What to be aware of
There are a number of important issues to be aware of when using the UKCP09 Weather Generator (WG), including:
- It is not a physically-based climate model
- Quality of output depends on quality of input data
- Translating climate projections into time series introduces an additional source of uncertainty
- Potentially large data file sizes
- Validation is crucial
- Outputs are not spatially consistent
- Estimation of some types of extreme events not necessarily possible
- Weather Generator output is not a weather forecast!
- Poor representation of large scale climate processes
See the UKCP09 Weather Generator report, in the Reports & guidance section, for an in-depth discussion of these, which are also summarised below.
The WG is a statistical model, based on empirical relationships between and within daily variables from the observed climate dataset (1961-1995). This is quite distinct from Global Climate Models (GCMs), which provide a mathematical representation of the processes that are understood to govern the climate system.
The WG output is reliant on the quality of the observed climate information (used to allow the Weather Generator to learn about baseline climate) and the UKCP09 projections (used to perturb the baseline climate to produce plausible future time series).
It is important to validate the WG output relative to the observed data before using the generated outputs from future time periods. See the UKCP09 Weather Generator report, in the Reports & guidance section, for more information on validation.
The WG adds an element of stochastic uncertainty into a time series because it is based on random number generation. For the baseline, this uncertainty is easy to quantify, because it can be calculated from the set of 100 WG samples. However, for future projections, a similar exercise will likely underestimate the full uncertainty range. The 100 30-year WG sequences for the future are each generated with different change factors, so encompass some of the additional future uncertainty, but this does not sample the full space of uncertainty.
The WG, by default, produces 100 simulations of 30 years of daily time series for the baseline period and a minimum of 100 time series for a future 30-year time period. The files produced could potentially be up to 1 GB in size. This would typically take about 6 hours to download with a 1 Mb/sec broadband connection.
Users should also consider that to analyse the full range of time series appropriately (e.g. validation and uncertainty analyses) may require significant IT resources.
All WG outputs should be validated. By comparing output for the baseline period with observed statistics users can begin to understand how reliable future projections may be. See the UKCP09 Weather Generator report, in the Reports & guidance section, for more information on validation.
The weather generator time series are produced for a single location representing a 5 km grid square or larger area up to 1000 km2.
Time series for neighbouring locations, however, will not be consistent because they are created independently. For example, there is no correlation between rain fall in two neighbouring locations (it may be raining on one day in one place but not the other).
WG output should therefore not be used where climate information is required simultaneously at multiple locations. In such cases, the 11-member RCM output might be more appropriate.
Since the WG allows users to produce long time series (e.g. 1,000 years of daily output), it is possible, in principle, to investigate very extreme events (such as the incidence of long heatwaves or a 1 in 100 year rainfall event). In practice, the WG has been trained using a much shorter record of climate observations (i.e. 35 years; 1961-1995) which does not necessarily include such long return period events.
It is therefore unreasonable to assume that these extreme events can be accurately simulated in future time series, regardless of how many years of output are simulated.
Using the example of cars on a motorway, if no yellow cars pass a particular point in a five minute period, it is difficult to predict how many yellow cars would pass the same point in a 30 minute period in subsequent days.
Weather generator output is not a weather forecast
Despite what its name might suggests, the purpose and design of a WG is not to provide a weather forecast for the future.
A weather forecast gives an indication of what the weather is predicted to be on a particular day.
A WG provides multiple plausible daily time series all of which are statistically consistent with both the baseline climate (1961-1995) and with the UKCP09 probabilistic projections of future climate change.
The series generated may be thought of as sequences that closely mimic the characteristics of real daily and hourly climate that could happen.
Regional scale climate fluctuations related to for example, the North Atlantic Oscillation (NAO), or El Niño Southern Oscillation (ENSO) events, are not explicitly represented in the WG.
Consequently, there is little or no incorporation of seasonal or annual climate variability in the generated time series from these sources.
For example, changes in sea-surface temperature which can influence warm or cold winters or multi-season dry spells like the 2003 and 2006 droughts in southern England.
The WG is only capable of reflecting seasonal and annual variability consistent with whatever is in the baseline observed climate data set.