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