Post on 20-Jan-2016
transcript
Experimental real-time seasonal hydrologic nowcasting and forecasting in the western U.S.
Andy Wood
Department of Civil and Environmental Engineering
SeminarUniversity of Arizona
Department of Hydrology and Water Resources
April 27, 2005
Outline
Introduction to seasonal forecasting and forecasting systembackgroundclimate forecastsVIC model spin-up
index station approachSNOTEL assimilation
Selected results for 2004 and current forecast season
Ongoing work (hydrologic nowcasting & …)
The importance of Seasonal Hydrologic Forecasting
water management hydropower
irrigationflood controlwater supply
fisheriesrecreationnavigation
water quality
Aug Dec Apr
Res
ervo
ir S
tora
ge
Aug
Introduction: A definition of useful terms
A short poem by someone not primarily known as a poet
The Unknown
As we know,There are known knowns.There are things we know we know.
We also knowThere are known unknowns.That is to sayWe know there are some thingsWe know we do not know.
But there are also unknown unknowns,The ones we don't knowWe don't know.
Feb. 12, 2002 Department of Defense news briefing Donald Rumsfeld
(Disclaimer: the use of this ‘poem’ does not represent a comment on US military policy…)
Apr 1 SM
Apr 1 SWE
Apr-Sep climate
Seasonal Forecasting: What do we know and when do we know it?
The primary factors determining summer runoff are generally: SWE at the start of melt (e.g., April 1) Soil moisture at the start of the forecasted runoff period (e.g, April 1) April-September Climate
the
Do
n
Ru
msf
eld
sca
le
Sn
ow w
ater
con
ten
t on
Ap
ril
1
April to August runoff
McLean, D.A., 1948 Western Snow Conf.
SNOTEL Network
Introduction: Hydrologic prediction and the NRCS
PNW
Technical Advances related to Hydrologic Forecasting
1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s
snow survey / graphical forecasts /
index methods / i.e., regression
computing in water
resources
aerial snow
surveys
SNOTEL network
ESP method
snow cats
conceptualhydrologic
models
Introduction: Hydrologic prediction and ESP
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
Currently, some western RFCs and NRCS coordinate their seasonal forecasts, using mostly statistical methods.
ICsSpin-up Forecast
obs
recently observedmeteorological data
ensemble of met. datato generate forecast
ESP forecast
hydrologicstate
Technical Advances related to Hydrologic Forecasting
1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s
snow survey / graphical forecasts /
index methods / i.e., regression
computing in water
resources
satelliteimagery
aerial snow
surveys desktopcomputing
SNOTEL network
ESP method
ENSO / seasonal climate
forecasts
snow catsInternet / real-time
data
conceptualhydrologic
models
physicalhydrologic
models
UW Forecast System Overview
Forecast System Overview
NCDC met. station obs.
up to 2-4 months from
current
local scale (1/8 degree) weather inputs
soil moisturesnowpack
Hydrologic model spin up
SNOTEL
Update
streamflow, soil moisture, snow water equivalent, runoff
25th Day, Month 01-2 years back
LDAS/other real-time
met. forcings for spin-up
gap
Hydrologic forecast simulation
Month 6 - 12
INITIAL STATE
SNOTEL/ MODIS*Update
ensemble forecasts ESP traces (40) CPC-based outlook (13) NCEP CFS ensemble (20) NSIPP-1 ensemble (9)
* experimental, not yet in real-time product
Introduction: UW Experimental Hydrologic Forecasting
Soil MoistureInitial
Condition
SnowpackInitial Condition
Introduction: UW Experimental Hydrologic Forecasting
VIC model runoff is routed to streamflow gages, and verified against observations
Introduction: UW Experimental Hydrologic Forecasting
targeted statistics e.g., runoff volumes
monthly hydrographs
Introduction: UW Experimental Hydrologic Forecasting
SWE Soil MoistureRunoffPrecip Temp
Mar-05
Apr-05
May-05
Outline
Introduction to seasonal forecasting and forecasting systembackgroundclimate forecastsVIC model spin-up
SNOTEL assimilation
Selected results for 2004 and current forecast season
Ongoing work (hydrologic nowcasting & …)
Climate Forecasts: Operational Products
Climate Forecasts: Use in UW forecast system
ESP
ENSO/PDO
ENSO
CPC Official Outlooks
Coupled Forecast
Model (CFS)
CAS
OCN
SMLR
CCA
CA
NSIPP-1 dynamical
model
VIC Hydrology Model
NOAA
NASA
UW
Seasonal Climate Forecast Data Sources
Climate Forecasts: Spatial Scale Issues
Seattle
Climate Forecasts: Bias Issue (prior NCEP model)
Sample GSM cell located over Ohio River basin
obs prcp GSM prcp
obs temp GSM temp
JULY
Regional Bias: spatial example
obsGSM
Climate Forecasts: Bias Correction Scheme
from COOP observations
from GSM climatological runsraw GSM forecast scenario
bias-corrected forecast scenario
month mmonth m
Climate Forecasts: CPC Seasonal Outlooks
e.g., precipitation
spatial unit for raw forecasts is the Climate Division (102 for U.S.)
CDFs defined by 13 percentile values (0.025 - 0.975) for P and T are given
Climate Forecasts: CPC Seasonal Outlook Use
Climate Forecasts: CPC Seasonal Outlook Use
probabilities => anomalies
precipitation
VIC initial state: Merging of SNOTEL obs with model SWE
The pattern of observed SWE values, which are merged with the forecast initial conditions, is usually in pretty good agreement with the VIC simulated snow state.
The PNW currently has very low snowpack, while the Southwest and California have record high snowpacks.
VIC initial state: SNOTEL 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
spatial weighting function
elevationweightingfunction
SNOTEL/ASP
VIC cell
• 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
VIC initial state: SNOTEL assimilation
April 25, 2004
Outline
Introduction to seasonal forecasting and forecasting systembackgroundclimate forecastsVIC model spin-up
SNOTEL assimilation
Selected results for 2004 and current forecast season
Ongoing work (hydrologic nowcasting & …)
Final comments
Results for Winter 2003-04: initial conditionsSoil Moisture and Snow Water Equivalent (SWE)
Results for Winter 2003-04: initial conditionsSoil Moisture and Snow Water Equivalent (SWE)
Results for Winter 2003-04: initial conditionsSoil Moisture and Snow Water Equivalent (SWE)
Results for Winter 2003-04: initial conditionsSoil Moisture and Snow Water Equivalent (SWE)
Results for Winter 2003-04: initial conditionsCPC estimates of seasonal precipitation and temperature
very dry hot
March
Results for Winter 2003-04: initial conditionsSoil Moisture and Snow Water Equivalent (SWE)
Results for Winter 2003-04: streamflow hydrographs
By Fall, slightly low flows were anticipated
By winter, moderate deficits were forecasted
Results for Winter 2003-04: volume runoff forecasts
UPPER HUMBOLDT RIVER BASIN
Streamflow Forecasts - May 1, 2003
<==== Drier === Future Conditions === Wetter ====>
Forecast Pt ============ Chance of Exceeding * ===========
Forecast 90% 70% 50% (Most Prob) 30% 10% 30 Yr Avg
Period (1000AF) (1000AF) (1000AF) (% AVG.) (1000AF) (1000AF) (1000AF)
MARY'S R nr Deeth, Nv
APR-JUL 12.3 18.7 23 59 27 34 39
MAY-JUL 4.5 11.3 16.0 55 21 28 29
LAMOILLE CK nr Lamoille, Nv
APR-JUL 13.7 17.4 20 67 23 26 30
MAY-JUL 11.6 15.4 18.0 64 21 24 28
N F HUMBOLDT R at Devils Gate
APR-JUL 5.1 11.0 15.0 44 19.0 25 34
MAY-JUL 1.7 7.2 11.0 50 14.8 20 22
Results for Winter 2003-04: volume runoff forecastsComparison with NWRFC forecast for Columbia River at the Dalles, OR
UW forecasts made on 25th of each month
RFC forecasts madeseveral times monthly:1st, mid-month, late
(UW’sESP unconditional and CPC forecasts shown)
UW RFC
79%obs
Results for Winter 2003-04: volume runoff forecastsComparison with RFC forecast for Feather River, CA
UW forecasts made on 25th of each month
RFC forecasts madeon 1st of month
(UW’sESP unconditional forecasts shown)
UW
RFC
78%obs
Results for Winter 2003-04: volume forecasts
for a sample of PNW locations
JAN 1, 2004 Summer Runoff Volume Forecasts compared to OBS
50
60
70
80
90
100
110
MIC
AA
DU
NC
A
LIB
BY
HH
OR
S
JLA
KE
LG
RA
N
DW
OR
S
DA
LL
E
pe
rce
nt
of
av
era
ge
OBS %avgRFCUW ESP
Results for Winter 2003-04: volume forecasts
for a sample of PNW locations
APR 1, 2004 Summer Runoff Volume Forecasts compared to OBS
50
60
70
80
90
100
110
MIC
AA
DU
NC
A
LIB
BY
HH
OR
S
JLA
KE
LGR
AN
DW
OR
S
DA
LLE
per
cen
t o
f av
erag
e
OBS %avgRFCUW ESP
WY2005, Dec. 1 hydrologic conditions
WY2005, Dec. 1 hydrologic conditions
WY2005, Jan. 1 hydrologic conditions
WY2005, Feb. 1 hydrologic conditions
PNW in crisis?: Headlines from February, 1977
Comparison: Columbia R. basin upstream of The Dalles, OR
WY2005
WY1977
WY2005
WY1977
Results: WY2005, Feb. 1 streamflow forecasts
SNOTEL / Env. Canada ASP network is a valuable source of snowpack information.
The observed SWE values, which are merged with the forecast initial conditions, were in good agreement with the simulated snow state.
98%
80%85%1977: 55%
WY2005, Mar. 1 hydrologic conditions
WY2005, Apr. 1 hydrologic conditions
4/15 ESP forecast: WY2005 Precip, Temp
Yakima R. Basin near
Parker, WA
plots show current + forecast(ESP; min, max and quartiles)
against historical 1971-2000min, max and quartiles
Apr-Sep % of avg3/15 4/15 chg
max 61 570.75 46 500.50 41 450.25 39 42min 31 39
4/15 ESP forecast: WY2005 SM, SWE, RO
Ongoing Work: Involvement with Operational Groups
A major goal of this work is to transition successful approaches intooperational settings.
Progress on this front:
Involvement with NRCS National Water and Climate Center (Tom Pagano!)
Involvement with NOAA Climate Prediction Center (CPC) e.g., related to Drought Outlook
Outreach to PNW water and power community through the UW Climate Impacts Group
e.g., to WA State Executive Water Emergency Committee
Outline
Introduction to seasonal forecasting and forecasting systembackgroundclimate forecastsVIC model spin-up
SNOTEL assimilation
Selected results for 2004 and current forecast season
Ongoing work (hydrologic nowcasting & …)
Ongoing work (nowcast & …)
snow state update using MODIS
multi-model (land-surface in addition to climate)
west-wide expansionmore forecast pointsmore comprehensive outputsreorganized web-sitemore verificationmore involvement with operational groups
surface water monitor
Ongoing Work: use of remote sensing
Trenchant remarks on remote imaging of the land surface
Oh my goodness gracious,What you can buy off the InternetIn terms of overhead photography!
June 9, 2001, following European trip
Satellite products with potential for hydrology (short list):MODIS snow-covered areaAMSR-E passive microwave-based SWEvarious microwave based soil moisture products
Ongoing work: MODIS snow cover assimilation (Snake R. trial)
Snowcover BEFORE update
Snowcover AFTER update
MODIS update for April 1, 2004 Forecast
snowadded
removed
Ongoing work: Rationale for Multi-model forecast framework
Single-IC ensemble forecast:
early in seasonal forecast season, climate ensemble spread is large
errors in forecast mainly due to climate forecast errors
ensemblemember
ensemblemean
OBS
Ongoing work: Rationale for Multi-model forecast framework
Single-IC ensemble forecast:
late in seasonal forecast season, climate ensemble is nearly deterministic
errors in forecast mainly due to IC errors
ensemblemember
ensemblemean
OBS
dailyupdates
1-2 day lag
soil moisture& SWEpercentiles
½ degreeresolution
archive from1915-current
uses ~2130index stns
Ongoing Work: UW SW Monitor
trends:1 week2 week1 month
Ongoing Work: UW SW Monitor
Archive from 1915-current
current conditions are a productof the same simulation (samemethods, ~same stations) ashistorical conditions
allows comparison of current conditions with historical ones
can navigate by month or year
Ongoing Work: UW SW Monitor
Ongoing Work: UW SW Monitor
Yakima R. Basin near
Parkr, WA
Ongoing Work:UW SW Monitor
Yakima R. Basin near
Parkr, WA
why the high soil moisture percentiles?
appears to have been relatively cold in last several weeks
Ongoing Work: UW SW Monitor
soil moisture actually decreasing, but relative to normal conditions at this time, percentile still high
Ongoing Work:UW SW Monitor
runoff is nearer to normal than soil moisture
Final Thought
Words of inspiration for researchers everywhere
A trained ape can know an awful lotOf what is going on in this world,Just by punching on his mouseFor a relatively modest cost!
June 9, 2001, following European trip
END
Framework: Downscaling CPC outlooks
downscaling uses Shaake Shuffle (Clark et al., J. of Hydrometeorology, Feb. 2004) to assemble monthly forecast timeseries from CPC percentile values
VIC model spinup methods: index stations
Example for daily precipitation
Index stn pcp pcp percentilegridded to 1/8
degree
1/8 degree dense station monthly pcp distribution(N years for each 1/8 degree grid cell)
1/8 degree pcpdisagg. to dailyusing interpolated daily fractions from index stations
monthly
0 1 0 1
Ongoing Work: UW SW Monitor
1920s 1990s
Average Flow, Columbia R. at The Dalles, OR
0
100000
200000
300000
400000
500000
600000
700000
jan feb mar apr may jun jul aug sep oct nov dec
cfs
coop avg
coop stdev
raw cpc avg
raw cpc stdev
cpc bc avg
cpc bc stdev
Results: CPC-based flow w.r.t. UW obs dataset
Answer:
YES, with help from bias-correction..........(but)
mean
std dev
Average flow, Sacramento R. (input to Shasta Reservoir)
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
jan feb mar apr may jun jul aug sep oct nov dec
cfs
coop avg
coop stdev
cpc raw avg
cpc raw stdev
cpc bc avg
cpc bc stdev
Results: CPC-based flow w.r.t. UW obs dataset
Additional examples show similar results
Mean pretty well reproduced; variability improved
mean
std dev
Results: CPC temp/precip w.r.t. UW obs dataset
based on 1960-99
Results: CPC temp/precip w.r.t. UW obs dataset
based on 1960-99
0
500
1000
1500
2000
2500
3000
3500
19
72
19
73
19
74
19
75
19
76
19
77
19
78
19
79
19
80
19
81
KA
F/M
ON
OBS VIC biascorr VIC raw
Statistical bias correction can dramatically improve streamflow simulations for use with reservoir models.
Natural Flow in the Snake River at Milner