Resolving inconsistencies in the dry period properties of bias-corrected daily precipitation timeseries: an example from the TRANSLATE climate projections for Ireland.

Data from the TRANSLATE project was used to examine the behaviour of standard climate indices associated with prolonged deficits in precipitation, particularly those that rely on the property of “consecutiveness”. Ireland has a temperature oceanic climate and usually gets a large amount of rain, but it is not uncommon for shortfalls to occur also. Having access to reliable climate information about these phenomena is important for planning of future infrastructure and for water-reliant sectors, such as agriculture and utilities.

The TRANSLATE dataset is a bias-corrected ensemble of regional climate model projections over Ireland, with quantile-mapping used to produce the bias-corrections. The regional climate model data comes from a combination of the EURO-CORDEX simulations and an ensemble of COSMO and WRF downscaled simulations over Ireland, both of which were driven by CMIP5 global model data. Bias-correction techniques, such as quantile-mapping, can successfully adjust daily time series to remove overall biases, but generally do not account for “consecutiveness”, and so can introduce occasional wet-day interruptions into otherwise dry periods in the detrended and bias-corrected daily precipitation time series. This has the potential to produce inconsistent results between the raw and bias-corrected projections if the bias-corrected daily time series is used to calculate indices which rely on this property.

Results for the standard climate index Consecutive Dry Days (CDD), the related “Dry Periods” (count of periods of more than 5 consecutive dry days) and the Standardised Precipitation Index (SPI) are presented to highlight where these inconsistencies may arise, and the steps taken to address them. It was found that calculating the indices first, using the raw projections, and then applying the bias-correction to this index data resolved this apparent contradiction. Meanwhile results based on single-day extremes and occurrence frequencies were unaffected by the choice of technique. This research helps to provide the best representation of future projected dry periods in Ireland, while adequately highlighting the uncertainty in these projections.