NOMADS Global Ensemble Weather
Prediction Forecast Probabilities
Table of Contents
- Introduction to Global Ensemble Forecast Probabilities
- How to use the Prediction Probability tool
Introduction to Global Ensemble Forecast Probabilities
One of the most powerful improvements to Numerical Weather Prediction (NWP) in many years is ensemble forecasting. The NOMADS Global Ensemble Weather Prediction Probability page allows users to interrogate
the entire NCEP Global Ensemble model suite with a single request. Ensemble models use slightly different initial (starting) conditions that reflect
the fact that the current state of the atmosphere is not well known. There are only so many observations taken, and at great distances between them.
Currently there are 14 different models that provide forecasts to 7.5 days (180hours). For more information, please see the AMS paper on the
subject: "A Client Application for Real Time NOMADS at NCEP to Disseminate NOAA's Information Data Base" by Alpert and Wang.
The complex nature of the atmosphere, coupled with inadequate observations, and current computer models, results in forecasts that always
contain uncertainties. NWS forecast products today are for the most part deterministic providing only one prediction of the future state of a
system. Probabilistic forecasts convey uncertainty in the prediction (e.g., a Hurricane track .cone. of probabilities), but are usually not
included in NWS forecasts and products today. Probabilistic forecasts, generated by ensemble models, convey uncertainty in the prediction.
The volume and complexities of ensemble datasets present an unparalleled access and use problem for even the most sophisticated end user. However,
the NOMADS model archive and access system employing a distributed Service Orientated Architecture is well suited to provide easy to use temporal,
variable, and geographic subsets of these data over the Internet.
How to use the Prediction Probability tool
Start the Prediction Probability tool:
HERE
- Select the geographic location you want to retrieve forecast information. This list is alphabetical and global. Typing in the first
letter of the City in the pop-down list will scroll you to your selected City. Once selected, the latitude and longitude of that City is
displayed on the page.
- Select the model run, or start time you want to use. The default is the most recent run of the models.
- Select the atmospheric variable or variables you wish to interrogate. To see if a City will experience a freezing event, select
Temperature "less than" 0°C (or 32°F). Another example would be selecting "Precipitation"
and select the amount of rainfall you are seeking a probability that will occur. In millimeters of rainfall.
- Combinations of weather variables of temperature, precipitation, and wind force can be selected to see if such combined events will
occur (temperature below 0°C AND winds in excess of 20 knots).
Note: Variables like precipitation are accumulated, in these cases, over 6 hourly intervals. The probability of events with precipitation are over
6 hourly periods so the amount of rain equivalent is over a 6 hour period. We didn't add up to get rainfall per day. The rest of the variables
are snapshots.
The first bar in the graph represents the probability of the event at the time of the end of the first 6 hour forecast. Zero does not have a mark
since precipitation amounts are only generated by the models for forecasts.
The results are displayed as percentages. For example, if 7 of the 14 Ensembles met the parameters of your request, the probability is 50%.
Finally, the results are for grid point locations on a longitude-latitude one-degree mesh. Note: For the GFS Ensembles (that these are derived from) the
resolution is 1 degree (approximately 111km). Soon will will develop this capability for the 1/2 degree (50km) Global models.
Top of Page
|
