EGU General Assembly 2020
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Evaluating the small-scale space time structure of rainfall in the Convection Permitting Model of UKCP18

Yuting Chen, Athanasios Paschalis, and Christian Onof
Yuting Chen et al.
  • Imperial College London, Civil and Environmental Engineering, London, United Kingdom of Great Britain and Northern Ireland (

Sub-daily precipitation at fine temporal resolution (~1 km2) is critical for a wide range of hydrological applications, such as flooding estimation, urban drainage design. In recent years, a step-change was given by Km-scale Convection-permitting models (CPMs), allowing for the first-time climate change projections at hydrologically relevant scales. CPMs have been now introduced in the operational climate change projections of the Met Office in the UK (UCKP18). The high-resolution hourly precipitation at a 2.2 km scales is currently available for the historical period (1980-2000) and future period (2020-2080) for the RCP8.5 scenario. It is perceived to provide a plausible tool for detailed climate impact studies. However, a question remains unanswered: is the local projection of precipitation from UKCP18 credible for hydrological use? 

To answer the question, simulated hourly precipitation from the UKCP18 for the historical period is compared statistically with the observed rainfall data. Observation rainfall was obtained from UK Met Office C-band Weather Radar network and Gridded estimates of daily areal rainfall (CEH-GEAR). These were used to assess the spatial-temporal structure of rainfall, including spatial spectra, distributions of rainfall cell sizes and intensities, and their temporal growth/decay dynamics, and rainfall extremes. The statistical evaluation was performed for all climatologically distinct regions of the UK on a seasonal basis.

The results show that hourly precipitation in UKCP18 has a realistic spatial correlation structure compared to observations. However, the extreme areal mean precipitation is overestimated, particularly at scales finer than 6.6 km. Significant differences between the size and temporal dynamics of observed and modelled rainfall cells were identified, with distinct differences amongst climate regimes, highlighting the limits of applicability of current generation CPMs for hydrological forecasting.

How to cite: Chen, Y., Paschalis, A., and Onof, C.: Evaluating the small-scale space time structure of rainfall in the Convection Permitting Model of UKCP18, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8007,, 2020


Display file

Comments on the display

AC: Author Comment | CC: Community Comment | Report abuse

displays version 1 – uploaded on 04 May 2020
  • CC1: Comment on EGU2020-8007, Angeline Pendergrass, 05 May 2020

    Dear Yuting,

    For many of the evaluation metrics you show, the different observations have differences that are as large as the distance of the model to any observational datasets. Do you have any thoughts on how to deal with such large differences in observational datasets in the course of model evaluation?



    • AC1: Reply to CC1, Yuting Chen, 05 May 2020

      Dear Angie,

      Thanks for your comments.

      Yes, I agree that the difference in observation datasets can not be ignored. Besides more quality control, radar-rain gauge merging will be a good solution. Spatial structure can be different due to selected merging techniques. So, in the next step, a suitable merging method which can preserve 'true' spatial structure or a well-checked merged dataset will be beneficial to my study.

      All the best,


      • CC2: Reply to AC1, Angeline Pendergrass, 05 May 2020

        Thanks, Yuting!