An experimental seasonal hydrologic forecast system for the western U.S.

1
An experimental seasonal hydrologic forecast
system for the western U.S.

• Dennis P. Lettenmaier
• Department of Civil and Environmental Engineering
  
• University of Washington
• for
• Cincinnati Earth Systems Science Seminar Series
• and
• Advanced Environmental Seminar Series
• University of Cincinatti
• Apr 30, 2004

2
Outline of this talk

• Introduction seasonal hydrologic forecasting
  historical development current operational
  methods
• Modeling framework and implementation for
  long-lead seasonal streamflow forecasting
• Estimating the hydrologic initial conditions
• Approaches based on ensemble climate prediction
• Data assimilation example using the MODIS snow
  cover product
• Westwide streamflow forecast system experience
  in winters 2002-3 and 2003-4
• Conclusions and unsolved problems

3
Introduction seasonal hydrologic forecasting
historical development
4
Introduction Seasonal Hydrologic Forecasts
water management hydropower irrigation flood
control water supply fisheries recreation
navigation water quality
5
Introduction Hydrologic prediction long
history
Snow water content on April 1
should add my personal pics of - snow
sampling snotel sites (and scan in curve method
figure)
SNOTEL network
NRCS SNOTEL Network
McLean, D.A., 1948 Western Snow Conf.
April to August runoff
6
Introduction Hydrologic prediction and NWS

• NWS River Forecast Center (RFC)
• approach rainfall-runoff modeling
• (i.e., NWS River Forecast System,
• Anderson, 1973
• offspring of Stanford Watershed Model, Crawford
  Linsley, 1966)
• Ensemble Streamflow Prediction (ESP)
• used for shorter lead predictions
• used for longer lead predictions
• The RFC final seasonal
• forecasts also incorporate
• NRCS results.

Neither NWS or NRCS objectively use contemporary
climate forecasts
7
Introduction Current Operational Methods
A primary seasonal forecasting product is the
runoff volume forecast
UPPER COLUMBIA BASINS




Mar-04MIDMN W A T E
R S U P P L Y F O R E C A S T S

FORECAST RUNOFF
AVERAGE RO PREV STREAM AND STATION
PERIOD PROBABLE MAXIMUM MINIMUM
RUNOFF PERIOD COLUMBIA RIVER
MICA RESERVOIR INFLOW, BC JAN-JUL
9100.0 95 10400.0 108 7770.0 81 9619.
1049
FEB-SEP 12200.0 94 13600.0 105 10900.0
84 12960. 13000
APR-SEP 11800.0 94 13200.0
106 10500.0 84 12500. 12590
REVELSTOKE, BC JAN-JUL
13200.0 95 14400.0 104 12100.0 87 13880.
15070 ARROW LAKES INFLOW
JAN-JUL 19300.0 92 21300.0 102 17200.0
82 20960. 21970
FEB-SEP 24400.0 92 26500.0
100 22300.0 84 26460. 26070
APR-SEP
23100.0 92 25100.0 100 21000.0 84 25110.
24950 BIRCHBANK, BC (1)
JAN-JUL 34900.0 90 40100.0 103 29600.0
76 38930. 42950
APR-SEP 39000.0 90 44200.0
102 33700.0 77 43500. 46150
GRAND COULEE, WA (1) JAN-JUL
55500.0 88 64000.0 102 46900.0 75 62900.
68020
APR-SEP 56400.0 88 65000.0 102 47900.0
75 63990. 68220 ROCK
ISLAND DAM BLO, WA (1) JAN-JUL 61000.0 89
70400.0 102 51600.0 75 68910. 74830
APR-SEP
61500.0 88 71000.0 102 52100.0 75 69540.
74300 THE DALLES NR, OR (1)
JAN-JUL 92000.0 86 106000.0 99 78400.0
73 107300. 103800
APR-AUG 79800.0 86 93400.0
100 66200.0 71 93090. 93800
APR-SEP
84600.0 86 98200.0 100 71000.0 72 98650.
98080
8
Introduction Research Rationale
Are current seasonal hydrologic forecasts all
that they can be? How can ongoing research on
land-atmosphere interactions help to improve
seasonal streamflow forecasts in the western U.S.?

• Potential sources of improvement since inception
  of regression/ESP methods
• operational seasonal climate forecasts
  (model-based and otherwise)
• greater availability of station data
• computing
• new satellite-based products (primarily snow
  cover)
• distributed, physical hydrologic modeling for
  macroscale regions

9
Modeling framework and implementation
10
Variable Infiltration Capacity (VIC) Model
11
(No Transcript)
12
Estimating the initial hydrologic conditions
13
Estimating relative impact of initial conditions
and forecast accuracy
Retrospective ESP-type simulations can shed light
on the relative value of initial conditions to a
given forecast application.
14
Initial Conditions Balancing IC and forecast
accuracy
Columbia R. Basin
fcst more impt
ICs more impt
Rio Grande R. Basin
RMSE (perfect IC, uncertain fcst) RMSE (perfect
fcst, uncertain IC)
RE
15
Initial Conditions Hydrologic Simulations
start of month 0
end of mon 6-12
1-2 years back
forecast ensemble(s)
model spin-up
initial conditions
climatology ensemble
NCDC met. station obs. up to 2-4 months from
current 2000-3000 stations in west
LDAS/other real-time met. forcings for remaining
spin-up 300-400 stations in west
climate forecast information
data sources
Forecast Products streamflow soil
moisture runoff snowpack derived products e.g.,
reservoir system forecasts
obs snow state information (eg, SNOTEL)
16
Initial Conditions estimating run-up conditions
Problem met. data availability in 3 months
prior to forecast has only a tenth of long term
stations used to calibrate model S
olution use interpolated monthly index station
precip percentiles and temperature anomalies to
extract values from higher quality retrospective
forcing data -- then disaggregate using daily
index station signal.
sparse station network in real-time
dense station network for model calibration
17
Initial Conditions snow state assimilation
Problem sparse station spin-up period incurs
some systematic errors, but snow state estimation
is critical Solution use SWE anomaly
observations (from the 600 station USDA/NRCS
SNOTEL network and a dozen ASP stations in BC,
Canada) to adjust snow state at the forecast
start date
18
Initial Conditions Initial snow state
assimilation

• Assimilation Method
• weight station OBS influence over VIC cell based
  on distance and elevation difference
• number of stations influencing a given cell
  depends on specified influence distances

• distances fit OBS weighting increased
  throughout season
• OBS anomalies applied to VIC long term means,
  combined with VIC-simulated SWE
• adjustment specific to each VIC snow band

19
Initial Conditions Snow state assimilation
SWE state differences due to assimilation of
SNOTEL/ASP observations, Feb. 25, 2004
20
Initial Conditions final product
Snow Water Equivalent (SWE) and Soil Moisture
21
Description and Products of MODIS Updated
Forecasts
local scale weather inputs
Initial Conditions soil moisture, snowpack
Hydrologic model spin up
Hydrologic simulation
Ensemble Forecast streamflow, soil moisture,
snowpack, runoff
NCDC met. station obs. up to 2-4 months from
current
LDAS/other real-time met. forcings for remaining
spin-up
25th Day of Month 0
1-2 years back
End of Month 6 - 12
Change in Snowcover as a Result of MODIS Update
for April 1, 2004 Forecast
Snowcover before MODIS update
Snowcover after MODIS update
22
Result of MODIS Update on Streamflow and
Storage April 1, 2004 Forecast
Unadjusted
MODIS Unadjusted
MODIS Adjusted
23
Unadjusted vs adjusted forecast errors,
2001-2003, for reservoir inflow volumes (left
plot) and reservoir storage (right)
24
Forecast approaches based on ensemble climate
prediction
25
Climate Forecasts Seasonal prediction

• Climate prediction has markedly advanced in the
  last several decades
• better monitoring of oceans and atmosphere
• deeper understanding of ocean-atmosphere
  teleconnections
• Monthly / seasonal climate forecasting has become
  operational at a number of research centers

Circulation Features
Sea Surface Temps
Typical climate model spatial resolution
e.g
El Nino / La Nina
26
Climate Forecasts Scale Issues
27
Overview Hydrologic Forecast Approach
28
Overview 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

29
Overview climate model forecast processing
sequence
a) bias correction climate model
climatology ? observed climatology b) spatial
interpolation e.g., GSM (1.8-1.9 deg.) ? VIC
(1/8 deg) c) temporal disaggregation (via
resampling of observed patterns) monthly ?
daily
30
Climate Forecasts forecast use challenges
31
Climate Forecasts Bias Correction

• numerous methods of downscaling and/or bias
  correction exist
• the relatively simple one weve settled on
  requires a sufficient retrospective climate model
  climatology, e.g.,
• NCEP hindcast ensemble climatology, 21 years X
  10 member
• NSIPP-1 AMIP run climatology, gt 50 years, 9
  member

32
Climate Forecasts Operational Products
33
Climate Forecasts skill deficit problem

• Retrospective skill analysis for NCEP-GSM found
• bias-correction sufficient to put T P forecasts
  into plausible range w.r.t. observations
• where/when model shows good rank correlation with
  observations, this can produce a better (than
  ESP) streamflow forecast
• sweet spots are few and far between
• -- strong ENSO-anomaly conditions
• -- ENSO-sensitive regions

34
Skill Assessment Retrospective analysis
tercile prediction skill of GSM ensemble forecast
averages, JAN FCST
35
Westwide streamflow forecast system experience
in winters 2002-3 and 2003-4
36
Forecasting Project Background
1998-9
Ohio R. basin w/ COE First tried climate
model-based seasonal forecasts on experimental
(retrospective) basis
Eastern US First attempted real-time seasonal
forecasts during drought condition in
southeastern states -- results published in
Wood et al. (2001), JGR
2000
Columbia R. basin Implemented approach during
the PNW drought, again using climate model based
approach
2001
Western US Retrospective analysis of forecasts
over larger domain (for one climate model and for
ESP)
2002
Columbia R. basin New funding for
pseudo-operational implementation for western
US began with pilot project in CRB
2003
(Funding from NASA NSIPP IRI/ARCS NOAA
GCIP/GAPP)
Western US expanded to western U.S domain for
real-time forecasts working to improve and
evaluate methods each forecast cycle
2004
37
Current season forecasts
September/October 2003 Soil Moisture
38
Current season forecasts
November 25 Snow Water Equivalent (SWE) and Soil
Moisture
39
Current season forecasts
December 25 Snow Water Equivalent (SWE) and Soil
Moisture
40
Current season forecasts
January 25 Snow Water Equivalent (SWE) and Soil
Moisture
41
Current season forecasts
February 25 Snow Water Equivalent (SWE) and Soil
Moisture
42
2003-04 Spatial Forecasts
43
Current season forecasts, locations
44
Overview Streamflow Forecasts
hydrographs
targeted statistics
raw ensemble data
45
Streamflow products
targeted statistics
hydrographs
46
Ensemble forecasts official CPC product
47
Comparison with RFC regression forecast for
Columbia River at the Dalles
UW forecasts made on 25th of each month RFC
forecasts made several times monthly 1st,
mid-month, late (UW ESP unconditional forecasts
shown)
UW
RFC
48
Some obstacles and opportunities in hydrological
application of climate information

• The one model problem
• Calibration and basin scale (post-processing as
  an alternative to calibration)
• The value of visualization
• Opportunities to utilize non-traditional data
  (e.g. remote sensing)

For more information www.hydro.washington.edu/Le
ttenmaier/Projects/fcst/