Sunday, September 02, 2007

sCAST Launched


News about this LR winter forecast method has dribbled out once and a while over the past few years. Appears the model is ready for prime-time. The sCAST developers have a paper about their model in the current Journal of Climate.

Improved Skill of Northern Hemisphere Winter Surface Temperature Predictions
based on Land-Atmosphere Fall Anomalies
.

Abstract
A statistical forecast model, referred to as the sCAST model, has been developed using observed October mean snow cover and sea level pressure anomalies to predict upcoming winter land surface temperatures for the extra-tropical Northern Hemisphere. In operational forecasts since 1999, snow cover has been used for seven winters and sea level pressure anomalies for three winters. Presented are skill scores for these seven real-time forecasts and also for thirty-three winter hindcasts (1972/73-2004/05). The model demonstrates positive skill over much of the eastern United States and Northern Eurasia; regions that have eluded skillful predictions among the existing major seasonal forecast centers. Comparison with three leading dynamical forecast systems shows that the statistical model produces superior skill for the same regions. Despite the increasing complexity of the dynamical models, they continue to derive their forecast skill predominantly from tropical atmosphere-ocean coupling, in particular from ENSO. Therefore, in the Northern Hemisphere extratropics, away from the influence of ENSO, the sCAST model is expected to outperform the dynamical models into the foreseeable future.



From TerraDaily...
"(Dr. Judah) Cohen and colleagues outline the link between October snow cover in Siberia and the Northern Hemisphere's winter temperatures, and snowfall.

"October is the month when snow begins to pile up across Siberia. October is also the month that the Siberian high, one of three dominant weather centers across the Northern Hemisphere, forms.

"In years when Siberian snow cover is above normal, a strengthened Siberian high and colder surface temperatures across Northern Eurasia develop in the fall.

""The result is a warming in Earth's stratosphere that occurs in January," said Cohen. "This eventually descends from the stratosphere to Earth's surface over a week or two in January, making for a warmer winter in Northern Hemisphere high latitudes.

"However, in mid-latitudes it turns colder, so winters in the northeastern U.S. and eastern Europe are likely to be colder and snowier than normal."

Summer Cancel - Snow Observed in North Finland




Several inches fell near Lapland on the last day of August. Locals report the last time that happened was back in '86.


View Larger Map

Tuesday, August 28, 2007

Farmers/ Almanac 07-08 Winter Outlook


"The 2008 Farmers’ Almanac long-range forecasts are calling for colder than normal conditions for the eastern half of the country – chiefly those areas east of the Mississippi –with snowier than normal conditions expected.

"The 2008 Farmers’ Almanac predicts that skiers in eastern New York and New England will enjoy some great conditions. The winter should also average as much as three degrees below normal down most of the Atlantic Coast, from New England through the Mid-Atlantic and the Piedmont into the Southeast. Cold conditions are on tap for the Southeast, with at least four cold frosts predicted all the way to Florida."

More...

Hope springs eternal.

Sunday, April 15, 2007

Interim Standings


Click image to enlarge.

To be ranked in the latest Interim Standings...forecasters must have entered at least four forecasts to be eligible under the 'two-thirds' rule.

After five storms...
First Place: Donsutherland1 with an average SUMSQ Z-score of -0.786
Second Place: Raven -0.749
Third Place: TQ -0.653

The third interim summary finds donsutherland1 maintaining his hold on 1st place. Raven moves up a notch to second. TQ jumps from 5th to 3rd.

A data table with the complete interim standing statistics...including Sum Squared Error (SUMSQ)...Storm Total Precipitation (STP)...Total Absolute Error (TAE)...Average Absolute Error (AAE)..and R-Squared (RSQ) at the web site.

The chart shows the distribution of forecaster SUMSQ Z-Scores (plum) and Total Absolute Error Z-Scores(cyan). Lower (higher) Z-Scores indicate better (worse) forecasts compared to all other forecasts made for each storm.

A Z-Score of 0 means the forecaster's error was equal to the average of all forecast errors. A Z-Score of -1 (+1) means the forecaster's error was 1 standard deviation below (above) the average of all forecast errors.

Contest # 5 - Results


Click image to animate.

Full forecaster verification table and contest results at the web site.

Forecasters: 10
8 veterans
2 rookies

Station forecasts: 110
Average stations per forecaster: 11
Stations with observed snowfall: 11

Storm-total precipitation (STP), all stations
Forecast
Minimum: 32.5” (shanabe)
Consensus Median: 39.75”
Maximum: 65.8” (Donald Rosenfeld)

Observed: 27.35”
Max single station: 11.4” CAR
Shutout: HYA…BDR…ISP…JFK…ABE…MDT…PHL…ACY…EWR…BWI…IAD…DCA…SBY…RIC…ORF…and RDU

***************************************************
Storm #5 - Results Summary

1st Place - TQ
SUMSQ Error: 43.9”
SUMSQ - Z: -1.065
STP: 13.15 (6th)
TAE: 17.85” (1st)
AAE: 1.79” (2nd)

2nd Place - Raven
SUMSQ Error: 47.9”
SUMSQ - Z: -0.971
STP: 10.15” (4th)
TAE: 17.95” (2nd)
AAE: 1.99” (4th)

3rd Place - shanabe
SUMSQ Error: 58.0”
SUMSQ - Z: -0.734
STP: 5.2” (1st)
TAE: 20.6” (4th)
AAE: 1.72” (1st)

Honorable Mention - Mitchel Volk
SUMSQ Error: 64.0”
SUMSQ - Z: -0.596
STP: 11.65” (2nd)
TAE: 19.25” (3rd)
AAE: 3.21” (10th)

Congratulations to the Winners!

SUMSQ: sum of square errors
STP: storm total precipitation
TAE: total absolute error
AAE: average absolute error

Saturday, April 14, 2007

Contest # 6 - Canceled


Today/s NWP guidance strongly suggests too few stations are likely to get too little snowfall to warrant a forecasting contest.

Contest # 6 is canceled.

Contest # 5 - Verification


Preliminary storm-total snowfall based on final CDUS41 bulletins from Friday and Saturday.

Report any errors in 'Remarks' along with a link to the correct data. Final results and contest summary will be posted tomorrow evening.

Rank Station Snow
1 CAR 10.90
2 BGR 5.30
3 PWM 3.00
4 CON 2.60
5 BTV 2.20
8 BOS 0.05
12 HYA 0.00
8 ORH 0.05
8 PVD 0.05
12 BDR 0.00
8 BDL 0.05
7 ALB 1.25
6 BGM 1.40
12 ISP 0.00
12 JFK 0.00
12 ABE 0.00
12 MDT 0.00
12 PHL 0.00
12 ACY 0.00
12 EWR 0.00
12 BWI 0.00
12 IAD 0.00
12 DCA 0.00
12 SBY 0.00
12 RIC 0.00
12 ORF 0.00
12 RDU 0.00
TOT 26.85

PWM and CON appeared to be low; however...after comparing the reported snowfall with hourly METAR liquid reports (P/////)...they seem reasonable.

Friday, April 13, 2007

Contest # 6 - Call for Forecasts


Who/d have thought there/d be a 6th Contest storm this year…but here we are looking at a rabid…late season nor’easter with a halfway decent snowfall potential for inland stations over NNE.

Given this storm is ~48 hours away from affecting the first forecast station …there/s a possibility this event will not live up to its advance NWP billing. Saturday/s 12z model runs may force the cancellation of Contest # 6.

Entry Deadline: Saturday...14 April 2007 @ 10:30 PM EDT.

Verification begins 12:01 AM EDT Sunday...15 April 2007 and ends 11:59 PM EDT Tuesday...17 April 2007.

Your forecast must be entered via the Contest/s web site. Follow the link to ‘Enter Storm Forecast.’

All forecasts will be posted to the NE_Wx Google Group by the Contest Administrator before 11 PM EDT Saturday...14 April 2007 and to the Contest web site by Sunday afternoon.

Forecasters will need to register once before entering...even if they were registered last year. Registration is simple...requiring only a user name and password. If you provide a valid e-mail address...a copy of your forecast will be sent to you immediately after your entry is submitted. Please ensure your browser is enabled to accept first-party cookies.

Updates and announcements are posted on the Contest/s web log.

< boiler plate >

The NE.Wx Snowfall Forecast Contest (NEWxSFC) is a multi-month event that continues into late March or early April. In general...contests are held whenever a decent... synoptic-scale storm rears its head and threatens at least a half-dozen forecast stations with more than nuisance snowfall amounts. Forecasters are called to post their 'storm total' snowfall predictions...on deadline...for 27 NWS / FAA observing stations scattered about New England and the Mid-Atlantic regions.

You can find the Contest Rules and additional information about Error Scoring... current monthly snowfall climatology from NCDC...daily CPC teleconnection indices... daily NESDIS N-Hemi snow cover...and NWS Daily Climate Bulletins (CDUS41) at the NEWxSFC web site.

< /boiler plate >

Contest # 5 - Day Two Leader Board


Based on Thursday and Friday/s CDUS41 bulletins and PNS reports.

Mitchel Volk
Raven
donsutherland1
TQ
shanabe
herb@maws
weathafella
ilibov
bruced39
Donald Rosenfeld

Still snowing at CAR.

Thursday, April 12, 2007

Contest # 5 - The Forecasts


Like Contest #4...this snowfall is an interior NE event where the forecast consensus places the +SN axis from BTV - CAR - CON - ALB - BTV with a max of 7.5" at CAR.

Forecasters: 10
Rookies: 2
Veterans: 8

Minimum STP: 32.6" (Shanabe)
Maximum STP: 65.8" (Donald Rosenfeld)
Average STP: 42.7"
Median STP: 39.8"
10th percentile: 33.9"
90th percentile: 55.8"

All individual forecasts have been posted to the NEWxSFC web site. Follow the link to 'Contest # 5 - The Forecasts'. The forecasts are ranked from minimum to maximum storm total snowfall (STP) for all stations.

Verification snowfall reports will be posted Saturday morning. They are considered preliminary and subject to change / challenge for 24 hours...if updated information becomes available. Final results will be posted on Sunday.