How can we make use of all the UK’s weather radar data to improve flood forecasting? Multifrequency Dual-Polarisation Doppler Radar Rainfall Retrievals for the UK

Flooding is one of the highest impact natural hazards and is forecasted to become increasingly frequent and extreme due to climate change. Accurate, timely, and high-resolution precipitation observations underpin the Met Office’s and Environment Agency’s joint approach to forecasting, warning, and responding to flooding. Continuous improvement of precipitation detection and the forecasting data that follows enables the UK to effectively safeguard people and property from these high impact weather events. Weather radars are a critical component of our precipitation observations, offering wide areal coverage at high spatial and temporal resolution.

The overall goal of this PhD is to improve the compositing of available UK weather radars to increase the accuracy of quantitative estimates of precipitation (QPE). The most significant barrier to this goal is how you combine the observations from radars of differing frequencies (wavelengths) where they overlap and create an optimum estimate of rainfall. Differences in radar technology, the retrieval method and other observational parameters of the individual radars all enhance the underlying difficulty of compositing data from overlapping radar systems. Therefore, the underlying physics needs to be (re)considered and combined with the optimal strategy that makes the best use of the strengths of each type of retrieval from each kind of radar to create the most accurate estimate of precipitation possible.

Six different ways to calculate rainfall using the same set of radar observations.

This project will use dual polarimetric radar observations to improve our ability to make accurate QPEs in near real-time. To do so, the PhD student will work closely with the Met Office Radar R&D team and the NCAS Weather Radar Group at the University of Leeds, developing novel precipitation estimation algorithms. This will initially use the archive of Met Office radar data combined with the observations from the NCAS X-band but could expand to include other novel data sources such as the Newcastle Urban Observatory weather radar.

The student will evaluate the skill of the algorithms by validating it with other observations, which include observations from the Met Office, SEPA, NRW and the EA rain gauge networks. Through NCAS, the student will also work with CEH to assess the new precipitation estimates using the Grid to Grid (G2G) hydrological model.

Dual-frequency observations also unlock more information about the microphysical composition of the atmosphere. This has the potential to enhance the QPEs generated from a combined approach and to increase our understanding of the underlying atmospheric processes, which in turn could improve QPEs and hazardous weather prediction.

In the first phase of the project, the student will be trained in the application of dual-polarisation QPE algorithms and differing approaches taken throughout the world. They will then apply this knowledge to an assessment of the available radar data sources in the UK and their potential. This will result in a publication that describes the current state of the UK precipitation observations and how this compares to other observing systems.

In the project’s second phase, the student will apply this knowledge to combine QPEs from available UK sources, validating the output against the UK rainfall observing network. The method(s) employed will need to explore the full potential of the data sources and will most likely involve machine learning to create an optimal solution. There is probably no single solution to this problem, so an essential step in this phase of the project will be to establish a set of rigorous metrics (in partnership with the CASE advisor) to compare the results to understand the improvements gained. The exact methods used will make up a significant part of the research involved in the work and will result in a high impact paper that will inform the Met Office and other national meteorological organisations on how to improve their QPEs.

The project’s third phase will focus on unlocking the potential of multi-frequency observations for scientific understanding of the conditions that lead to hazardous flooding. This will use the results from the second phase to identify cases of flooding that can be further examined with the new data sources and methods.

The research done by this PhD has a direct path to impacting the entire UK. Specifically, the Met Office could use this research to help inform the creation of the next generation national precipitation product.

Through this process and explicit training provided through the PhD,  the student will gain skills and hands-on experience in the use of high-performance computing (JASMIN) and how to analyse “big data” (the volume of data to be examined here is on the order of ~100TB and spans all the observations collected by the 16 Met Office and NCAS radars over the last decade) using cutting edge techniques that include machine learning tools.