Jet Propulsion Laboratory California Institute of Technology
Progress in Subseasonal Weather/Climate Forecasting Duane Waliser Jet Propulsion Laboratory/Caltech Pasadena, CA
Drought Response Workshop February 25-26, 2015 Irvine, CA
Weather Forecasts ~ O (10 Days) (Mid-Latitude Baroclinic Instability & Cyclone Lifetime)
Dynamic Forecasts root back to 1910
What about the forecasting between “weather” & climate ~ 2 weeks to 2 months? (aka sub/intra – seasonal)
Seasonal Forecasts ~ O (100 Days) (ENSO phenomena & Local/Remote Circulation Impacts)
Dynamic Forecasts root back to Mid1980’s
Progress and Importance in “ISI” Prediction Warranted a 2010 National Research Council Study ISI = Intraseasonal to Interannual aka S2S = Subseasonal to Seasonal
Useful predictions depend on having and being able to model “sources of predictability” with time scales that exceed typical weather patterns • • • • • •
El Nino – Southern Oscillation (ENSO) Madden-Julian Oscillation (MJO) Soil Moisture & Snow Cover Sea Ice Stratosphere–Troposphere Interaction Other slow ocean & atmosphere “waves” e.g. PNA, AO, IOD, CCEWs
Madden-Julian Oscillation
B&W “Photo” of whole tropics every 6 days
Madden & Julian, 1972
We have a Daily Index of MJO Activity Since 1979 Wheeler & Hendon 2004
Every day the MJO is characterized by a Phase (1-8) & Amplitude
Multi-week Insight into Monsoon Onsets & Breaks
Two Examples of Why Accurate MJO Forecasts Would be of Exceptional Value Multi-week Insight into Tropical Cyclone Frequency
Summer
MJO & Tropical Cyclones MJO Enhancement
MJO Suppression
1949-97 Tropical Storms & Hurricanes
Westerly/ Cyclonic Phase Higgins et al. 2000
Maloney and Hartmann, 2000
Easterly/ AntiCyclonic Phase
MJO Modulation of US (Dec-Feb) Temperature Anomalies Zhou et al. 2012
MJO Modulation of CA Winter Precipitation Seasonal Precipitation Rates ~2-4 mm/day MJO Modulation ~ +/- 1 mm/day Predictability Value!
Courtesy B. Guan/UCLA
The MJO & CA Winter Precipitation Typical Impact of MJO on California Precipitation Phase 1
Phase 5
Typical CA Wintertime Precipitation Rates ~2-4 mm/day (not shown); MJO Modulation ~ +/- 1 mm/day (~25-50%)
Left Figure CLIVAR MJO Working Group Right Figure Courtesy B.Guan, JPL/UCLA
MJO Occurrence and Frequency Jones et al. 2012
A few events per year – providing “forecasts of opportunity”
MJO Prediction Evolution
1990
2000
2010
Late 1990’s – Early 2000’s
• Operational MJO forecasts didn’t exist. • Focus on El Nino and Seasonal Forecasting • Weather and climate prediction models had poor representations of the MJO • MJO forecasts only good to a few days
Assessment of Intraseasonal to Interannual (ISI) Climate Prediction and Predictability, NRC 2010
See recent review Waliser, 2011
MJO Model and Forecast Improvement from the European Center for Medium Range Weather Forecasting (ECMWF) OLR; 15-day Lead Time; 10N-10S; TOGA COARE
Courtesy F. Vitart ECMWF
Promising gains from continued model improvements Resolution, Data Assimilation, Model Physics (Tomkins et al. 2007; Bechtold et al 2008) – M. Miller.
MJO Model and Forecast Improvement from the European Center for Medium Range Weather Forecasting (ECMWF) OLR; 15-day Lead Time; 10N-10S; TOGA COARE
Courtesy F. Vitart ECMWF
Promising gains from continued model improvements Resolution, Data Assimilation, Model Physics (Tomkins et al. 2007; Bechtold et al 2008) – M. Miller.
80 ensemble mean start dates, daily 1st Feb., May, Aug and Nov 1989-2008
Courtesy, F. Vitart ECMWF
MJO Forecasts from ECMWF
Useful MJO prediction skill out to 4+ weeks!
Progress attributed to • • •
better models more/better observations for initial conditions more computer power
Operational MJO Forecasts Many operational weather prediction centers now routinely produce MJO predictions
Gottschalck et al. BAMS, 2010
10 operation centers, 20 data streams, 13 ensemble forecasts (with 4 – 51 members)
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/CLIVAR/clivar_wh.shtml
Operational MJO Forecasts Prediction of a strong MJO event
NCEP
ECMWF
UKMO
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/CLIVAR/clivar_wh.shtml
Operational MJO Forecasts The MJO represents a “forecast of opportunity”
This week
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/CLIVAR/clivar_wh.shtml
NOAA Global Tropics Hazards & Benefits Outlook
Based in part on MJO forecasts
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/ghazards/
Subseasonal Forecasting Atmospheric Rivers?
ARs : Key to Beneficial & Hazardous Water Delivery
ARs Provide beneficial rain and snow for water supply
25% - 45% of annual precipitation in the west coast states fell in association with atmospheric rivers
Atmospheric Rivers are to the west what Hurricane Hazards are to the Southeast
From Dettinger et al. in Water, 2011
Ralph & Dettinger, 2012 An average AR transports the equivalent of 7.5 times the average discharge of the Mississippi River, or about 10 Million acre feet 22 per day. Of this, 20-40% may become precipitation.
Forecast Models Need to Improve their Forecasts of Landfall Location For example: at 5-6 day lead time, global weather forecasts cannot determine if it will hit LA or San Francisco
RMS Error in Forecast AR Landfall Location
Wick et al. 2013
The Unusually Snowy Winter of 2010/2011
2010/2011 winter • Largest total seasonal snow (~170% above normal) • Largest number of AR dates (twice normal) • Largest AR-related snow accumulation
On average 9 AR dates per winter contribute 37% total snow
December 18 to 22 – Five Straight Days of AR 17–19 December 2013
20–22 December 2013
• >13 feet of snow in the Sierras • >6 inches of rain in LA and >21 inches in parts of the foothills • Spread into Nevada/Arizona/Utah ; Zion NP evacuated
14 out of the season’s 20 AR dates occurred in one month
* No. of ARs in Dec. 2010 well above normal range
Right: Ralph and Dettinger 2012,
Climate Conditions of the 2010/2011 Winter –AO and –PNA tend to be associated with more stormy weather in California
500 mb Geopotential Height Anomalies
“Arctic Oscillation” (AO)
“Pacific North American” (PNA)
Circulation anomaly when both PNA and AO in “negative” phase
Phasing of AO/PNA vs. AR Frequency in California
When the AO and PNA are both in the negative phase, ARs are significantly more likely to occur.
Does the MJO influence ARs? A Little Bit More high-impact ARs are observed during MJO phase 6 – convection in the W. Pacific Ocean, including the top two events during WY1998-2010.
Guan, B., et al. 2012 AR events during WY 1998-2010 are plotted in relation to the phase and amplitude of the MJO. The amplitude of the AR is shown in terms of DSWE as size of green circles. AR dots/events inside the unit circle occur during weak/no MJO. AR dots/events outside unit circle occur during strong MJO events in the given phase of the MJO life-cycle.
New National Research Council Study “Developing a U.S. Research Agenda to Advance S2S Forecasting” S2S = Subseasonal to Seasonal
Recommend a bold strategy to increase the nation's scientific capacity for research on S2S forecasting • • • • • •
sources of predictability process studies & improving models sustained and new observations uncertainty quantification communication & decision support computation & infrastructure
Subseasonal to Seasonal (S2S) Project An International Experimental Operational Forecast Activity
Targets improving accuracy and use of forecasts at lead times of 2 weeks to 2 months
WMO Bulletin
S2Sprediction.net
• • • • •
5 – year project, starting 2014 11 operational forecast centers database of ~45 day forecasts focus on science & applications learn to tailor forecast products
Summary • Subseasonal (e.g. MJO) forecasting has become operational and starting to be useful – great progress in last 5-10 years. • MJO impacts on west coast wintertime precipitation and snowpack are evident and to some extent becoming predictable and lead-times of order weeks. • Significant community activities working towards improving global weather/climate model representation of the subseasonal phenomena (e.g. MJO, sea ice, soil moisture) to improve subseasonal predictions. • This work is really just beginning, and more “R” as well as “R2O” is needed before these sorts of relationships and their predictability can be “O”perationally exploited for weather and water related decision support.
• The S2S Project and the US NRC Study on S2S are poised to help improve the accuracy and utilization of S2S forecasts.