CBRFC WFO CallJune16
Outline
• Survey Results (Kevin – 5 min)• CBRFC Summer Development Plan (Michelle –
10 min)• CHPS update (Andy, John – 10 min)• ET project update (Mike – 10 min?)• June 5-10 event post-mortem (Various – 30
min)• Other topics?
SH/MIC Survey Results• N=6 (offices participating: Las Vegas, Flagstaff, Tucson, Grand
Junction, and Boulder)• Rating of CBRFC products and services:
– All offices value flood forecasts, flash flood support, and daily forecasts
– Northern offices valued water supply, peak flow forecasts– Drought information and recreation forecasts less valuable
• Key comments:– Request QPF on par with CNRFC’s QPF (Las Vegas)– Improved FFG (Grand Junction, Boulder)– Ability to specify duration for precipitation analysis (Grand Junction)– Mixed comments on quality of streamflow forecasts; positive
feedback on flooding event support and forecast quality (Flagstaff). Room for improvement on forecasts during spring melt (Grand Junction) and data sparse basins (Meadow Valley Wash and Muddy River in Vegas HSA)
CBRFC Development Plan
• Improve daily operations: – plot previous forecasts– revisit shifts to allow for forecasts to be issued quicker– Post-mortem study on June 2010 event
• Water Supply: – peak flow re-analysis– Improvements to weekly ESP web display– Study to relate snow and runoff for small basins– Revamp pub– Documentation (stick diagrams, IFPcompanion basin information)
• Distributed Model: – Work with WFO TWC over summer to evaluate FSR (FFG)
CBRFC Development plan - cont
• Dam failure program: – integrate EAPs into dam failure web site– training.
• Outreach: – Open house (Aug 18)– Utah User Engagement workshop (Aug 2)– visit WFO Riverton (July)– Arizona Hydrologic Symposium booth (September)
CBRFC Development plan - cont
• Improve web services– Web speed– QC web page– Web services USU and CLDSS– Precipitation display (more flexibility to choose dates to accumulate over, etc.)
• Improve products and services for USBR• CHPS
– HEFS evaluation– Revisit CPC consolidation forecast– Begin systematic daily runs of CHPS over sub-area
• Calibrations
CHPS (John L, Andy W)
• CHPS – Community Hydrologic Prediction System– The whole system, ie data collection, quality
control, future forcing creation– Includes FEWS
• FEWS – Flood Early Warning System– The hydrologic modelling software– Within CHPS– Replaces NWSRFS
Current IFP interface with NWSRFS
New IFD interface for FEWS
Highlights of new features• More logging of system• Easier navigation of model plots• Better integration of segment information• Ability to view all data in system
Gridded data into FEWS QPF example
Gridded Temperatures and Freezing Levels will be imported too
• Ability to provide Stand Alone systems.• Ability to add new models to the system.• Facilitate collaborations with academia, and
research-to-operations implementations.• International hydrologic community.
Other things to come with CHPS
Forecasting ETrc across NWS Western Region
(Mike Hobbins)
ET-related work in NWS:WFOs, RFCs, WR-SSD, ABRFC…
Calls for standardized ET, Ep, or ETrc output on www from forecasters, RFCs, extension agents, and within WR:
Colorado Basin RFC:• goal to improve RFC simulations and forecasts.
WFOs’ goals are often to provide ET grids to satisfy AG users:• Great Falls WFO (MT) experimenting with Penman Ep.• Pendleton WFO (ne OR) publishing Kimberly Penman ETrc,• Hanford WFO (SJ valley, CA) publishing Penman-Monteith ETrc.• Sacramento WFO (int. n CA) publishing Penman-Monteith ETrc,
working with UC Davis and California DWR.• Arizona WFOs All AZ MICs agreed to host ET forecasts on their www
Western Region:• regionwide team from various WFOs to create standardized ET webpage, primarily for ag users,• wants ET knowledge-base in-house,• MSD wants standardized procedure by spring 2010,• WR wants to coordinate their efforts with CBRFC ET work and goals.
Arkansas Basin RFC:• calculates observed Ep at selected METAR sites using MAPE preprocessor and associated SYNTRAN program. Each basin's calculation is related to one of these sites, then modified by Ep adjustment curve.
ET = evapotranspirationEp = potential evaporation
ETrc = reference crop evapotranspiration
ETrc across Western Region:RTMA-driven Penman-Monteith ETrc
asatnrc eeUTU
QU
ET
222
9.0
34.0134.01
• in Penman-Monteith, ETrc is a specific case of ET calculated for specific biological and physical conditions (reference crop):
• hypothetical, well-watered crop of height h = 0.12 m, rs = 70 sec/m, a = 0.23,
• Uz, T, and ea data from 2-m height,
• ra implicitly specified as 208/U2 sec/m.
• international standard (FAO-56).
• very commonly used in agricultural community.
24hrs from 12Z 5/3/2010
RTMA = Real-Time Mesoscale Analysis
asatnrc eeUbaQET
24.86
43.6KPKP
•adjusts Penman Ep to model seasonality in vapor transfer process through aKP and bKP:
•seasonal functions of day of year,
•calibrated in Kimberly, ID.
ETrc across Western Region:RTMA-driven Kimberly Penman ETrc
24hrs from 12Z 5/3/2010
ETrc across Western Region:RTMA-driven ETrc difference: Kimberly Penman – Penman-Monteith
24hrs from 12Z 5/3/2010
dudd ABGHETLWLWSWt
W
1
LWd LWu
GGround heat flux
Shortwave radiation balance, SWn
Wt
Longwave radiation balance, Ln
SWdSWd ET H
Net availableenergy, Qn
B
Ad
Radiative parameterization
HETQGLWSW nnn
Western Region ETrc forecast goals:
Forecastoperations
• real-time• daily/weekly
Climatology• Jan 1980 – Dec 2009• high resolution• unbiased wrt forecasts
ETrc(climo)
ETrc(forecast)
statistical analysis
ETrc(t)
Value-added ETrc(forecast)
experimental www publication
www publication
feedback from users
verificationProject goals:
•provide end-users with web-disseminated, fine-resolution, accurate, daily-to-weekly forecasts of ETrc across the NWS Western Region domain,
•add value to the ETrc forecasts by generating and comparing to a 30-year climatology.
Data sets:Climatology and forecasts
climatology
forecast
+ 1
day
+ 7
days
climatology
Jan
‘79
Aug
‘06
Pres
ent
Dec
‘09
NLDAS• T• Tdew• Wind• SWd• LWd
• Hourly• 12 km• also 4 km (HRAP)
NDFD• T• Tdew• Wind• AEC
• 3-, 6-, hourly• 2.5 → 5 km
RTMA• T• Tdew• Wind• AEC
• Hourly• HRAP grid
NLDAS = North American Land Data Assimilation SchemeNARR = North American Regional Reanalysis
NDFD = National Digital Forecast DatabaseRTMA = Real-Time Mesoscale Analysis
AEC = Areal Extent of Cloud
serially inconsistent
biased wrt NDFD?
verifies NDFD
Potentiallyunavailable
Unresolved issues
• What products do the users want and/or expect?• forecast periods,
• spatial resolution,
• how often updated?
• What do we tell the users we can offer?
• Climatology• accessing RTMA history – SKY
• generating unbiased RTMA/NLDAS climatology
• low-res NLDAS climo in experimental period, replacing with high-res
RTMA/NLDAS unbiased climo later?
• time-step for climatology statistics: daily? multi-day? weekly?
• if weekly: moving average (365 surfaces)? static weeks (52)?
• Verification:• ETrc vs. Epan to check radiative parameterization?
• anomaly behavior,
• synthetic or observed Epan?
• how do we check our forecast skill?
Post-Mortem for June 6-10 flooding
• Forecasts generally poor and under simulated for peak flows that occurred June 6-10, 2010 in northern Utah and western Colorado
• General conditions leading into event:– Very cool May– Warm, moist air mass beginning June 5– Temperature forecasts generally good– SNOTEL sites in flooding catchments near average for this time of year – Streamflow forecasts were almost uniformly too low
• Ongoing study to understand why• Preliminary results focus on Little Cottonwood Creek
– What happened in the real world? – What happened in the model world?
Recent local Temperatures
• May was generally cool, delayed melt
• First week in June was warm
From NOAA CPC
A cool May
A warm week
Little Cottonwood at Crestwood Parkflow forecasts under-simulated
forecasts
observed
simulated
bankfull
flood
weekend
WFO Watches and warnings for Little Cottonwood (Cottonwood, Crestwood Park hydrograph shown)
Flood Warning543 PM MDT SUN JUN 6
Hydrologic Outlook (ESF)330 PM MST SAT JUN 5“FLOODING IS NOT ANTICIPATED”
Flood Watch1132 AM MDT SUN JUN 6
weekend
Little Cottonwood snow year
• Snowmelt rate extraordinarily large? (no)
• Snowmelt extraordinarily late? (no)
2010avg
Little Cottonwood at Crestwood Park: Temp. forecasts
forecasts
observed
Averaged over all elevation zones: not bad!
weekend
Snotels in Cottonwoods• Mill-D North
(8960’, southwest face)
– “middle”
• Brighton (8750’, southeast face)
– “middle”
• Snowbird (9640’, northeast face)
– “high” Snowbird
Mill-D North
Brighton
SWE in Cottonwoods
2010 compared to 2006
Nearly identical melt!
2006
2010
SWE melt comparisonCompared to 2005 and 2006: 2010 had less SWE, but melt was more synchronized
2010
20052006
Lower vs
upper meltout
20 days
30 days
5 days
SWE/Snotel Comparison• June Snowmelt at Snowbird in 2006 & 2010 were similar, but in 2006, snow at lower
elevations was gone – in 2010 there was synchronized, rapid melt at both elevations. • What about May 2006, when both zones were also melting? Melt rates were lower,
particularly in upper zone (and flow response was smaller).
2010
~1.8 in/day
~2 in/day
~1.5 in/day
2.5 to 3 in/day
2006
Snow Distribution – corroborates presence
of lower elevation snow in 2010 at start
of event
2006 Snowbird SNOTEL trace almost identical to 2010 trace from June 1-10
However, NOHRSC indicates south facing slopes had already melted out in 2006
2010 June 1
2006 June 1, for comparison
Streamflow Year Comparisons• Both 2005 and 2006
had bigger snowpacks
• Flow peaks in prior years occurred after lower snow was gone
• In 2005 and 2006, imagine combining the two melt pulses into one – you’d get a flood!
2010
2005
2006
Rainfall
What did the model think? • Model SWE at time of runoff was sufficient, and melt rates were close to observed rates• Overall shape of seasonal snowpack was good (compare to ~38’ peak at Snowbird)
2.8 in/day
2.6 in/day
Jun 3-8
What did the model think? • Model soil was on pretty dry in upper and middle levels entering event, and stored water
Deficits in model soil filling
Jun 3-8
3-4 days meltsoaked up
upper
lower
middle
Was the model right about soil moisture?
• Partly right – 2010 was relatively dry• Partly wrong – throughout winter, SM steadily moved upward toward normal
levels• In model, this recovery started late and didn’t get as far• 3-4 days of melt went by before model really started generating runoff
Preliminary Conclusions The snowmelt that drove the flooding was …
More rapid than normal (by about an inch per day) From more area than normal (both middle and high elevations)
The cool May 2010 … Delayed the melt of snow into June, holding lower elevation snow Atmospheric moisture effects contributed little during the event, but prior month of cool, relatively moist
conditions may have helped it ripen
CBRFC models under-simulated the streamflow response The snow model performed well, simulating proper amounts and melt rates (with adjustment by forecasters) SNOTEL soil moisture confirms that 2010 SM was relatively dry, but: The Sacramento hydrologic model soils may have been too dry, storing enough snowmelt during the event to
buffer the flow response
SWE related mods? LCTU1 (Cottonwood @ SLC)only melt factor was used
MFC 0.69
CF
S MFC 1.70
MFC 1.80 – deleted 6/7
MFC 2.35 – deleted 6/8, 6/10
MFC 1.47 – deleted 6/9
Note, appears we did only one soil water mod during May-June