operational drought information system kingtse mo climate prediction center ncep/ nws/noaa...
TRANSCRIPT
Operational Drought Information System Kingtse Mo Climate Prediction Center
NCEP/ NWS/NOAA
Operation--- real time, on time and all the time
1
MissionCPC issues operational monthly and seasonal drought
outlook and participates in the Drought Monitor These products are used by government, NIDIS, local
state government , regional centers and private sectors To support the CPC mission, we give drought briefing each
month to review the current conditions and forecasts
2
Definition of drought– persistent dry conditions
Colorado basin
SE
3
A wet region
drought
6 mo running mean black line
3 mo running mean (black line)
SM 1-2 months delay
No smoothing
Red line: monthly mean, no smoothing
75-85W,31-35N
4
SM has much lower freq. over the western region
A dry region
5
Good indices
• 1. They do not depend on season• 2. They do not depend on location• 3. They are accessible in real time• 4. It measures the spatial and temporal
scales of drought • 5. All indices should be able to pick upstrong drought evens
6Kelly Redmond
Different faces of droughtdefine drought by impact• Meteorological drought– P deficit• Agricultural drought--- soil moisture
deficit• Hydrological drought_ runoff or
streamflow deficit
Using index to define drought7
SPI fcsts 20130831
Recent rain diminishes drought over the Southwest and California
Continuous rainfall events causes floods over the Southeast and the East cast
The SPI24 still shows the strong drought events of 2012
SPI gives the historical development of drought/floods
8
SPI SPI Advantages:• Easy to use and only need station data• Cover all time scales• Do not need a hydrologic model. • (Other indices are model derived products)SPI Disadvantages:1.It does not contain snow information2.Areas where soil moisture feedback is important or
large E, SPI may not be representative (e. g. Amazon)
9
North American Land Data Assimilation system They are not TRUTH
• Surface land model- Noah, SAC, VIC, Mosaic and Catchment model
• They are driven by forcing which consists of precipitation (P), Max and min Tsurf and wind speed for a water balance model
• Some models like Noah and Mosaic have the energetics –radiation terms. VIC has both versions
• Outputs: Evaporation, Soil moisture, soil temperature, runoff , Snow water equivalent. And many others
10
Don’t worry, be happy!!
Even though the total soil moistures differ from one model to another, their anomalies (or percentiles) are very similar!!
(Robock et al 2004;
Dirmeyer et al. (2004)
Koster et al. (2008)
All models were driven by the same forcing 1511
12
Multi model SM information
U Washington
NCEP/EMC
Both captures the wetness over the Eastern and East central United States and dryness over the Southwest and the Plains,But intensity differs
•Differences between two systems are larger than the spread among members of the same system
•The differences are not caused by one model. They are caused by forcing.
• In general, extreme values from the UW (Green) are larger than from the NCEP (red)
NCEP(red),UW(green)
standardized SM anomalies for area 38-42N,110-115W 13
The EMC NCEP system
• Four models: Noah, VIC, Mosaic and SAC• Climatology: 1979-2007• On 0.125 degrees grid• P forcing : From the CPC P analysis based on
rain gauges with the PRISM correction. (all stations reports within cutoff time
• Other atmospheric forcing: From the NARR•
14
University of Washington system• Four models: Noah, VIC, SAC ,CLM Catchment• (models may have the same name, but versions
may not be the same)• Climatology: 1915-2007• P, Tsurf and low level winds from
NOAA/NCDC co-op stations• P from index stations
15
Forcing
• Since the differences among the members of the same system are small, the differences do not come from models.
• Differences come from forcing.• There are two major forcing terms: precipitation
and temperature.• Their differences are larger after 2002
16
Experiments The VIC model of 0.5 degrees resolution from the UW system
was chosen for experiments. All experiments started from Jan 1 1979 using the same initial
conditions from the UW VIC model in the UW system. Experiments end on 31Dec 2008
Forcing terms have two components1. P forcing :Precipitation 2. F forcing : Tmax, Tmin and wind
speed 17
Four experimentsComparison between• Exp (P uw F uw) vs Exp(Pncep,Fuw) and• Exp (Puw, Fncep) vs Exp(Pnecp,Fncep) indicates the differences caused by Precip
Comparison between• Exp (P uw F uw) vs Exp(Puw,Fncep) and• Exp (Pncep, Fuw) vs Exp(Pnecp,Fncep) indicates the differences caused by F forcing (Tsurf and winds)
18
Experiments :RMS differences of SM %Same F forcing Same P forcing
19Large differences between experiments with the same F forcing but the same P forcing are large
Number of station reports averaged over a year
20
Number of reports /month averaged over the box
Large drop in real time
21
Challenges: improving drought monitoring• Improve historical and real time
Precipitation data and analyses• Improve NLDAS model forcing: P,
downward short wave radiation etc• Improve hydrologic model • Improve and integrate satellite
observations with station data.• Link to attribution
22
NMME/IMME seasonal fcsts
23
We have 6 models:CFSv2 24 members;GFDL, CMC1 and CMC2 : 10 membersNASA: 11 membersNCAR : 6 members
Hindcasts from 1982-2010P, Tsurf monthly means
JAS 2013
ASO 2013ASO 2013
Hydroclimate FCSTs
• SPI forecasts based on the National Multi Model ensemble (NMME)
• ESP forecasts from the UW• Cfsv2_VIC forecasts from the Princeton,
EMC and MSU• NASA SM from their Coupled model
forecasts
24
25
SPI forecast
If you have precip monthly mean fcsts, you can have the SPI forecasts
Yoon et al. JHM 2012
CGCM
SPI fcsts (201308)
26
verification
ESP (Ensemble streamflow prediction) vs NMME_VIC Fcsts
IC s
Run VIC with observed P and Tsurf
Jan 1,1915 from UWJan 1, 1979
ESP- P T inputs taken from randomly selected observations
Both ESP and NMME_VIC have the same initial conditions, but ESP has no climate forecast information of P and Tsurf
Fcst forward
Starting date
Feb 5 Feb 6---
27
NMME_VIC :forcing s were taken from error corrected T P from CGCM
ESP FCST UW ICs=20130802
28
\August lead=1mo
Sep lead=2mo
Oct lead=3mo
Acc ro lead=1mo
SM fcsts EMC_MSU_Princeton
29
AUGUST lead=1mo
September lead=2mo Oct lead=3mo
Same as the ESP, but climate forcing is given by the CFSv2 forecasts
NMME_VIC forecasts
• Initial conditions from the VIC simulation taken from the UW NLDAS_VIC (perfect)
• Climate forcing derived from the members of the NMME for each model
• Drive VIC to get SM and Runoff• For a given model and given lead time, we took the
ensemble mean of all members. The climatology of the forecasts is corrected in the cross validated way.
• SM /Runoff or SRI3 ensemble mean is the equally weighted mean of all 6 models
30
Fcst skill for SM
31
Lead-1 : correlation >0.8 (WOW!!!)
Lead-3: Over the western interior dry region, the fcsts are still skillful for all seasons and the North Central for January (high skill regions)
Low skill regions are circled
Differences btw NMME-ESP
32
1. No significant differences for Lead-1 and Lead-2
2. Only October and January forecasts pass
the Livezey Chen field pass3. Differences are in the areas that the skill is low and dynamically active areas
4. Oct fcsts are helped by skillful P forecasts
Lead-3
Two regimes
Dry: western interior & eastern Texas• Forecast skill of SM and Runoff are high at lead-3, • Contributions are from the initial conditions• ESP_VIC also has high skill• Areas with low P mean and Low P variability
WET: Eastern region and monsoon region• Wet areas with large mean P and P variability• Skill is low even at Lead-1• Dynamically active and P depends on the moisture transport• NMME has higher skill than the ESP 33
Problem with hydroclimate prediction- low P fcst skill
34
No skill after Lead-1Except the SoutheastIn Oct
When the CF starts to contribute at Lead-2 or higher, the skill of P forecasts are so low, it does not make a difference
Issues of hydroclimate fcsts• At Lead-1, the initial conditions dominant the forecast skill.
The NMMS precipitation forecasts have some skill, but it competes with the initial conditions
• At Lead-2 and Lead-3, the impact of forcing starts to contribute to skill, but the skill of P fcsts decreases . In the western region, the Ics still contribute but over the dynamically active region such as the Southeast or the monsoon region, the P forecasts need to be good enough to contribute to SM or RO forecasts at higher lead
35
Conclusions
• GIVE Me :
Better P forecasts at Lead-2 and Lead-3You will have Better SM and Runoff forecasts over the dynamically active region • Give me
better station data reporting in real timeYou will get : better NLDAS with less uncertainties and better forecasts over the dry areas
36
Measure the differences among models
Rm for a group of models Wm :the mean intermodel variance (or spread) Wint (m): interannual variance of the ensemble mean
)(int mR mm
Similar formula was used by Dirmeyer et al (2004). to assess Global wetness products except we use variance instead of standard dev.
1737
R values for SM %1.The spread among the
members from the same system (UW or NCEP) is small. It is less than 0.4. (Fig. a and b)
2. R values with all UW and NCEP members together is much larger (Fig.c).
This implies that the mean differences between two systems are large
38
1.The RMS difference (Fig.d) between the ncep and the UW ensemble SM means are large over the western U. S. (> 20%).
2. Largest differences occur after 2001 as indicated by the mean differences for two periods (Fig. f and g)
39
PredictionOceanic conditions• ENSO normal • the positive SSTAs over the North Pacific will continue through
summer.Precipitation above normal rainfall over the East above normal rainfall over the SouthwestDrought• All forecasts indicate that drought over the Central U. S. and
Texas will improve • A normal to slightly above normal monsoon will improve
drought conditions over the Southwest
40
RMSE (NMME) R(NMME/ESP)
Lead=1mo
Lead=2mo
Lead=3mo
Spread lead=2 and 3 mo
Over the central and western U.S., the ESP has advantages up to lead=2mo
Over the eastern U.S., the NMME has advantage than the esp
For lead=3, skill overall is very low and the NMME and ESP have comparable skill
The spreads are small and located in the area with low skill
.
Comparison with ESP
41