How can I use the Weather Generator to provide counts of days with rain?

Users requiring estimates of the number of days with rain under projected climate change will be able to develop such counts using the multiple time series available from the Weather Generator available within UKCP09.

In detail

These counts may also be of interest to those users who require estimates of the length of dry spells and perhaps wet spells. Prior to using this count information, users should clearly understand how a day with rain has been defined. Associated with this definition are two important issues:

1. the smallest possible observed value (observational threshold); and
2. the smallest amount of rainfall that has an effect (effectiveness threshold).

With respect to the first issue, although rainfall amounts can be measured down to 0.1 mm, many observers often do not report such small amounts. A measure of the rigour of observers’ practices can be obtained by looking at the rainfall amounts within observational records (counts of specified amounts of rainfall) over, say a 30-year period. These counts and accompanying cumulative plots show that many observers often round the precipitation recorded to 0.2 mm and some to 0.5 mm. Furthermore, prior to metrication, the smallest measurable amount was 0.3 mm (equivalent 0.01 in). From an observational limitation perspective therefore, the lowest threshold that would be consistently measured by different observers is 0.5 mm.

In terms of the effectiveness of rainfall, most meteorological services use 1 mm as a threshold to determine a day with rain. Additionally, most users would recognise that the timing of the rainfall relative to antecedent conditions would also play a role in defining the effectiveness of rainfall. For most, it would be clear that 0.5 mm of rainfall in the middle of two 10-day dry spells will make little if any difference and the spell in reality should be recognised as one dry spell with a length of 21 days.  

A further consideration is that climate model output is averaged over a grid square of area 25 x 25 km in the case of UKCP09. Since rainfall can be quite spatially variable, agreement between rain-day counts estimated from a rain gauge and averaged over a grid square can be poor if a low threshold amount is used, and better correspondence is found using higher thresholds.

Find out more

• Weather Generator report