Embedded Files
Model output comes in many forms. Here, we have easy access to two forms of data: gridded fields and point forecasts. We've been using forecast maps from the NAM all semester. Other modules have described how to read the forecast maps; this section will describe how to read the point forecasts.
There are three types of point forecasts: direct model output (FOUS, rhymes with noose), interpolated model output (EXT), and model output statistics (MOS, rhymes with loss). Direct model output is computed and transmitted by NCEP. It includes a set of forecasted fields, such as wind, temperature, and precipitation, interpolated to various forecast cities. Interpolated model output is similar, except we do the interpolation here using the gridded forecast data we receive from NCEP. Finally, model output statistics are computed by NCEP based on what has happened at particular stations in the past given particular model forecasts. In effect, model output statistics remove biases and systematic error, and attempt to correct for errors in particular situations.
FOUS and EXT are similar. When possible, use FOUS: it was generated from the original model output by NCEP. EXT is less accurate because it is generated from the coarser grids that NCEP actually transmits to us. But we can interpolate data to any station, not just the FOUS stations, so EXT is available sometimes when FOUS isn't. Also, EXT includes more types of data and is partially decoded.
When making a forecast for the WxChallenge, a lot of this information is useful. Of greatest value are the MOS temperature forecasts and the FOUS precipitation amounts. These numbers are used to construct "guidance", which is scored along with your forecasts.
You can always make a forecast simply by writing down the FOUS and MOS numbers for the station. However, this is of no practical value whatsoever. The public has no need to hire you to make forecasts when it can just access the computer forecasts directly. The true value of a forecast is measured by how much it improves on objective guidance.
Report abuse