| Do you think better weather forecasting
| |
| | Check the forecast highs and lows for the
|
| requires a degree in meteorology? Maybe a
| |
| | last 365 days, and check the actual
|
| degree in statistical analysis would be
| |
| | temperatures for those days. See what the
|
| more helpful. Consider this following new
| |
| | predicted probabilities of rain or snow
|
| way to forecast the weather with more
| |
| | were, and what actually happened.
|
| accuracy and less knowledge.
| |
| | Suppose that of the last 24 times a
|
| Friday, February 2, 2007 - Canon City,
| |
| | forecaster predicted a 50% chance of
|
| Colorado. I brought in my Canon City
| |
| | rain, it actually rained 18 times. He may
|
| Daily Record from the porch when it
| |
| | have the best data, but he may be too
|
| arrived, at about 3 in the afternoon. I
| |
| | conservative in how he uses it. Suppose
|
| opened the newspaper to the page with the
| |
| | this was not a fluke - which can be
|
| weather forecast, wondering how cold it
| |
| | determined by doing more statistical
|
| would be the following day.
| |
| | analysis. You could know nothing about
|
| The paper forecast a high temperature was
| |
| | weather forecasting and provide a more
|
| 13 degrees Fahrenheit. I knew this was
| |
| | accurate forecast simply by saying "A 75%
|
| way too low. Forecasts on television and
| |
| | chance of rain tomorrow" every time he
|
| on the internet said that we would reach
| |
| | said there was a 50% chance, right?
|
| 23 or 27 degrees the following day. I
| |
| | That's the basis for this new forecasting
|
| knew they were also too low, and I told
| |
| | model. First gather the statistical
|
| my wife it would be in the 30s at least.
| |
| | information on the forecasts of several
|
| The actual high temperature the next day
| |
| | weather forecasting services or
|
| was 53 degrees Fahrenheit.
| |
| | meteorologists. Compare this to the
|
| No, that's not a typo. The various
| |
| | actual weather that happened, and look
|
| weather forecasting "experts" were off by
| |
| | for any consistencies in the
|
| as much as 40 degrees - and that was for
| |
| | inaccuracies. Then you create a computer
|
| a simple 24-hour forecast. How could they
| |
| | program. As you enter each of these
|
| be so far off? And how could I be better
| |
| | forecasts into it, they are adjusted for
|
| than them at forecasting the weather?
| |
| | known tendencies. The result is a more
|
| I can't answer the first question. The
| |
| | accurate forecast.
|
| weather here is more unpredictable than
| |
| | If Forecaster A has managed over the last
|
| in most places. Also, perhaps the
| |
| | year to forecast a high that averages 4
|
| meteorologists follow there computer
| |
| | degrees over the actual high, the
|
| models too slavishly, even when their
| |
| | computer adjusts for that. More
|
| experience and intuition tell them to
| |
| | sophisticated analysis might show that
|
| adjust the forecast.
| |
| | Forecaster B is consistently predicting a
|
| However, I can answer the second
| |
| | higher probability of rain than there is
|
| question. My guess was closer because the
| |
| | in the fall, but a lower probability of
|
| forecasters were so consistent in the way
| |
| | rain than there actually is in the
|
| they made their errors. Around this time,
| |
| | spring. The computer can take this into
|
| I remember counting something like 15 out
| |
| | account. Finally, it may work best if the
|
| of 20 days when all the various weather
| |
| | adjusted forecasts of three or more
|
| forecasts predicted a high temperature
| |
| | sources are then averaged.
|
| that was 5 degrees or more too low. All I
| |
| | There really is no need to know anything
|
| needed to do was take the highest
| |
| | at all about weather forecasting. This is
|
| temperature forecast and add five
| |
| | based on the idea that even when experts
|
| degrees.
| |
| | have all the best knowledge and data,
|
| A New Weather Forecasting Model
| |
| | they sometimes apply it incorrectly, and
|
| Consistency in their errors was the key
| |
| | do so in consistent ways. Don't be
|
| to my success. They weren't forecasting
| |
| | surprised if some television stations get
|
| too high one day and too low the next.
| |
| | rid of their meteorologists and take
|
| They were wrong in the same ways over and
| |
| | advantage of this new weather forecasting
|
| over.
| |
| | idea.
|
| Are the errors as consistent in other
| |
| | "Now your electronic weather forecast,
|
| parts of the country? That could be
| |
| | from our Statistical Analysis Weather
|
| determined by looking at the statistics.
| |
| | Machine.
|
