On Friday, March 16, 2018, I provided an updated forecast to a client stating that on Tuesday, March 20, there was a threat for 3" or more snow in northern parts of their area of concern, a threat for minor to moderate icing (thin glaze to as much as 0.10-0.20" possible) due to freezing rain in northwestern parts of their area, and a threat for 1" or more rainfall in southeastern parts of their area. All three of these concerns ended up happening in the areas I highlighted.
At the time, looking five days out, there remained considerable uncertainty about which precipitation type would be most dominant. In addition, factor in the late March timing, meaning lighter snow rates would be slower to accumulate (if at all). The heavier snow rates, 1-2"/hr, would be sufficient. However the occurrence of such rates would be within localized bands, which are difficult to pinpoint where they would set up even 24 hours ahead of time, but nearly impossible five days ahead. From a prediction standpoint, I feel as though I made a good forecast given how far in advance I alerted the client to potential issues of concern to them.
Then I started thinking more deeply about what decisions, if any, are practical to make given the level of confidence. Do people change their staffing plans for a low confidence forecast in a high impact event? Do they really need that level of precision five days out, or could they wait until two days before the storm? How much money do they waste if something changes and it's not as bad as forecast? How do you prepare for a high-impact event that has very short lead time, such as a severe squall line? Is there even a solution to these issues?
At the time, looking five days out, there remained considerable uncertainty about which precipitation type would be most dominant. In addition, factor in the late March timing, meaning lighter snow rates would be slower to accumulate (if at all). The heavier snow rates, 1-2"/hr, would be sufficient. However the occurrence of such rates would be within localized bands, which are difficult to pinpoint where they would set up even 24 hours ahead of time, but nearly impossible five days ahead. From a prediction standpoint, I feel as though I made a good forecast given how far in advance I alerted the client to potential issues of concern to them.
Then I started thinking more deeply about what decisions, if any, are practical to make given the level of confidence. Do people change their staffing plans for a low confidence forecast in a high impact event? Do they really need that level of precision five days out, or could they wait until two days before the storm? How much money do they waste if something changes and it's not as bad as forecast? How do you prepare for a high-impact event that has very short lead time, such as a severe squall line? Is there even a solution to these issues?
Comments
Post a Comment