Welcome to American Weather
Roger Smith

2017 North Atlantic Tropical Storm / Hurricane Forecast Contest

35 posts in this topic

(Aug 31, 2359z) _ The discussion posted earlier today has been edited as Irma rapidly reached major hurricane intensity before August ended (even in z time although I use EDT as the actual determinant of months). So this discussion now reads correctly for the final count which is 4/4/2. The year is now at 9/4/2.

 

August will finish 4/4/2 -- no alternate scoring table is therefore needed (yet). An alternate scoring table will be provided if any late September or October systems reach higher intensity levels in following months.. We've had this discussion almost every year for six or seven years now, and of course there is no easy (contest related) answer, for instance, theoretically Irma could have been a hurricane in both August and September and a major only in September -- but that was cut off by the rapid intensification -- however, our alternate scoring does not add duplicate status entries, just any new status. I've never been able to determine absolutely if the NHC keeps track the same way as system (a) or (b), or if they keep track at all. But plan (a) is our official scoring system anyway.

Scores for August are based on a maximum of 12 points and full error deductions are used (June-July had half-error deductions from their smaller base scores).

For 4/4/2 the scores are as follows ... this table only shows forecasts that were actually made in the contest. For example, nobody predicted 3 3 1.

Interesting that nobody went as high as 4 hurricanes or 2 major hurricanes. The scoring system as explained in post 1 reduces scores from the base of 12, by the average of error and error squared in each category. An error of 1 is a reduction of 1, an error of 2 is a reduction of 3, and an error of 3 is a reduction of 6. 

If by any chance potential T.D. 10 (Carolinas to south of New England) was later declared to be a tropical storm and/or hurricane in retrospect, but during the calendar year 2017, then that would count in the contest too and August could then theoretically be scored from 5/4/2, or 5/5/2. If they made this or any similar declaration after January 1st it is null and void for the contest (same applies to seasonal total) although I might post alternate scores if it was before the 2018 season.

 

Aug Forecast __ score _____ forecasters

4 3 1 ________ 10 _______ RJay, Stebo, Rockchalk83, Roger Smith, Ser Pounce

4 2 1 _________ 8 _______ Dmillz25, Matt Petrulli, NCForecaster89, NWLinnCountyIA, "Normal"

4 2 0 _________ 6 _______ Windspeed

3 2 1 _________ 7 ________NHC mid-range (see CSU below), Kurzov, NJwx85, ldub23, Consensus

3 2 0 _________ 5 _______ Yoda

3 1 1 _________ 4 _______ CSU (arbitrary based on scaled seasonal forecast)

3 1 0 _________ 2 _______ radarman

2 2 1 _________ 5 _______ Kalasea

2 2 0 _________ 3 _______ pcbjr

 

3.4 2.2 0.8 ____ 7.7 _____ Contest average

 

=============================================

After September, will post a table of partial total scores based on June to September, plus a reasonable October and Nov-Dec outcome applied to seasonal and future monthly scores. Then that can be adjusted as October and Nov-Dec realities roll in. Much depends on September which gets 16/50 monthly points, then October gets 10/50. Nov-Dec we are back to half-error deductions from a measly 2 points. 

 

 

 

Share this post


Link to post
Share on other sites

(updated Sept 7 21z)

Jose and Katia have both reached hurricane intensity and Jose has reached major hurricane status, so this brings the annual count so far to 11/6/3 and September already at 2/2/1. Although all of Irma's rampage will be in September, it did reach major hurricane intensity on August 31 and this means that its major hurricane status does not fall in the September portion of our contest. 

 

Share this post


Link to post
Share on other sites

(edits on Sept 24, 27) _ Lee went dormant for several days and then regenerated, and has attained hurricane intensity. And then it attained major hurricane strength. 

The annual count is now 13/8/5. September is now at 4/4/3.

If there's no sign of further activity by Saturday, I will post September scores. 

 

Share this post


Link to post
Share on other sites

The September count is 4/4/3.

September scores are from a base of 16  for a perfect forecast.

The actual scores for forecasts made  (those not in the table are not shown here):

Forecast _____ Score ______ Forecaster(s)

6 4 3 ________ 13.0 _______ Windspeed

6 4 2 ________ 12.0 _______ rockchalk83, dmillz25

5 4 3 ________ 15.0 _______ Roger Smith

5 4 2 ________ 14.0 _______ NJwx85

5 4 1 ________ 12.0 _______ NCForecaster89

5 3 2 ________ 13.0 _______ Ser Pounce, ldub23

5 3 1 ________ 11.0 _______ NHC mid-range, Contest Normal

5 2 1 _________ 9.0 _______ Matt Petrulli

4 3 2 ________ 14.0 _______ RJay, Yoda, Consensus

4 2 1 ________ 10.0 _______ Stebo, NWLinnCountyIA, CSU

3 2 1 _________ 9.0 _______ kalasea, radarman, pcbjr

3 0 0 _________ 0.0 _______ Kurzov

____________________________________________________________________________

Following is a provisional total scoring table based on 3 2 0 in October (upgraded to 3 2 1 on Oct 14) and 1 0 0 in Nov-Dec.

This will be edited whenever (a) better estimates can be made or (b) reality confirmed is different at end of month,

Seasonal score was 13/8/5 when the table was posted in early October, and would finish 17/10/6 if these estimates verify. Of course, they may not and your score could improve or worsen as a result. No scores are given for April which was a "given" for all entries.

(ranks are among 17 entrants, based on these assumptions, and ranks for other scores such as normal, consensus, NHC and CSU do not affect ranks of the 17 entrants).

 

FORECASTER ____ Season  __APR__ JUN ___ JUL ____ AUG ____ SEP ___ OCT ___ NOV-DEC  __ TOTAL

 

Kurzov ________ 22 8 2_ 22  _100_ 110_3.0 _220_4.5 _321_ 7 _300_ 0 _211_ 8 _ 100_2.0 __ 46.5 (15)

Windspeed _____21 11 5_38 _100_ 210_3.5 _320_4.0 _420_ 6 _643_13 _422_ 8 _100_2.0 __ 77.5 (8)


Roger Smith ____19 13 6_41 _100_ 110_3.0 _210_5.5 _431_10 _543_15 _432_7 _210_1.0 __ 82.5 (3)

Kalasea _______ 18 13 6_43 _100_ 100_3.5 _421_2.5 _221_ 5 _321_ 9 _ 443_ 3 _330_0.0 __ 66.0 (13)

rockchalk83 ____18 10 5_48 _100_ 000_2.5 _211_5.0 _431_10 _642_11 _421_ 9 _100_2.0 __ 87.5 (1)

dmillz25 _______17 7 3 _38 _100_ 200_4.0 _310_5.0 _421_ 8 _ 642_11 _200_ 5 _000_1.5 __ 72.5 (9)

Stebo _________16 9 4 _45 _100_ 200_4.0 _321_3.5 _431_10 _421_10 _211_ 8 _110_1.5 __ 82.0 (4)

NJwx85 _______ 16 9 3 _42 _100_ 100_3.5 _210_5.5 _321_ 7 _542_14 _320_ 9 _100_2.0 __ 83.0 (2)

NCforecaster89 _ 16 8 3 _40 _100_ 100_3.5 _210_5.5 _421_ 8 _541_12 _211_ 8 _100_2.0 __ 79.0 (7)

 

Consensus* ____ 16 8 3 _40_100_ 100_3.5 _210_5.5 _321_ 7 _432_14 _321_10 _100_2.0 __ 82.0 (4)

 

Ser Pounce _____15 9 3 _40 _100_ 100_3.5 _210_5.5 _431_10 _532_13 _220_ 8 _000_1.5 __ 81.5 (5)

 

Contest Normal _ 15 8 3 _38 _100_ 100_3.5 _100_5.5 _421_ 8 _531_11 _321_ 10 _000_1.5 __ 77.5 (8)

 

RJay __________ 15 8 3 _38 _100_ 100_3.5 _210_5.5 _431_10 _432_14 _210_ 7_100_2.0 __ 80.0 (6)

NWLinnCountyIA _15 6 3 _31 _100_ 100_3.5 _100_5.5 _421_ 8 _421_10 _321_10 _100_2.0 __ 70.0 (10)

Matt Petrulli ____ 15 6 2 _27 _100_ 100_3.5 _210_5.5 _421_ 8 _521_ 9 _210_ 7 _000_1.5 __ 61.5 (14)

.NHC (mid-range) 14 7 3 _32 _100_100_3.5_ 100_5.5 _321_ 7 _531_11 _321_10_000_1.5 __ 70.5 (10)

Yoda __________ 13 8 3 _31 _100_ 000_2.5 _210_5.5 _320_ 5 _432_14 _321_10 _000_1.5 __ 69.5 (11)

ldub23 _________13 8 3 _31 _100_ 100_3.5 _110_5.0 _321_ 7 _532_13 _220_ 8 _000_1.5 __ 69.0 (12)

radarman _______12 5 2 _10 _100_ 100_3.5 _211_5.0 _310_ 2 _321_ 9 _210_ 7 _000_1.5 __ 38.0 (16)

pcbjr __________ 11 5 2 _ 4_ 100_ 000_2.5 _000_4.5 _210_ 0 _ 321_ 9 _421_ 9 _100_2.0 __ 31.0 (17)

 

..... CSU ________11 4 2 _ _100_ 000_2.5 _100_5.5 _311_ 4 _421_10 _210_ 7 _000_1.5 __ 30.5 (17)

_______________________________________________________________________

* Consensus is derived from means for 17 forecasts not incl NHC, CSU, Normal, and the means are listed below to one decimal (rounded in table above)

Scores for mean _______  42.3 __ --- ____ 2.9 ____ 5.4 ____ 7.7 ____ 12.7 ___9.3 ___ 1.8 ___ 82.1

Mean _______ 16.0,8.4,3.4 _ 100 _ 1.0,0.2,0 _ 2.1,1.1,0.2 _ 3.4,2.2,0.8 _ 4.4,2.8,1.6 _ 2.8,1.6,0.8 _ 0.8,0.3,0

Share this post


Link to post
Share on other sites

(original post) With Ophelia likely to reach hurricane status, the count for October would then be 2/2/0. The seasonal count will then become 15/10/5.

I had posted a provisional scoring table earlier (previous post) and today I revised that slightly with the assumptions of one more named storm later in October, a count of 3/2/0, and the consensus value of 1/0/0 for Nov-Dec. That would give us an annual total of 17/10/5.

This provisional scoring table will be adjusted if necessary from later developments or lack of developments.

Basically, the scoring can improve for those of us who have room left in our October and seasonal forecasts to handle two more storms in October and in some cases more hurricanes and one major. Differences in November forecasts are slight and that component is only worth a maximum of 2 points. However, anybody has the option of adjusting their Nov-Dec forecast if they see any advantage in doing so (up until Nov 1 06z). 

(added Oct 14) ... Ophelia has reached cat-3 and thus the count for October is 2/2/1 and the seasonal count is 15/10/6. Changes in the provisional scoring table will be made on assumptions of a final October 3/2/1 and seasonal 17/10/6. 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.