Nowcasting vs. Numerical Model
Numerical weather prediction models provide most of what we know from websites and TV about the weather one hour into the future. Numerical models are statistic tools with a tendency to predict precipitation in the wrong places, and at the wrong times. These errors are a function of the data that begins a forecast rather than the quality of the forecast models specifically. Since models refer to macro weather i.e. weather front, high altitude winds, massive storms and the like in for a large domain of about 10 square km. they are the best tool available for mid- and long range weather prediction of few hours to few days.
When using models to predict precipitation for the next 1 – 3 hours, the drawback of an over predicting of rainfall show its face. In order to convey a bit more reliable rainfall prediction, sites and application often attach probability for precipitation with each prediction. While these prediction have good (above 60%) and some time very good probability (of above 80%), it is very common to see also 20% – 40% probability for precipitation, which basically means that this percentage of a region will experience rain on that day, or that it will rain at a particular location once in 3-5 days. Overall, for people planning outdoor activities, or working outdoors, these percentage forecasts are useless.
In order to improve accuracy and avoid embarrassing predictions (for example, where a mobile application icon indicates partly cloudy skies at a user location, while in fact it is pouring rain), leading weather forecast providers are incorporating observations from radar; these observations represent current and recent data. Most of these “model corrections” nowcasting efforts are base on 5 – 20 minutes old radar loops representing where it is raining now (or was raining or snowing 20 minutes ago). Due to the large level of complexity in the atmosphere, and the huge number of simultaneous processes in a weather system, and within a single cloud, these radar loops are useful when making a few minutes correction to the numerical weather prediction model.
For high quality accurate weather forecast of less than one hour, weather companies using trained meteorologists and employ semi-automatic / manual methods also known as Manual Nowcasting. Due to the high level of expertise required for such high accuracy short-range forecast, these predictions are very expensive and not available to the public upon demand.
Automatic Nowcasting is the next generation technique for short-range weather forecasting (up to 3 hours). Nooly is a pioneer and technological leader in short-range Nowcasting aiming to bring minute-by-minute accurate weather to a user where ever he/she is, a technology that once was available only to the government and other organizations that could pay the high price.
