Why Don't We Have a Specific Forecast for Inauguration Day?

You may have noticed that the discussion of the Inauguration Day weather outlook, at least so far, has been expressed only in terms of average probabilities for an entire week including the day of interest. If climate predictions can be made with some confidence well into the remainder of the century, why can't we make an accurate weather forecast less than 2 weeks ahead?

Early last year, we posted an analysis of the limits of daily weather forecasting. The upper chart to the right shows that there has been a steady increase in the skill of computer weather models, but there is essentially no value in a forecast beyond about 8 days. In spite of this, some current forecast models are routinely run out as far as 16 days. Is this just a waste of computer resources at taxpayer expense? No, it serves to generate a base of experience from which the models can continue to be improved in the future.

Because the model output is available, however, snow freaks looking for the next school cancellation, commercial weather services looking to make a quick buck, and corporate media attempting to capture more eyeballs for their pop-up, animated, cycle-sucking advertising continue to peddle the notion that a forecast for a specific day at a particular location 2 weeks or more in advance makes sense. This is not only foolish, it's also unprofessional on the part of those meteorologists who enable such activity (despite their disclaimers and caveats), and it confuses a public already being bombarded by misinformation and even deliberate disinformation on the relationships among weather, climate, and policy issues.

As an example, the 4 maps to the right are the computer model predictions for 7 am, January 20, 2009 made on each of the last 4 days (most recent at the top, oldest at the bottom). If you're not familiar with reading weather maps, the solid lines are surface air pressure, the green and blue shaded areas are precipitation, and the dashed red and blue lines are an indirect representation of temperature. You don't need to be a weatherman, however, to see that these are very different patterns.

Images (click to enlarge): Weather model skill as a function of year, from The Emergence of Numerical Weather Prediction, Peter Lynch, Cambridge University Press, 2006. Chart © ECMWF. GFS model output maps for sea level pressure, 1000-500 mb thickness, and precipitation, 7 am, January 20, 2009, from NCEP/NWS/NOAA.