Forecast Evaluation for snow event: February 29, 2012- March, 01, 2012
Time of evaluation: 9:20AM Thursday March 01, 2012
Storm type: Two phase Miller B Type
10PM Forecast Probabilities for February 29, 2012: 0%D, 0%C, 40% ED
Outcome: Early Dismissal
10PM Forecast Probabilities for March 01, 2012: 30%D, 10%C, 0%ED
Outcome: Delay initially, upgraded to Closing
Forecast rating (1-5) Phase 1: 5 (Excellent)
Forecast handling (1-5) Phase 1: 4 (Good)
Forecast rating (1-5) phase 2: 1 (Poor)
Forecast handling (1-5) Phase 2: 3 (Fair)
Phase one of a two phase system was handled very well. The focus period of the first phase was the morning/afternoon of Wednesday February 29, 2012. Vigorous low pressure over the upper midwest spread warm air advection well ahead of main low. Main low drifting eastward, but nearly stationary at the time. Main region of precipitation was located over Illinois and
Indiana Tuesday evening. Scattered radar signatures ahead of main low, but surface reports indicated that no precipitation was associated with these areas. Straight line, extrapolation of the main precipitation had it through the Kingston area around 9am Thursday. Forecasted snow to begin during the 10 o'clock hour, possibly a bit earlier- - compromise between radar and NWS forecast of 30% sn at 10am. Expected steadier snow during the 12PM hour with moderate snow possible as the precipitation became established. Accumulations to begin surpassing an inch by 2pm. Total accumulations through 8pm forecast to be 3 to 5 inches. Winter weather advisory 12pm Wednesday 12pm Thursday.
Overall, the forecast was accurate. There was little historical support for a cancellation, as early dismissals had been recorded when snow began at or prior to 8am. However, it would have been wise to note a slight chance of a cancellation (10%), in the event that the snow set in earlier than 9/10pm- - that the radar signatures ahead of the low might have begun to precipitate- - also the 3 to 5 inch forecast was far greater than the other pre-recorded EDs (generally in the 1-2 inch range). Also slight chance closing due to over district anticipation. The 40% ED was respectable (opposed to 60) because a delay in arrival time of 2 hours could have resulted in no pre-3PM accumulation. Early Dismissal probability upped to 60% at 6am. ED called at 630am. Sn started at 10am. Grass accumulations started around 11pm. road accumulations by 12pm.
Phase two of a two phase system was handled decently, however the associated probabilities were poor.
Overall, the forecast was poor. The majority of plowed roads were in good condition by 8pm, but side roads were likely still messy. rationalized that- in the event of no additional accumulation- there would be time to clear all roads, thus normal school at that point. Light drizzle had been falling since ~5/6pm. Correctly extrapolated energy over kentucky to miss MHV- - forecast lull after 4pm-9PM. Post-12am forecast- very poor. At the 10pm update the main low was over western michigan (Lower) moving east. proceeded by developing convection over ohio. straight line extrapolation had the convection through MHV between 2am and 4am,accompanied by a portion of the ULL. Based on precipitation around ULL and rate of circulation, expected most of associated precip to trek north of the MHV, thus main period of precip 2am-~7am. Temps were just above freezing at 10pm- - drizzle. Expected convection to result in light to moderate precip- - continued drizzle/light rain with wet snow in heavier convection, but predominantly rain- - thus only slushy accumulations incapable of significant road impact. NWS hourly grid showed rain mixed with sleet and up to .2" rain & .7" sleet. NWS forecast warm noose around low to gradually receed- - temps 33-34 overnight. secondary low development after 12am was seen as crucial for baroclinicity. high pressure to the northeast expected to provided cool air flow. based on NWS forecast, and given anticipate lighter precip. went with predominantly rain scenario. - - didnt completely understand/have confidence/ in the secondary low roll, and focused mainly on direct/developing convective returns over ohio- - underestimated /largely dismissed/ additional convective development- - mistake. Mistake realized after convection started to really deepen after 1030pm- - too late. Forecast was generally correct in proper extrapolation, but far too dismissive /biased/ regarding additional storm development.
Probability forecast of 10C 30D was poorly constructed. Did not expect significant morning accumulation impact, but there was a lot of doubt/discrepancy in the forecast- - wasnt reflected in probabilities. Considering just radar returns alone- - the probabilities did not allow for any system organization increase- - probabilities were overly restrictive. Potential for increased organization- - heavier precip/more sleet- - should have at least been noted, especially since even though ohio convection was established it was still in the process of developing. Also- - statements over the northeastern PA indicated the possibility of heavy thunder- - dismissed in forecast due to anticipated Catskills interference- - probabilities left no room for lack of interference. Also- - not evaluated at the time of forecast, but relevant- - numerous/frequent lightning returns over ohio- - indicative of potentially heavier precip across MHV. Also, the affects of the ULL its self were dismissed- assumed to pass northward/impact after sunrise- - potential for atlantic enhanced moisture not considered. Main focus, though, should have been heavy/developing convection over ohio, and its potential to maintain punch through MHV- - overlooked/underestimated. Much of the predominintly rain mentality resulted from perception of poor handeling by NWS during evening lull, and rain bias in NWS hourly grids- - ohio convection should have overrode this- - or at least the possibility. Probabilities should have read 40% delay, 30% cancellation. Outcome: Delay issued at 5:45am, upgraded to cancellation at 750am.
- Important to keep in mind for future events that 30% (cancellation) still implies an event not occurring, just as 10% does, but 30% is less restrictive in doing so- - probabilities were to restrictive, and ultimately failed as a result.
- March 1 was a highly publicized event due to the proceeding early dismissal. Without affecting forecast quality, it is especially important to not be restrictive during high profile events, as forecasts are relied on more heavily.
- Don't use 0% when there is a slight percentage… use 10%- - it implies 0% but is less restrictive.
- Don't dismiss the potential for additional development, even when radar suggests minimal impact.
Overall storm handling: 3 (Fair)