Using Analogs as Medium Range Guidance for Organized Snow Events
Note - much of the text below was provided by the
CIPS Winter Storm Analog Guidance page.
Researchers at the Cooperative Institute
for Precipitation Systems (CIPS) along with the
NWS St. Louis, MO and the
NWS Springfield, MO are
engaged in a project to show how the knowledge of past events that exhibit similar characteristics
to the current forecast can assist forecasters with a range of potential scales and intensities of the
upcoming event. The project is entitled "Using analogs as medium range guidance for organized snow events."
The goal of the analog forecast approach is not to make a forecast; but to provide
medium-range guidance for events by using a historical dataset. In addition, a forecaster can
quickly gain historical experience and become familiar
with the meteorological patterns associated with certain events. The analog forecast
approach can be applied to any meteorological event as long
as a control run can be created.
The methodology below is used by the CIPS Winter Storm
Analog Guidance Project which currently uses 36, 48, 60, and 72 hour forecasts from the 40 km GFS.
The basis for analog guidance is to search a climatological dataset for maps
that resemble the current forecast, and then assume that the atmosphere will evolve
similarly to the historical analogs.
Search the 28 year North American Regional Reanalysis (NARR) dataset against the model
forecasts provided by the 40km GFS (GFS212) for potential analogs based on the constraints below:
6 months over the winter season (October - March)
Remove “duplicate” times for the same analog event by choosing the best analog over a 24-h period.
6 hour temporal resolution
Potential database of 20,384 analogs (28 winters, 6 months, 4 per day)
Refine and rank the resulting analogs to create products that are useful for medium range guidance.
Determining what constitutes an analog is done statistically using the following techniques:
During the first pass through the NARR dataset, the statistics are computed on a large domain (REGN)
and then on a smaller domain (MESO). If certain thresholds are not exceeded, the date/time is not considered
a potential analog.
Mean Absolute Error
To reduce the approximately 20,000 potential analogs, threshold values were determined
based on the control run for the following fields:
300 HGHT COR 0.85 REGN
Once the approximately 20,000 potential analogs are reduced, “duplicate” times are removed.
“Duplicate” times occur due to the variability in system speed (e.g., a slow historical
system may exhibit similar patterns to the forecast over a longer period of time).
500 HGHT COR 0.83 REGN
850 TMPC MCOR 0.88 MESO
850 TMPC MMAE 3.8 MESO
PMSL MCOR 0.83 MESO
850 HGHT MCOR 0.70 MESO
The best analog is found over a 24-h period by using the following formula:
After the potential analogs are reduced, the program is
rerun to find statistics on the following variables:
300 HGHT COR REGN
After new statistics are determined, a results score is computed using the following formula:
500 HGHT COR REGN
700 FRNT COR MESO
850 HGHT COR MESO
850 TMPC COR MESO
850 TMPC MAE MESO
850 FRNT COR MESO
850 THTEADV COR MESO
2m TMPC COR MESO
2m TMPC MAE MESO
PMSL COR MESO
PWTR COR MESO
850HGHTCOR*3 + PMSLCOR*2 + SUM(COR) - SUM(MAE/3)
Finally, in order to catch possible system propagation, statistics are computed for ± 12 h from the
time of the best analog using the matching forecast (the 48 hour forecast is used in the example below):
(-12 h Analog: GFS 036h FCST, Analog: GFS 048h FCST, +12 h Analog: GFS 60h FCST). Using statistics from
the ± 12 h times, results scores are also computed.
The final analog rank is determined by using the results scores in the following formula:
AVERAGE(m12ANALOG, ANALOG, p12ANALOG)
Example Analog for this Event
The CIPS Heavy Snow Analog Guidance for 12 UTC on 2009/12/19 across the
East Coast 2 domain
was examined for this event and a brief review of the process and
data is shown below. The analog was based off the 48 hour forecast
from the 40 km GFS (212 grid) valid at 12 UTC on 12/19. The
four panel GFS 48 hour forecast indicated
a strong subtropical jet over the eastern Gulf of Mexico and Florida along with the
exit region of the polar jet over the New England states and a well developed surface low
off the North Carolina coast with surface pressures less then 992 mb.
The CIPS Heavy Snow Analog Guidance Table for the East Coast 2 domain from the GFS 40km
12 UTC 20091217 48 hour forecast is shown below (click to enlarge). Note the
statistics for GFS212 20091217/1200F048 forecast:
Total number of potential analogs: 21140
Number of potential analogs that did not exceed set thresholds: 21005
Number of potential analogs that exceeded set thresholds: 135
Number of distinct analog events: 73
Percentage of potential analogs that did not exceed the set threshold at each step of the analog process:
300-mb HGHT correlation: 45.08 %
500-mb HGHT correlation: 4.80 %
850-mb TMPC correlation: 25.37 %
850-mb TMPC mean-absolute-error: 6.31 %
PMSL correlation: 15.82 %
850-mb HGHT correlation: 2.61 %
The 1983/02/11 18 UTC analog had
the greatest final analog rank based on the averaging of that forecast along with
the statistics from 12 hours prior to and after the 1983/02/11 18 UTC forecast. Examining the
CIPS Heavy Snow Analog Guidance Table shows that
this event had the greatest similarities for the 300-mb HGHT correlation, the 500-mb HGHT correlations,
and persistence with slightly reduced similarities for the lower tropospheric fields such as PMSL correlation and
the 2mTMPC correlation.
The CIPS Heavy Snow Analog page provides
analysis maps of the COOP snowfall from the various analogs along with some
probabilities based on the analogs. The COOP snowfall analysis map from the CIPS Heavy Snow Analog page for the 1983/02/11 18 UTC analog is
shown below. In addition, the mean and median of the top 15 analogs are also provided on the CIPS Heavy Snow Analog page.
The CIPS Heavy Snow Analog page also provides images displaying measures of center and spread guidance for
various fields including mean sea level pressure, 850 mb height, 500 mb height,
300 mb height, 850 mb temperature, and 2 meter temperature.
Some of the more insightful products are the
probabilistic forecasts of COOP snow greater then various thresholds. The
probabilistic forecast of COOP snow greater then 2, 4 or 6 inches based on the
top 15 analogs is shown below. Additional probabilities for COOP snow amounts greater then
8 inches are also available along with probabilities of snow to liquid ratios
from COOP observations greater then 10:1, 12:1, 14:1, and 16:1.
Utility of Analogs for this Event
Most of the forecast shift preceding the event utilized analogs in one form or another.
Analogs were referred to by HPC in the PMDEPD products and local forecasters took advantage of
the new Medium Range Guidance for Organized
Snow Events from CIPS.
The use and application of analogs, in particular
the CIPS project are growing rapidly, most likely because they are becoming easier to incorporate into our decision-making given
the advances in technology and communication. Many operational forecasters have long gravitated to their
own "analogs", or personal experiences, when preparing forecasts anyway.
These statistical analogs are simply a more objective way of drawing on past cases.
During the December 18th winter storm, a CIPS
graduate student was monitoring communications in the
in the NWSChat and they may do more monitoring for potentially big storm in the future.
During the event he let us know that the strong analog case for that event (February 10-12, 1983)
was amazingly persistent (popping up run after run) and that it was ranked by NCEP as the 10th
biggest snowstorm with 45+ deaths.