space weather prediction center ncep psr 2011 doug biesecker

31
Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Upload: ami-harmon

Post on 13-Jan-2016

224 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Space Weather Prediction Center

NCEP PSR 2011Doug Biesecker

Page 2: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Outline

• SWPC’s GPRA and Geomagnetic Storms• The first Operational SWx Model

– WSA-Enlil

• What’s coming next?– Geospace Modelling– Whole Atmosphere Modelling

Page 3: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

SWPC’s Proposed GPRA• Geomagnetic Storm Forecast Accuracy

– Percentage of geomagnetic storms occurring for which a forecast was successfully issued

• Storms equal to or exceeding the Minor Storming level as defined by the Daily Geomagnetic A-index ≥ 30

– equivalent to ≥ G1 Level on the NOAA Space Weather Scales.– Solar Cycle 23 (5/1996 – 12/2008) GPRA accuracy was 30%– Statistics are tracked for last 30 A ≥ 30 storms

• (10/14/2003-10/31/2011)

FY11 Target

FY12 Target

FY13 Target

FY14 Target

FY15 Target

FY16 Target

GPRA POD Goals 30%

FAR=70%

40%FAR=60%

40%FAR=60%

45%FAR=55%

50%FAR=50%

50%FAR=50%

POD Actual 38%FAR=63% 3

Page 4: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

The Geomagnetic Storm SWx Scale

Page 5: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Geomagnetic Storm Impacts

Manned SpaceflightIncreased radiation risk

Power Grid OperationsGrid failure, Grid capacity, Component Failure,

GPS Timing

Impacts from geomagnetic storms are wide-ranging

with potentially significant consequences.

GPSPrecision Agriculture,

Surveying, Drilling, Military

Satellite OperationsLoss of mission, reduction in capability

Aircraft OperationsPolar Flights, WAAS, NextGen,

Airline Communication 5

Page 6: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

SWPC Customer Growth is Accelerating

Page 7: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

WSA-Enlil Improves Geomagnetic Storm Prediction

• Accepted in FOC Dec 5, 2011?• Provides perspective on co-

rotating structures 1-27 days in advance, CME’s 1-4 days

• Reduces error in geomagnetic storm onset time from ±12 hrs to ±6 hrs

• Reason for early GPRA success?

7

Page 8: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

The Results

‘Average error’ is calculated as ‘average absolute error’, which was used by CCMC in Taktakishvili et al. 2010.

‘RMS error’ is the community preferred measure.

The ‘community’ accepted error during Solar Cycle 23 is ±12 to ±15 hours

EVENT START Shock at ACEWSA/ENLIL

NOAA DIFF

02/13/2011 01:44 02/18/2011 00:49 02/17/2011 15:00 9:49

03/08/2011 20:14 03/10/2011 06:10 03/10/2011 08:00 1:50

06/02/2011 07:57 06/04/2011 19:58 06/04/2011 08:00 11:58

06/21/2011 03:25 06/23/2011 02:26 06/23/2011 12:00 9:34

08/02/2011 06:19 08/05/2011 17:22 08/05/2011 17:00 0:22

09/06/2011 00:00 09/09/2011 11:49 09/09/2011 17:00 5:11

09/14/2011 02:00 09/17/2011 02:56 09/16/2011 21:00 5:56

09/24/2011 10:00 09/26/2011 11:53 09/26/2011 16:00 4:07

10/01/2011 00:00 10/05/2011 06:47 10/05/2011 16:00 9:13

10/26/2011 10:00 10/30/2011 08:55 10/30/2011 10:00 1:05

11/09/2011 13:54 11/12/2011 05:30 11/12/2011 02:00 3:30

11/26/2011 08:00 11/28/2011 21:15 11/29/2011 12:00 14:45

AVERAGE ERROR 6:26

RMS ERROR 7:48

8

Page 9: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Customers

Enlil CONOPS

NSO

SWFO

WSA-Enlilmodel run

Modelresults

NGDC

NASA

MonitorCME event

NCEPCCS

Ambient & CME inputs

GONGdataSOHO

LASCOdata

SWFO Forecast products

Archiveoutputs

GenerateCME cone

data

Process model results

Generate graphicalproducts

Inform improved forecast

STEREOdata

NASA

Name CME

9

Page 10: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

T = 0 T = 5 days

CME 1

10 day model startup forecast5 days CME injection

T = -15

CME 2

1.5 hours Wallclock time on NWS CCS

WSA-Enlil run schematic

10

CME’s are parameterized as a simple ‘cone’

Page 11: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

WSA-Enlil production run cycle at NCOmodel runs every 2 hours

00Z 02Z 04Z 06Z

Enlil Enlil Enlil with CME Enlil

PreprocessedGONG data

PreprocessedGONG data

PreprocessedGONG data

PreprocessedGONG data

CME Detected

CMECone Data

WSA WSAWSAWSA

11

Page 12: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Inputs drive the performance• ENLIL propagates CME’s from the corona out to Earth

– Driven by the empirical WSA-Model• WSA errors in wind speed of 50-100 km/s are common• WSA errors in background wind speed of order 100km/s can

change arrival time by up to 6 hours– Driven by the parameterization of Coronal Mass Ejections

observed in near-real-time• An educated guess would be the CME parameter estimates are

good to no better than 20%• 20% error in CME parameters can change arrival time by more

than 6 hours• This is where SWPC efforts will be devoted in FY12 and

likely beyond

12

Page 13: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

WSA-Enlil in action

13

Page 14: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

One of the better results1:05

0:22

14

Page 15: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Today’s Forecast

15

“The forecast is for rain, somewhere on Earth, sometime today.”

Is this an analogy to geomagnetic disturbance products today? Perhaps a slight exaggeration, nevertheless there is a need for regional forecasts with

longer lead time.

www.ruralwellbeing.org.uk/ images/weatherman.gif

Page 16: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Geospace Model Transition

Howard SingerNOAA Space Weather Prediction Center

Safeguarding Our Nation’s Advanced Technologies

Geospace Models

Protecting Power Grids

(and other services)

Page 17: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Regional Geomagnetic Activity Prediction

• Need for both continuous activity prediction and storm prediction (location, onset time, duration, magnitude, probability of exceeding threshold)

• Focus on dB/dt and Kp• dB/dt: demonstrated

customer need (e.g. power utilities)

• Regional K: to serve customers and demonstrate improvement over current global products

17

Page 18: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Secondary GeomagneticActivity Products and Metrics

MHD Model Auroral Products

Latitude, width, local time, and intensity of the auroral electrojets

Related to locations of large dB/dt’s

Related to location of HF radio absorption

Provides location of polar cap where Solar Energetic Particle’s have access and can disrupt HF radio communication

Energetic particle precipitation

Metrics need to be developed

Potential data sources for comparison include: AMPERE, DMSP, POES, ground-based magnetometers

Polar Visible Aurora: High Solar Wind Conditions on

April 17, 1999 over the North Pole

Geosynchronous orbit magnetopause crossing Ionosphere: products and disturbances; e.g TEC

Page 19: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

19

Models at CCMC Participatingin Geospace Evaluation

MHD Models: 1. Space Weather Modeling Framework (SWMF) - U. of Michigan (delivered to CCMC)2. The Open Geospace General Circulation Model (Open GGCM) - University of New Hampshire (delivered to CCMC)3. Coupled Magnetosphere-Ionosphere-Thermosphere (CMIT) - BU CISM, Dartmouth, NCAR (delivered to CCMC)4. Grand Unified Magnetosphere-Ionosphere Coupling Simulation (GUMICS) - Finnish Meteorological Institute (recently parallelized, not ready for full evaluation for selection process)

Empirical Models:5. Weimer Empirical Model, Va. Tech (delivered to CCMC/may update)6. Weigel Empirical Model, George Mason (delivered to CCMC)

Page 20: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

SOLAR WIND – INDUCED ELECTRIC CURRENTS FLOWING IN THE

MAGNETOSPHERE

20

Credit: Kivelson and Russell, Introduction to Space Physics

Page 21: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

21

Geospace Model Transition: Recent Activities and Current Schedule

• 4/25/11: Geospace modeler meeting focused on evaluation metrics, selection process and initiate discussion to understand resource requirements

• May-June 2011: Spatial, temporal, and window sensitivity testing at CCMC to refine and iterate on metrics, event selection, verification measures

• June 26 – July 1 2011: GEM-CEDAR Workshop including Modeling Challenges and discussions with modelers on sensitivity tests and schedule

• July – Nov 2011 : Empirical model tests, gathering data for additional events, tool to integrate currents, comparisons of db/dt calculated by CCMC and modelers (SWMF and GGCM)

• Dec: Presentations and discussions with modelers at Metrics and Validation Session at Geospace Modeling Workshop (day before AGU meeting)

• Jan 2012: Runs and post processing

• Feb – March 2012: Analysis and Report Writing

• April-May 2012: SWPC circulate draft report for comments

• June 2012 : Model Selection at SWPC (I suppose this is now a choice for Louis?)

Page 22: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Geospace Model Plans

• FY11 Model Selection– Metrics– Community wide testing

• Mid-FY12 Begin Transition• FY15 Begin Operations

Page 23: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

23

Computational Requirements

• Models under consideration run in real-time or near-real time on 64 processors• This configuration used for model evaluation• Detailed conops will be developed during FY12

• Current vision includes:• Porting codes to NCEP research computer and testing in configurations

used during evaluation period. • Test runs using 64 processors for several intervals of solar wind

conditions of up to 2-week duration• Input and output data have 1-minute time resolution; however, time

steps can be on the order of 5 seconds• Codes produce on the order of 100 Gbytes/day output (it will not be

necessary to store the entire output data stream, although details need to be worked.)

• Test models on different types of events: 64 processors, for 2 to 3 day runs

• Test models with different resolutions and code settings: 128 processors, for 2 to 3 day runs

Page 24: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Input for Geospace

• NASA/ACE Satellite at L1– Upstream of Earth 1

million miles• 15-60 minutes

– Solar wind velocity, density and magnetic field

– WSA/ENLIL can provide 2.5/3 inputs

• Hoped for NOAA replacement (DSCOVR (Triana)) in FY11 Budget

Page 25: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Now for a right turn…

• Space Weather meets Terrestrial Weather– Where?

• At 60km

– Why?• To better describe the

ionosphere– GPS– Communications– Satellite Drag

Page 26: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

26

Integrated Dynamics in Earth’s Atmosphere (IDEA)

Whole Atmosphere Model (WAM)• Collaborators

– NCEP Environmental Modeling Center (EMC)

– Naval Research Laboratory (NRL)

– National Center for Atmospheric Research (NCAR)

– Others

• Sponsored by– AFOSR Multidisciplinary

University Research Initiative (MURI) program

– NASA Living With a Star (LWS) and Heliophysics Theory programs

26

WAM = Extended GFS

Team: R. Akmaev, F. Wu, T. J. Fuller-Rowell, H. Wang

Page 27: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

The advantage of Whole Atmosphere Modelling

• The Whole Atmosphere Model (WAM) is an extension of the operational Global Forecast System (GFS) model currently used operationally– The operational version of this model is run four times a day but is limited to

60 km.• The WAM, currently under development for space weather, is an

extension of GFS up to 600 km with additional atmospheric chemistry and dynamics appropriate for the upper atmosphere.– The WAM has been tested and validated at low spatial resolution and has

shown great promise in capturing many space weather features and phenomena critical to GPS users, satellite drag/ orbit prediction, and satellite communication

– Vertically extended models and data assimilation systems have also been shown to benefit conventional weather prediction by removing artificial boundaries and eliminating the limitations of existing operational systems by better specification of upper layers through which the weather satellites observe the lower atmosphere.

Page 28: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

The basic tasks• There are three critical research areas that need to be

addressed:– (1) the development and implementation of the

Ionosphere-Plasmasphere-Electrodynamics (IPE) module– (2) understanding the impact of increasing spatial

resolution of the model– (3) implementation and testing of new data assimilation

techniques applicable to the middle and upper atmospheres and ionosphere.

• These activities will help in the assessment and understanding of the impact of the lower atmosphere on the structure and irregularities of the ionosphere.

Page 29: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Funding in FY12(13?)• $800K to the National Weather Service, NCEP Central Operations

– for additional computation resources (CPU and disk)– allowing WAM to be run at higher spatial/temporal resolution which would resolve waves and

structures propagating from the troposphere which are critical for initiating ionospheric structures and scintillation

• $400K to NCEP/EMC– to support the development of Ionosphere Plasmasphere Electrodynamics module and the

integration of IPE into GFS/WAM. The IPE module adds additional physics related to the ionized component of the upper atmosphere. The neutral and ionized components are highly coupled and both are critical to the full representation of them ionospheric gradients and irregularities

• $800K to SWPC– to develop data assimilation techniques for satellite data above 60 km. There are several

assimilation techniques (Kalman Filter, 3-D-VAR, 4-D-VAR, etc) which have strengths and weaknesses at various altitudes in the atmosphere. These techniques need to be implemented, explored, and evaluated.

– to support evaluation of the model improvements and validation and verification of the results of each of these improvements (higher resolution, IPE module, and data assimilation) data being implemented into WAM.

• The application of these additional resources will accelerate the model development by at least two years.

Page 30: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Computational Requirements• WAM estimate for annual computer requirement

for 2012 on NCEP machines.– The first 6 months on vapor, the following 6 months

presumably on the new SGI machine, zeus. If there is a delay in porting WAM to zeus it could impact the expected usage.

– Total requirement, all projects: 34050 node-hours,or ~4250 hours on 8 nodes, or roughly 8 nodes 50% of the time. Margin of error 50%.

– Disk space requirements ~30 TB for 2012.• 2013-2015 usages are estimate to increase by a

factor of two per year.

Page 31: Space Weather Prediction Center NCEP PSR 2011 Doug Biesecker

Future Plans

• Space Weather has come into numerical prediction in a small way in FY12 with WSA-Enlil– Bringing a dramatic improvement in prediction of

geomagnetic storms• FY12-15 will see a large increase in SWPC

needs for computational resources– Geospace and WAM modeling will bring dramatic

new capabilities that will benefit many areas of space weather

31