwestern United States shows higher correlation Coefficient (greater than 0.6) with SST than Mexico region.
• June shows larger area with higher coefficient than other months.
Predictability of Soil Moisture by SST First and Second PC
Southwestern US area shows highest predictability (the highest variance explained is about 0.45)
Predictability of Soil Moisture by Persistence
• Soil moisture shows significant persistence even in at 6-month lead time especially for June soil moisture.
• Mexican part of the domain also shows high persistence for June soil moisture
June Soil Moisture Predictability by Persistence and SST PCs
The highest variance explained is more than 90%. For June, over 40% of the variance is explained over most of the study domain, including Mexico.
June soil moisture predicted by Persistence and SST PCs
• Last December can explain 66.7% June soil moisture
LDAS Data
Predicted
Introducing SST PCs benefits long-time lead predictability (of June soil moisture), but no significant benefits for less than 6-month lead
time predictability.
SST and Persistence
Persistence
Region with 90% significant correlation between last winter’s JFM precipitation and JJAS precipitation (1965-1999) in given
Monsoon Region
JJAS monsoon precipitation with statistically significant correlation to previous winter’s (JFM) precipitation
by region
0
40
80
120
160
1950 1960 1970 1980 1990
Year
Pre
cip
itat
ion
(m
m)
0
50
100
150
200
250
1950 1960 1970 1980 1990
Year
Pre
cip
itat
ion
(m
m)
Monsoon West Monsoon North
The significant period is around
1970-1999.
15-year moving correlation between last winter JFM precipitation index and JJAS Monsoon precipitation
Monsoon West Monsoon North
Hypotheses
• For Monsoon West, the region for which monsoon precipitation is (negatively correlated with) the previous winter’s precipitation is the Southwestern U.S. (at least for the period 1965-1999).
• For Monsoon North, the region for which monsoon precipitation is (negatively correlated with) the previous winter’s precipitation is the Midwest and Southwestern U.S. (for the period 1970-1999).
• The land surface feedback mechanism could be: anomalous winter precipitation leads to anomalous spring soil moisture, hence lower early summer surface temperature, and weaker monsoon precipitation.
2. Examples b) Seasonal to interannual streamflow forecasting
climate model forecastmeteorological outputs
• ~1.9 degree resolution (T62)• monthly total P, avg T
Use 3 step approach: 1) statistical bias correction 2) downscaling3) hydrologic simulation
General Approach
hydrologic model inputs
streamflow, soil moisture,snowpack,runoff• 1/8-1/4 degree resolution
• daily P, Tmin, Tmax
Models: Global Spectral Model (GSM) ensemble forecasts from NCEP/EMC
• forecast ensembles available near beginning of each month, extend 6 months beginning in following month
• each month:• 210 ensemble members define GSM climatology for
monthly Ptot & Tavg• 20 ensemble members define GSM forecast
Models: VIC Hydrologic Model
domain slide
Example Flow Routing Network
One Way Coupling of GSM and VIC models
a) bias correction: climate model climatology observed climatologyb) spatial interpolation:
GSM (1.8-1.9 deg.) VIC (1/8 deg)c) temporal disaggregation (via resampling of observed patterns):
monthly daily
a. b. c.
0
5
10
15
20
25
30
0 1Probability
Te
mp
era
ture
TGSM
TOBS
GSM Regional Bias:a spatial example
Bias is removed at the monthly GSM-scale from the meteorological forecasts
(so 3rd column ~= 1st column)
GSM Regional Bias:
one cell example
For sample cell located over Ohio River basin, biases in monthly Ptot & Tavg are significant!
GSM Regional Bias:
one cell example
Bias: Developing a Correction
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
20 member forecast ensemble
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
from 1979 SSTsfrom 1980 SSTs
from 1981 SSTs
from 1999 SSTs
from current SSTs
(21 sets)10 member climatology ensembles
Bias: Developing a Correction
10
15
20
25
30
0 0.2 0.4 0.6 0.8 1
percentile (wrt 1979-99)
deg
C
GSM
Observed
July Tavg, for 1 GSM cell
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
-5
5
15
25
35
Mon 1 Mon 2 Mon 3 Mon 4 Mon 5 Mon 6
deg
C
1979 SSTsetc.
from 1999SSTs
10 member climatology ens.
* for each month, each GSM grid cell and variable
*
Downscaling Test
1. Start with GSM-scale monthly observed met data for 21 years
2. Downscale into a daily VIC-scale time series
3. Force hydrology model to produce streamflow
4. Is observed streamflow reproduced?
Simulations
Forecast Productsstreamflow soil moisture
runoffsnowpack
VIC model spin-upVIC forecast ensemble
climate forecast
information (from GSM)
VIC climatology ensemble
1-2 years back start of month 0 end of month 6
NCDC met. station obs. up to 2-4 months from
current
LDAS/other met. forcings
for remaining
spin-up
data sources
A B C
Columbia River Application
CRB
Initial Conditions
late-May SWE &water balance
CRB
Initial Conditions
(percentiles)
CRB: May forecast
hindcast“observed”
forecast
forecast medians
CRB May forecast
basin avg. soil moisture
CRB May Forecast
Streamflow
CRBMay Forecast
cumulative flow averages
forecastmedians
2. Examples c) Hydrologic effects of climate change
Climate Scenarios
Global climate simulations, next ~100 yrs
Downscaling
Delta Precip,Temp
HydrologicModel (VIC)
Natural Streamflow
ReservoirModel
DamReleases,Regulated
Streamflow
PerformanceMeasures
Reliability of System Objectives
Overview of ColSim Reservoir Model
Physical Systemof Damsand Reservoirs
Reservoir Operating Policies
Reservoir StorageRegulated StreamflowFlood ControlEnergy ProductionIrrigation ConsumptionStreamflow Augmentation
0100000200000300000400000500000600000700000800000900000
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
Flow
(cfs
)
Streamflow Time Series
Dam Operations in ColSim
Storage Dams
Run-of-River Dams
0
100000
200000
300000
400000
500000
600000
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Month
Avg
Str
eam
flow
(cf
s)
0
100000
200000
300000
400000
500000
600000
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Month
Avg
Str
eam
flow
(cf
s) Virgin Regulated
Flow In=Flow out + Energy
H
GCM grid mesh over western U.S. (NCAR/DOE Parallel Climate Model at ~ 2.8 degrees lat-long)
ACPI: PCM-climate change scenarios, historic simulation v air temperature observations
ACPI: PCM-climate change scenarios, historic simulation v precipitation observations
Climate Change Scenarios
Historical B06.22 (greenhouse CO2+aerosols forcing) 1870-2000
Climate Control B06.45 (CO2+aerosols at 1995 levels) 1995-2048
Climate Change B06.44 (BAU6, future scenario forcing) 1995-2099 Climate Change B06.46 (BAU6, future scenario forcing) 1995-2099 Climate Change B06.47 (BAU6, future scenario forcing) 1995-2099
Climate Control B06.45 derived-subset 1995-2015
Climate Change B06.44 derived-subset 2040-2060
PCM Simulations (~ 3 degrees lat-long)
PNNL Regional Climate Model (RCM) Simulations (~ ¾ degree lat-long)
Future streamflows
• 3 ensembles averaged
• summarized into 3 periods;» Period 1 2010 - 2039
» Period 2 2040 - 2070
» Period 3 2070 - 2098
Bias Correction and Downscaling Approach
climate model scenariometeorological outputs
hydrologic model inputs
snowpackrunoffstreamflow
• 1/8-1/4 degree resolution• daily P, Tmin, Tmax
•2.8 (T42)/0.5 degree resolution•monthly total P, avg. T
Bias Correction
from NCDC observations
from PCM historical runraw climate scenario
bias-corrected climate scenario
month mmonth m
Note: future scenario temperature trend (relative to control run) removed before, and replaced after, bias-correction step.
Downscaling
observed mean fields
(1/8-1/4 degree)
monthly PCManomaly (T42)
VIC-scale monthly simulation
interpolated to VIC scale
Inflow
Run of River Reservoirs (inflow=outflow + energy)
Inflow
Inflow
Inflow
Inflow
Inflow
Storage ReservoirsReleases Depend on:•Storage and Inflow•Rule Curves (streamflow forecasts)•Flood Control Requirements•Energy Requirements•Minimum Flow Requirements•System Flow Requirements
System Checkpoint
Consumptive use
Consumptive use
Inflow +
ColSim
Regional Climate Model (RCM) grid and hydrologic model domains
PCM Business-as-Usual scenarios
Columbia River Basin(Basin Averages)
control (2000-2048)
historical (1950-99)
BAU 3-run average
RCM Business-as-Usual scenarios
Columbia River Basin(Basin Averages)
control (2000-2048)
historical (1950-99)
PCM BAU B06.44
RCM BAU B06.44
PCMBusiness-As-Usual
Mean Monthly
Hydrographs
Columbia River Basin@ The Dalles, OR
1 month 12 1 month 12
CRB Operation Alternative 1 (early refill)
15,000,000
20,000,000
25,000,000
30,000,000
35,000,000
40,000,000
45,000,000
50,000,000
55,000,000
O N D J F M A M J J A S
To
tal
En
d o
f M
on
th S
yste
m S
tora
ge
(acr
e-fe
et)
Max Storage
Control
Base Climate Change
Change (Alt. 1)
Dead Pool
CRB Operation Alternative 2 (reduce flood storage by 20%)
15,000,000
20,000,000
25,000,000
30,000,000
35,000,000
40,000,000
45,000,000
50,000,000
55,000,000
O N D J F M A M J J A S
En
d o
f M
on
th T
ota
l S
ys
tem
Sto
rag
e (
ac
re-f
ee
t)
Max Storage
Control
Base Climate Change
Change (Alt. 2)
Dead Pool
Columbia River Basin Water Resource Sensitivity to PCM Climate Change Scenarios
0%
20%
40%
60%
80%
100%
120%
Portland-Vancouver
Spring FloodControl
Reliability
Portland-Vancouver
Winter FloodControl
Reliability
Autumn FirmPower
Reliability(November)
% of ControlHydropower
Revenues
McNaryInstream
TargetReliability
(April-August)
Middle SnakeAgriculturalWithdrawalReliability
Grand CouleeRecreationReliability
Rel
iab
ility
(%
, mo
nth
ly b
ased
)
Control
Period 1
Period 2
Period 3
RCM
A v e r a g e M o n t h ly D e f i c i t a t t h e M c N a r y D a m T a r g e t ( c f s )
M o n th ly R e l i a b i l i t y a t t h e M c N a r y D a m T a r g e t
Per
iod
1
-
20,000
40,000
60,000
80,000
100,000
120,000
Apr May Jun Jul Aug
Control
Current Operations
Refill 2 w eeks earlier
Refill 1 month earlier
Per
iod
1
Per
iod
2
Per
iod
3
3) Weak links and research opportunities
1) How can macroscale hydrology models be evaluated/diagnosed at the spatial scales at which they are designed to be applied? What is the roll of large scale field experiments, and how should the next generation be designed?
2) Are there “interesting” research issues in climate change impact assessment (where most uncertainties are in the climate forcings) vs feedbacks (roll of the land surface in climate)? Planning under uncertainty?
3) Applications issues: a) We don’t have a good pathway for infusing science advances into operational model improvements; and b) Lack of skill score history for hydrologic prediction – no basis for audit
4) What is the role of remote sensing and data assimilation in large scale hydrologic modeling and prediction?