The analog guidance began in 2008 after science meetings between the Springfield, MO and St. Louis, MO NWS Forecast Offices and Saint Louis University. A plan was put together and the guidance started with the cold season and expanded to the warm season in the summer of 2011. The effort is completely driven by a grassroots movement since no outside funding or research grants have ever been awarded. Since November 2009, the CIPS analog guidance has been mentioned in over 300 AFDs by over 45 WFOs across the CONUS. The developers of the Analog Guidance, Dr. Charles Graves and Dr. Chad Gravelle, recently received the 2012 National Weather Association Dr. T. Theodore Fujita Research Achievement Award for "Revolutionizing forecaster decision aids and customer decision support services by creating the CIPS Analog Guidance for various high-impact weather events."
What are Historical Analogs and Why Use them in the Forecast Process?
Early Analog Definition: If the current state of the atmosphere resembles a previous state then the two are termed analogs, and for a period of time the current state may evolve in a similar fashion as the past state (Lorenz 1969).
Modern Analog Definition: Analogs are found using the perfect prognostic (“perfect prog”) approach. NWP forecast fields are used as a pattern-recognition tool in contrast to using analysis maps as a forecast tool (Root et al. 2007). This is the methodology behind the CIPS Analog Guidance. Take the upcoming forecast fields to find similar historical events and then extract the sensible weather impacts.
In the last few years, a shift has begun in the way that meteorologists forecast and warn society. An emphasis has been placed on impacts and the weather-related events that significantly affect people and their livelihoods. However, even for the most experienced forecasters this is difficult to accomplish with traditional meteorological forecasting techniques. The image below shows the snowfall from both the above 30-h GFS forecast and one of the most similar analogs. It's easy to see how this could be powerful information for weather partners and society.
The analog guidance can be used as a decision assistance tool for forecasters. Decision assistance tools can aid forecasters in identifying high-impact events and provide information to help understand the significance of the event. Some of the decision assistance methods are as follows...
- Give a conditional climatology of the forecast (i.e., given the selection criteria of an analog…this is what “typically” happens).
- Provide situational and forecaster awareness of similar events.
- Give confidence in NWP model output.
The first three bullets are the most common uses of the analog guidance, but when extreme and high-impact events occur the guidance can also assist forecasters in identifying them in the forecast.
The historical information gathered can then be used to provide impact-based decision support services since the historical impacts from NCDC Storm Data, COOP snow event maps, Unified Precipitation Dataset, Severe Storm Reports etc. have been extracted. If there is confidence in the forecast, the upcoming storm can then be communicated in a historical framework to the unscientific audience. This is powerful because society as a whole does not recognize or react appropriately to warnings. Those who have lived through the high-impact historical event now have a personal connection to the upcoming forecast. This approach of forecasting meteorological risk is the future of weather forecasting.
How Are Analogs Found?
The upcoming model forecast using the NAM or GFS output is used to search the 30-yr North American Regional Reanalysis (NARR) for potential analogs. The NAM is used with the 36 and 48-h forecasts and the GFS is used beyond 2 days. Analogs are found for a 3-month period (±45 days from current day) with a 6-h temporal resolution. This provides roughly 11,160 potential analogs (31 years, 90 days, 4 per day).
The fields used to assess the similarity between the forecasts and historical events are as follows:
300-mb Heights *
500-mb Heights *
850-mb Heights (weighted three times)
850-mb Wind Speed
Sea-level Pressure
850-700-mb Thickness
850-mb Temperature (weighted twice)
2-m Temperature
Precipitable Water
925-mb Mixing Ratio (weighted twice)
2-m Dewpoint
Correlation and mean-absolute error statistics are computed for each field against the forecast over the example domains below.
* 300-mb and 500-mb height field statistics are found on the large domain.
Once potential analogs are found, duplicate times are removed by choosing the “best” analog over a 24-h period. The analogs are then refined and ranked (using the analog time and ±12h from analog time) and the resulting 15 most similar analogs to the forecast are used to create impact-based products useful for impact guidance.
Real-Time Analog Guidance
The current webpage for the real-time Analog Guidance is located here. After a forecast and domain of interest is chosen, the statistics page is displayed with the top 15 most similar analogs. Top 15 analogs are ranked from first to 15th and this is roughly the top 0.1% similar historical events to the forecast. Table headers can sort the analogs by date, field, or score.
Below the statistics table are four buttons.
Analysis Guidance
The analog guidance button displays a gui that allows the user to analyze the representativeness of the top 15 analogs to the forecast. Height, wind, temperature, moisture, instability, and severe composite fields can be compared. In addition, mean standardized anomalies from certain fields that are based on the top 15 analogs can also be analyzed. The hypothesis is that if the synoptic-scale fields are similar, more confidence can be given to the impact guidance based on the top 15 analogs.
Impact Guidance
The impact guidance gui is one of the more powerful portions of the analog webpage. Here, impact guidance is available that is based on the top 15 most similar historical events. The categories are...
COOP Snowfall
Severe Thunderstorms
Unified Precipitation Data
Surface Temperature
Surface (other)
Aviation
Fire
Mean, median, and probability of exceedance threshold products are available to diagnose the meteorological risk of the above weather threats. This information can be used in many ways which is why feedback to the guidance developers is always encouraged.
Thumbnail Guidance
The thumbnail guidance can be used to quickly find the most similar individual analogs to the forecast and assess the range out outcomes for some sensible weather fields.
Individual Analogs
Individual analogs can be accessed by clicking on the analog date/time through either the statistics table or the thumbnail pages. It is important to use the impact guidance and the individual analogs together to hone in on potential threats. Often...the synoptic-scale pattern can be very similar to the upcoming forecast in many of the top 15 analogs. By analyzing these most similar analogs, a range of threats can be determined. For example...if the upcoming model forecasts suggest a snow event, the user can diagnose the most similar analogs to the forecast and determine if their is a large range of snow amounts from the historical events. If some events suggest 6-8" of snowfall and others suggest 2-4", the user can determine the differences in those events. It's possible the 2-4" events were warmer and had a lower snow-to-liquid ratio. If the model forecasts suggest a more efficient thermal profile for snow production that is more similar to the 6-8" events, then the 2-4" events could be discounted. This is one of the more powerful aspects of analyzing the individual analogs.
Some Things to Remember
- A “perfect” analog does not exist.
- Historical meteorological data from the North American Reanalysis Dataset is only available back to 1979.
- Quality of supplemental impact data deteriorates further back in time.
- Only a few, good quality analogs may exist for anomalous and record breaking events.
- Analog guidance is based on the atmospheric conditions forecast by the model.
- Some meteorological patterns are more “predictable” and easier to identify than others.
- This is only one approach to find analogs.
Forecaster Feedback
“The greatest strength is that it gives historical context to weather events. Sometimes…not always…you can utilize this historical context when providing a decision support briefing.”
“If 8 of 15 analogs on the CIPS Guidance are events that push further east and yield nothing for the Northeast…this gets forecasters thinking. It makes them ask questions like ‘What is different about those 8 systems than the one the models are forecasting?' Perhaps those 8 analogs have the H5 trof in a different location or a low level jet displaced etc. It is a learning process.”
“When using their personal analogs, forecasters tend to generalize the similarities in systems when it is the details that often separate significant events from routine ones. This is where I believe analogs can make some of the largest gains this field has seen in some time. Imagine having a digital database of all the details forecasters can forget of all past storms and the parameters that went with them and more importantly the results.”
“Using models as an analogy, it would be like going from a pure advection model to seeing multiple ensembles on AWIPS, all of this in a short period of time. This is how important and beneficial digital analogs can be and why this is our future as much or more as mesoscale modeling or ensembles are."
Please feel free to leave comments how you have used historical events in the forecast process.
No comments:
Post a Comment