Thanksgiving Rain Estimates Bad?
Blog reader David writes:
David, thanks for keeping me honest. I'll admit I didn't have much time to check that graphic before I published it in the blog, due to Black Friday duties. I did a quick check against local Doppler estimates and it seemed on target. But as you noted, they can be wrong too. After examining further data, I believe that it was inaccurate, by nearly a factor of two (see below). Had I examined it more closely, I wouldn't have posted it. But AccuWeather.com posts it every hour so here's a lesson on interpreting Doppler-Estimated precipitation maps:
The map I posted does seem quite overdone, saying that over 4 inches fell over much of the area, with maximums over 8 inches. I suspect that the map is nearly twice as high as it should be. In reality, I would say that around 2 inches of rain covered a wide area, with maximums over 4.70 inches. Here's my thinking on what went wrong and why.
1. LOCAL DOPPLER ESTIMATES: The "Storm-Total" Local Radar Doppler Estimates, which are generated by the NWS's NEXRAD 88-D Doppler Radars, indicated the following:
Albany: Wide Area of 2.0"+, Maximum of 3.5 - 4.0" over CT ("Max" Incorrect*)
Boston: Wide Area of 1.5"+, Maximum of 3.0-3.5" in CT, RI, MA (Max 5.9" Offshore)
NYC: Wide Area of 1.5"+, Maximum of 3.5" - 4.0" Over CT & RI ("Max" Incorrect*)
Philly: Wide Area of 2.0"+, Maximum of 3.0" - 4.0" On Long Island (Also 5.2" Off DE Coast)
I would say that these Doppler estimates were fairly accurate, though when one compares the maximums to gauge maximums, they come in a little low in the maximums. The rain in this system was very dense, like a Tropical Storm; if the NWS didn't change their radar "Z/R" (Height/Range relationships) to account for that, the estimates might be a little low. Normally they announce "Z/R" changes in their NEXRAD Outage Messages but a quick scan did not see anything from that day.
*The "Max" stated on Doppler Estimates is sometimes inaccurate due to ground clutter which is misinterpreted as heavy rain. This is especially bad in major cities and mountainous areas. Make sure the max amount corresponds to the highest color, and be on the lookout for false echos which will appear as sudden, tiny color changes.
2. THE STATE COMPOSITE DOPPLER-ESTIMATES:This map, the one I posted in my blog, is produced here at AccuWeather and adds up the Hourly Doppler Estimate graphics for 24 hours. I picked the 24-hour range which had the maximum extent of heavy precip amounts. Sometimes this map is off by a considerably amount; I believe it to be a bug in the system and I've yet to get anyone here to address it -- but Gauge Adjusted Precip is a similar and much more accurate product which will be available soon (see below). In any case, when you compare the state composite to the Locals discussed above, it seems 2 inches high off the coast of Mass. and 4 inches high over parts of CT & RI. Normally I catch this and simply don't publicize the image.
3. THE GAUGE ADJUSTED PRECIPITATION MAPS: The NWS is currently producing some maps which combine Doppler-Estimated information (as AccuWeather does), but take it one step further and adjusts the map based on actual rain gauge reports. AccuWeather will have this data on our state-level maps next month, for
Pro & Premium users.
How many rain gauges are taken into account? The NWS doesn't say, and I would imagine that it varies greatly by state. It sounds like it includes the Cooperative Observing Program but I'm not sure what else. It probably does not include the ROMAN database, even though it should (more on that below). So, what these maps show for the area, for a 7-day rainfall total, is, in theory, more accurate than the Doppler Estimates, but in this case seems a bit low, never showing more than 4 inches, even though some ROMAN gauges indicate that (see below). I also posted some additional gauge-adjusted maps (1-day totals and mid-Atlantic maps).
4. THE ACTUAL RAIN GAUGE DATA: As noted above, since we're not sure what's going into the gauge-adjusted maps, it's impossible to see the actual data (even COOP data is difficult to obtain online). Enter our
Pro site, from which I obtained a 3-day map of total rainfall for the New Jersey and Massachusetts areas. These stations are quality controlled by AccuWeather climatologists and generally include all the NWS official climatic stations (some COOP). You'll see amounts such as 3.56" and 3.63" on the latter map, confirming the highest amounts seen on the Local Doppler-Estimated maps.
I also published the old DIFAX maps which are official NWS stations, but really too crowded to make much out of. More important is this spreadsheet from ROMAN (The Government Mesonet) which includes amateur and unofficial stations. You'll see there the highest gauge amounts we've encountered thus far: 4.74", 4.73" and 4.17".
