Why Consider Droughts Beyond the Historical Record?

The draft Environment Agency planning guidelines for PR19 outline that water companies “need to know how much water is reliably available to supply your customers in each of your water resource zones through a range of challenging droughts.”

Traditionally for water supply planning a company uses the historic record to identify the worst drought events that have affected their area and then use these drought events to test the performance of their contemporary water supply system. In England and Wales the historical record typically goes back to the early twentieth century, although with greater uncertainty as data records are less widespread further back through time. This will typically provide up to five or six historical events, with differing degrees of magnitude and impacts, in order to design and plan a water supply system.

Drought events are rare and their characteristics are both spatially and temporally variable between different drought events. Planning to the events in the historical record alone does not reflect the fact that droughts could occur in the future with different characteristics to those that we have already experienced. The concern therefore is that a water supply system may appear robust to the historical events to which it is designed but it may not perform to the same standards when exposed to a previously unseen drought.

How to consider droughts with different characteristics

A number of methods and evidence sources exist for creating droughts with different characteristics from those in the historical record. Understanding the differences between methods can help to identify which is most suitable for testing a specific water supply system. Drought is a complex phenomenon whose occurrence and predictability, even today, is not fully understood by meteorologists, climatologists and hydrologists. There are a number of sources of evidence which can be used to generate drought events that are different from those observed historically, however, they are all typically constrained by our latest scientific understanding.

Stochastic weather generators

Stochastic weather generators are used to generate long sequences of artificial weather and are typically trained to do this based on historical data. The weather sequences are generated for many hundreds of years in which it is possible to identify drought periods for testing a water supply system. A weather generator can be used to produce drought events with a range of different characteristics to those from the historic record. Due to the long nature of the sequences developed, return periods can be calculated to be associated with each event. However, it should be noted that weather generators are calibrated using historical data and it is therefore difficult to validate the drought events that are more severe than the historical record. Stochastic weather sequences were used by Southern Water for WRMP2014 and HR Wallingford is currently applying stochastic weather sequences for two other water companies for WRMP2019.

Synthetic drought scenarios

Synthetic drought scenarios allow a water supply planner to decide what type of drought event they would like to test their system against. The methods HR Wallingford helped develop as part of the Environment Agency and UKWIR project “Understanding the performance of water supply systems during mild to extreme droughts” used drought duration and long term average rainfall deficit to define the design drought event. The drought events are created by resampling the historical record in order to produce the desired drought characteristics. The droughts events can be viewed as a stress test placed on a water supply system and can also be used to modify the characteristics of historical drought events. HR Wallingford is currently applying these techniques for a number of water companies to assess their system vulnerability to drought and to test the ability of future infrastructure development options to reduce the risk and impact of future droughts.

Climate models

Climate models are typically used to understand the impacts of climate change on mean climate throughout the twenty-first century. The latest climate models also provide time series of climate for the twentieth century with multiple versions provided to reflect different sequences of natural climate variability. The climate model outputs can therefore be analysed to identify drought sequences. However, climate models are generally considered to have limited skill in simulating some climate variables relevant to drought formation and persistence and so at the current time should only be used to identify drought signals which can then be combined with resampling of the historical record to provide an inventory of climate change impacted drought scenarios for testing a water supply system. A similar approach is currently being applied by HR Wallingford as part of the Water UK funded Water Resources Long Term Planning Framework. 

Attached link

Taxonomy

• Drought
• Climate Change
• Water Supply