Chapter 13

Methods of Forecasting by People

·         Folklore Forecasts are short saying that attempt to predict the weather based on sky conditions, special days of the calendar, or the behavior of animals.

·         Persistence and Climatology

o       Persistence Forecast-the weather you are having now will be the weather you are having later, tomorrow’s weather will be the same as today.

o       Climatology Forecasts- relies on the observation that weather for a particular day at a location doesn’t change much from one year to the next.

·         Trend and Analog

o       Trend Forecasts- acknowledges that weather does change, but it assumes that the weather-causing patterns are themselves unchanging in speed, size, intensity, and direction of movement.  Accuracy of these forecasts declines if made for longer than a few hours ahead.

§         Nowcasting is forecasting for a brief period of time, such as several hours.

o       Analog Forecasts acknowledge that weather changes, but, unlike the trend method, it assumes that weather patterns can evolve with time.  It is an empirical method of forecasting that uses past weather events that resemble the current conditions to create forecasts for the days ahead.

o       Requires many years of weather maps and an efficient way to compare one map to another.  One approach is to categorize according to Weather Types.

·         The methods are hit and miss and are inadequate for modern needs.

Numerical Weather Forecasting

·         Numerical Modeling is the technique of approximating tough real-world problems with numbers.

·         A Model is a simplified, but relatively accurate approximation of reality.  Today’s weather forecasts are computed by numerical models that approximate the behavior of the actual atmosphere.  The numerical formulas used are called a Model.

·         Numerical Weather Prediction Process

o       Step One: Weather Observations- a numerical forecasts is only as accurate as the observations that go into the forecast at the beginning of its run, the “initial conditions”.  Surface observations, radiosondes, and satellite measurements supply most of the data used for model initial conditions.

o       Step Two: Data Assimilation- The combining of observed data into a numerical forecast model.

§         Gridpoint Models are a class of numerical weather forecast models that divides the atmosphere into grids, which are a set of orderly arranged points on which variables are analyzed or predicted in a numerical weather forecasts.

§         The middle of the model is called the Gridpoint where the model actually calculates weather variables and makes forecasts.

§         Interpolation is the process of creating an evenly spaced data set from irregularly spaced observations.

§         Data Initialization is the smoothing out of errors and inconsistencies in the initial data that is to be ingested by a numerical weather forecast mode, which leads to more accurate weather forecasts.

§         Data Assimilation includes the multiple jobs of interpolating and balancing the data for use in numerical models.

o       Step Three: Forecast Model Integration

o       This step puts together the two key ingredients of a forecast: the observed data and the model’s formulas.

o       A 24 hour global forecast for just five weather variables can take more than 1 trillion calculations.

o       During this step, the physical distance between interpolated distance points becomes very important, with smaller spacing making it easier for the model to resolve, so it has good resolution.

o       Resolution is the spacing between gridpoints in a numerical weather forecast model.  Fine resolution results from close spacing of gridpoints, while coarse results from wide spacing.

o       A downside to good, or “fine” resolution is that it can take a very long time to compute because more gridpoints are used.

o       Step Four: Forecast Tweaking and Broadcasting

Modern Numerical Weather Prediction Models

·         Short Range Forecast Models

o       LFM (Limited Area Fine Mesh Model)- the first truly modern numerical forecast model

o       NGM (Nested Grid Model)- an improvement to the LFM

o       NAM (North American Mesoscale, formerly called Eta)-run four times a day

o       RUC (Rapid Update Cycle)-runs every 3 hours

o       WRF (Weather Research and Forecasting Models)

·         Medium-Range Forecast Models

o       Spectral Models are a class of numerical weather forecast models that divide the atmosphere in terms of waves rather than gridpoints.  Can run on a computer faster than gridpoint models.

o       The ECMWF (European Centre for Medium-Range Weather Forecasts) is probably the best model; it is a type of spectral model.

o       Ensemble Forecasting is a method of weather forecasting that uses the results of chaos theory to assess to the amount of confidence that should be placed in a forecast.  A forecast is run repeatedly with slightly different initial conditions.  If the resulting forecasts, or the “ensemble”, agrees closely, then the confidence in the forecast is high.  If the ensemble exhibits a wide range of different forecasts, then the confidence in the forecast is low.

Why Forecasts Go Wrong Today?

·         Imperfect Data

·         Faulty Vision and Fudges (Imperfect knowledge of how the atmosphere works)

o       Parameterizations are portions of numerical weather prediction models that are devoted to the approximation of phenomena that the model cannot calculate precisely.

·         Chaos: Because we don’t know the atmospheric conditions perfectly at any time, chaos means that the resemblance between a model’s forecast and reality will be less and less with each passing day.

·         Limits on computer power