45 The changing UK rainfall climate

I watched an interesting webinar (available here on YouTube) on the recent update to the UK Flood Estimation Handbook (FEH) rainfall model.?The model defines the depth-duration-frequency relationship for design storms for all of the UK.?There are further details in the report published by the UK Centre for Ecology and Hydrology (CEH) available online, from which I have borrowed the maps used in this blog.

Some history

• 1935.?The first standard method of representing depth-duration-frequency relationships for rainfall in the UK was the Bilham formula.?This used a simple equation to represent the average rainfall patterns across the whole country with no regional variation.?
• 1960.?The Bilham formula was updated with a better representation on high intensity rainfall.?It still did not include regional variation but identified that this was important.
• 1975.?The Flood Studies Report (FSR) was published with a comprehensive model of rainfall variation across the whole country and covering a wide range of event probabilities.?The model used three parameters to represent the variation in rainfall.?The focus of this report was river flooding.
• 1978.?The FSR rainfall model was adapted for urban drainage use as part of the Wallingford Procedure.?The main change from the fluvial model was to use the 50%ile summer storm profile, which had a sharper peak than the 75%ile winter storm used for fluvial studies.
• 1999.?The Flood Estimation Handbook (FEH) replaced the FSR with a more complex model using 6 parameters to represent the rainfall variation.?
• 2013.?The FEH was updated with more data and a new analysis.?The model parameters were no longer published, but the detailed results were provided as data tables on a website.
• 2022.?The FEH was updated again with even more data and taking into account some recent extreme rainfall events.

The updates are getting more frequent and at this rate, the next update could be due in just 4 or 5 years.

Some limitations of the model

The derivation of the FEH rainfall model does have some limitations.

Stationarity

It does not attempt to take into account past climate change to derive the DDF relationships.?For simplicity it assumes that the past climate has been stationery without any change.?New extreme events are therefore possibly diluted by older less extreme events that are still in the data used for the analysis.?However this is outweighed by the improved accuracy given by having very long data series by lumping all of the data together.

Partly because of this, the change in rainfall estimates from the older FEH13 to the new FEH22 is not primarily due to a change in climate but rather to the addition of extra data and improved quality assurance removing erroneous data from the entire record.

Probability and duration

The analysis uses annual maxima (AMAX) data, so it cannot directly give values for 1/1 year annual probability events without an adjustment.

It is based on daily and hourly rainfall data so rainfall depths for duration shorter than 1 hour are estimated using other information and may therefore be less accurate.

These two limitations would be particularly significant for urban drainage studies, but the adjustments made do overcome these problems.

Results of the model

Compared to the 2013 version of the model, the rainfall depths for durations longer than 6 hours have generally increased by a few percent.

However for short durations (1 hour to 6 hours) the depths have generally reduced by a few percent in the new model.?This is believed to be because of the greater accuracy from the significant increase in hourly read gauges over recent years.?This is good news for urban drainage assessments.?However the exception to this is that short duration depths have increased for Cumbria, Norfolk and Aberdeenshire.?(I am glad that I used a generous allowance for climate change for the flood study of Aberdeen that I did a few years ago).

The map below shows the change from the FEH13 model for a 6 hour 1/30 year event, which would be typical for urban drainage studies.

Compared to the even earlier FEH99 model, FEH22 estimates of rainfall depth are generally larger for shorter durations and return periods. See the map below again for the 6 hour 1/30 year event.

If you are still relying on an assessment using the old FEH99 rainfall model, then you should probably update your model soon.?If you have been using FEH13 then you might have a slight extra factor of safety depending on what assumptions you made on future climate change.

Using the new model

Most urban drainage engineers will be using the FEH rainfall model by linking the REFH2 hydrology software to their urban drainage modelling package.?Although REFH2 has not yet been fully updated for the changes to hydrology and runoff, it should accept the new rainfall data exported from the FEH website and use this data correctly.?(I haven’t tested this, so I suggest that you do.)

For urban drainage studies always generate rainfall depth data for a point rather than for a catchment area.?Using the catchment area automatically includes the area reduction factor (ARF) for the entire catchment area for all durations and will therefore underestimate rainfall for short duration storms.?Generating data for a point does not apply any ARF and this should be applied separately.

Remember to apply a multiplier for future climate change.?Guidance on this is not included in the FEH model but is provided by the Environment Agency for fluvial flows and UKWIR for urban drainage assessments.