Probable Temperature Deviations from Average 1961-90
from June to November 2007 at Berlin in Kelvin:
Jun Jul Aug Sep Okt Nov
0.8 -0.2 1.5 0.5 0.3 -0.6
standard deviations(K) for the analogue years approximately:
1.5 1.7 1.5 1.5 1.5 1.5
For further information about the forecasts for other regions
send a mail to firstname.lastname@example.org
May circulation models produce better longrange forecast results than
The simplest statistical or "natural" forecast method for the following month is to use the persistancy of the atmospheric circulation features. The best results for the temperature are obtained by the continuity of the last days´ values of a month, depending from the season. In order to avoid greater errors it is sometimes better to involve the average temperature of the whole month. The real values for an element as the temperature you will get by the persistancy correlation.
The longrange forecasts of the circulation models are based on the physical properties of the atmosphere, oceans, continents and so on. The results are comparable with the statistical methods, as the reliability of the outputs ends near 15 days. From this moment the forecasts are a mix of the normal seasonal change and atmospheric wave behavior, persistancy or change of ENSO, model properties and other features.
If there will exist well known influences, which determine the longrange weather, they may be involved in the statistical methods. The circulation model outputs over several months therefore
cannot be more successful
than those which you can get by conventional means. The statistical forecast has some great advantages: It can easily be adapted to each observation station, has no problems with the temperature level and it is much cheaper than model computations.
Para o próximo outono achoque podemos ter duas certezas, será mais frio e menos chuvoso que o outono do ano passado.