Taylor Mandelbaum School of Marine and Atmospheric Sciences Stony Brook NY Brian Colle School of Marine and Atmospheric Sciences Stony Brook NY Trevor Alcott Earth Systems Research Laboratory Boulder CO ID: 467156
Download Presentation The PPT/PDF document "Using Model Climatology to Develop a Con..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Using Model Climatology to Develop a Confidence Metric
Taylor Mandelbaum, School of Marine and Atmospheric Sciences, Stony Brook, NY Brian Colle, School of Marine and Atmospheric Sciences, Stony Brook, NY Trevor Alcott, Earth Systems Research Laboratory, Boulder, CO
* This work is supported by NOAA-CSTARSlide2
Outline
Background/MotivationMethodology: How the tool is constructedShow two cases as examples: One with large spread and another with smaller spread.Future work: what can be done to help improve the tool and prepare for real time use in December 2015.Slide3
Background
The Ensemble Situational Ensemble Table (ESAT) plots
anomalies in multiple formats (using NAEFS and GEFS).
The goal of ESAT is to provide
a
tool that can be used to assess
anomalies in
ensemble forecasts.
DSS
and forecasters can determine how anomalous a forecast is relative to previous forecasts.Slide4
Motivation
Mean anomaly can show how anomalous a forecast is but…
Is the model spread for that forecast greater or less than normal for the SA?
If the spread is less than normal this would translate to greater confidence the SA may occur..
Pictured: SA of
72h
GEFS mean M-Climate valid 8 Jan 1996,
0hSlide5
Motivation
This is the conventional way to view spread.. Calculated relative to the ensemble mean
But, there is no prior knowledge incorporated to understand if this forecast is more or less predictable than other cyclones at this same forecast lead time.
Pictured:
72h GEFS ensemble
spread valid 8 Jan 1996,
0h.Slide6
Motivational Questions
Can we use the model climate to calculate the climatological spread for more anomalous weather events? Can one use this climatological spread to determine how anomalous the spread may be for a particular forecast and
location?
How
can we best communicate these spread anomalies?Slide7
Terminology
M-Climate: Refers to model climatology, or how a model forecasts during a certain seasonal period at a certain forecast hour.Anomaly: F-Cm, or the difference between the forecast and M-Climate at each gridpoint.
Standarized
anomaly or z-score (SA): F-C
m
/
σ
, difference of forecast and climatology at each
gridpoint
normalized by
gridpoint
standard deviationSlide8
Datasets
GEFS Reforecast2 from ESRL every 6-h from Nov 1985 to March 2015. Create M-Climates for the winter (December-February) season.Obtain Ens mean + spread taken from GEFS Reforecast2.
Create 3d (
cases,x,y
) array for each day, centered about 21 day window on CONUS grid. For 1985-2015, array is (630,30,63).
Each forecast hour given unique array (0z, 3z, 24z, 168z…).
Use M-Climate to determine spread anomaly in GEFS this winter.Slide9
Method to Obtain Standardized Spread AnomalySlide10
Case 1: 11 February 2010
Taken from ESRL PSD Reanalysis 0.3x0.3 degree dataset
From NESIS
500mb
Hgts
MSLP
MSLPSlide11
Mean, Spread, SA, and M-Climate Anomaly
120h Forecast Valid 0000 UTC Feb 11 2010Slide12
Mean, Spread, SA, and M-Climate Anomaly
96h Forecast Valid 0000 UTC 11 Feb 2010Slide13
Mean, Spread, SA, and M-Climate Anomaly
72h Forecast Valid 0000 UTC 11 Feb 2010Slide14
Mean, Spread, SA, and M-Climate Anomaly
48h Forecast Valid 0000 UTC 11 Feb 2010Slide15
Mean, Spread, SA, and M-Climate Anomaly
24h Forecast Valid 0000 UTC 11 Feb 2010Slide16
Case 2: 8 January
1996Taken from ESRL PSD Reanalysis 0.3x0.3 degree datasetStrong offshore low which developed into a nor’easter – how confident was GEFS relative to storms of similar magnitude?
From NESIS
MSLP
500mb
HgtsSlide17
Mean, Spread, SA, and M-Climate Anomaly
120h Forecast Valid 0000 UTC 8 Jan 1996Slide18
Mean, Spread, SA, and M-Climate Anomaly
96h Forecast Valid Jan 08, 1996 at 0zSlide19
Mean, Spread, SA, and M-Climate Anomaly
72h Forecast Valid Jan 08, 1996 at 0zSlide20
Mean, Spread, SA, and M-Climate Anomaly
48h Forecast Valid 0000 UTC 8 Jan 1996Slide21
Mean, Spread, SA, and M-Climate Anomaly
24h Forecast Valid 0000 UTC 8 Jan 1996Slide22
Conclusions
The reforecast M-Climate is used to determine whether the forecast spread is greater or less than expected for a particular forecast anomaly.The tool shows promise towards being able to determine large spread vs small spread days relative to the M-Climate.
Case studies illustrate that there can be relatively large differences in spread from storm to storm along the U.S. East Coast.
The tool is only as good as the model – if the spread is
underforecast
(undispersed) this tool may yield too much confidence in the forecast.Slide23
Future Work
Sample size issues for larger anomalies (smoothing? Increase range of anomalies?)Testing approach with 21 member GEFSMore variables (geopotential height, winds, 700hPa RH)Assess ensemble members and identify clusteringClean up code and refine to be included on a webpage (perhaps ESAT page with help of WPC).
Assess further efficacy by expanding to year long M-Climate datasetSlide24
References
Hamill, T. M., G.T. Bates, J. S. Whitaker, D. R. Murray, M. Fiorino, T. J. Galarneau, Y. Zhu, and W. Lapenta, 2013: NOAA’s Second Generation Global Medium Range Ensemble Forecast Dataset. Bull. Amer. Meteor. Soc., 94, 15531565.
Anticipating
a Rare Event Utilizing Forecast Anomalies and a Situational Awareness
Display,
The Western U.S. Storms of 18–23 January
2010. Randy Graham,
Trevor Alcott, Nanette
Hosenfeld
, and Richard
Grumm
. Bull. Amer. Meteor.
Soc. BAMS-D-11-00181.1
Questions?
Email:
Taylor.Mandelbaum@stonybrook.edu