POP-Climo Training National Weather Service Western Region

1
POP-ClimoTrainingNational Weather Service
Western Region

• July 2006
• Developed by Western Region POP-Climo Team under
  direction from the WR Digital Service Management
  Team

2
Section 1 Goals of Training

• Improve the accuracy and usefulness of WR NDFD
  POP grids through the application of
• probabilistic forecasting
• verification feedback
• climatological POPs.

3
Section 2 Our Current Situation, Is It Cause
For Concern?

• National NDFD verification stats indicate that a
  significant dry bias exists in Western Region by
  day 5.

4
Section 2 Our Current Situation, Is It Cause
For Concern?

• A human low-POP bias is contributing to lower
  scores for the NDFD grids, especially for longer
  forecast ranges.
• Our partners and customers are not well served
  when a recognizable and correctable bias is not
  removed from one of our forecast systems
• Through the correct use of climatological PoPs
  and MOS, this low-PoP bias can be significantly
  reduced or even removed.

5
Section 3 What Is A POP?

• 12 hour NDFD POP (NWS Directive 10-506 Digital
  Data Products/ Services Specification)
• 12-hour Probability of Precipitation (PoP12) is
  the likelihood, expressed as a percent, of a
  measurable precipitation event (1/100th of an
  inch) at a grid point during the valid period.
  The 12-hour periods in NDFD begin and end at 0000
  and 1200 UTC.

6
Section 3 What Is A POP?

• Climatological POP
• Represents a long term average POP, which
  includes both wet and dry periods or upper level
  ridges and trough patterns. Conceptually, it can
  be compared to a long term temperature average.

7
Section 5 The Transition From Deterministic To
Probabilistic Forecasting

• As lead time increases, probabilities
• tend to be less extreme
• the use of the very low and especially the very
  high values is less frequent
• Eventually (5-10 days) the range has shrunk to
  the single value of the climate frequency or
  climate POP.

8
Section 5 The Transition From Deterministic To
Probabilistic Forecasting

• One effective way to estimate the uncertainty for
  a given forecast, and the forecast period past
  which it is no longer reasonable to make
  forecasts deterministically, is to use ensembles.
  
• Ensemble forecasting yields information about the
  magnitude and nature of the uncertainty in a
  forecast (Wilks, 1995).
• For more information about the best way to use
  ensemble forecasts, visit the following excellent
  training sites
• http//www.hpc.ncep.noaa.gov/ensembletraining
• http//meted.ucar.edu/nwp/pcu1/ensemble

9
Section 6 What Is A Good POP?

• Forecasters should strive for POPs that are
  accurate and unbiased
• So far as the scientific problem of weather
  forecasting is concerned, the forecasters duty
  ends with providing accurate and unbiased
  estimates of the probabilities of different
  weather situations. (Brier, 1944, Hughes, 1980)
• POPs should reflect both uncertainty and
  knowledge about future weather so customers can
  make appropriate decisions based on risk. (Wilks,
  1995)
• POP grids should reflect weather conditions,
  e.g., elevation differences should be much more
  distinct during convective or upslope
  precipitation events and less distinct during
  stratiform periods.
• Instead of trending towards 0 during a period of
  forecast uncertainty in the longer term
  forecasts, trend towards climatology (unless
  climatology is represented by zero).  This does
  not mean simply populate directly with
  climatology, but as a rough starting point. 
  Using collaboration with surrounding offices,
  trend above climo if the models are hinting at a
  long wave trough, or below climo if the models
  are hinting at a ridge.

10
Section 7 Final Recommendations

• Use Climatological Pops as a first-guess field
  when there is uncertainty, especially in the
  longer range forecast period.
• The range of usable POP values narrows with
  forecast time, eventually shrinking to the
  climatological POP at some time in the future.
• Use references to Climatological PoP grids in
  12Planet chats to make collaboration more
  efficient.
• MOS and Climatological POPs are not a perfect
  forecast. However, significant deviations in the
  longer term should be limited to known model bias
  periods and other known local confidence factors.
  
• Ensemble forecasts provide a means to help
  determine confidence.
• Know your skill, verify your forecasts, adjust
  accordingly.