Search site


All modelling methods incorporate assumptions, including the UKCP09 Weather Generator (WG). These include:

See the UKCP09 Weather Generator report, in the Reports & guidance section, for an in-depth discussion of these assumptions, which are also summarised below.

Rainfall can be used to estimate other variables

The WG methodology considers that rainfall is the primary variable. It starts by randomly generating realistic rainfall time series. Other weather variables, such as temperature and wind speed, are then determined by whether the day is wet or dry and the weather on the previous day.

It assumes that wet-dry sequences are an important component of the variability and character of daily climate. The success of the WG therefore depends heavily on the method of rainfall generation. 

Observed relationships between daily variables don't change over time

The WG works by 'learning' about the relationships between and within daily variables from the observed climate dataset (1961-1995).

These relationships are captured through mathematical/statistical equations which are location specific and recognise seasonal changes. The relationships between the variables are assumed to remain fixed for future time periods, as there is no evidence on which to base any changes.

Hourly values can be disaggregated from daily values

Within the WG, hourly time series are created from the daily time series using simple disaggregation rules. These rules are based on the observed climate and are assumed not to change in future time periods. 

The rules (e.g. no change in the diurnal cycle, day-time temperatures are generally higher than night-time temperatures) are used to convert the generated daily time series into sensible hourly time series. They resulting values are consistent with the daily figures (e.g. hourly rainfall totals add up to the daily rainfall total).

Time series for a point location are representative of a 5 km grid square (or an area up to 1000 km2).

The WG can be used to generate weather data for a point (more specifically a 5x5km grid square) or averaged over an area up to 1000km2.

The WG generated time series are for a point which is representative of a 5 km grid square. When time series are generated for an area the resulting time series still correspond to a single point, but a point, which is representative, on average, of the larger area. Care must therefore be taken when considering generating time series for larger areas, particularly where there are large differences in elevation.