The area of proposed research is the predictability of Arctic cyclones. The Arctic is a less well studied and observed region compared to other regions of the Earth, although it is crucial to the global climate and also sensitive to climate change with reduced sea ice extent and thickness in the past 30 years (Kwok & Untersteiner, 2011). Sea ice reduction is projected to continue opening opportunities in the Arctic for business in diverse sectors such as tourism, fisheries, shipping and mineral extraction. However, both synoptic scale cyclones and mesoscale “polar lows” (winter) occur in the Arctic and can induce strong wind-driven waves and precipitation, which can strongly impact on activities mentioned above. It is therefore imperative to better understand the current limitations in the prediction of these storms.
Numerical Weather Prediction (NWP) combines inhomogeneous quality controlled observations from diverse observing systems, with a short-range forecast using data assimilation, to produce the initial conditions for forecasts. However, given the sparse observations in the Arctic region and its seasonal variability, the extent of NWP’s capability of predicting Arctic cyclones on daily to weekly timescales is still an open question. Moreover, the dependency of the forecast quality of Arctic cyclones on large scale variability modes remains unresolved, e.g. the intensity of the Arctic Frontal Zone – a belt in the northern high latitude region of strong horizontal temperature gradients during the summer (Barry, 1967; Reed & Kunkel, 1960). Information on the quality of NWP forecasts of cyclones in the Arctic is crucially important to end users (e.g. coastal communities and MetOcean forecasting companies) and is becoming important to the insurance industry.
1. How well are cyclones (synoptic and mesoscale) currently predicted in the Arctic?
2. How does the predictability of Arctic cyclones depend on season and large-scale modes of variability?
3. How does the predictability of Arctic cyclones depend on the observational network?
In this project, the predictability of cyclones, on lead times of 0-15 days, in the Arctic and their sensitivity to the modes of variability will be explored using data from reforecasts (which provide a consistent view of forecast skill over a long period of time), Ensemble Prediction System and deterministic forecasts from ECMWF, and the Year of Polar Prediction forecast dataset that is designed specifically for investigating polar predictability. Forecast uncertainty will be explored using the ensemble forecasts. A cyclone tracking method, TRACK (Hodges, 1994), which was previously used to track extra-tropical and tropical cyclones in NWP data (Froude, Bengtsson, & Hodges, 2007; Hodges & Emerton, 2015), will be used to identify cyclones in the forecasts and analyses. This will then allow errors in the structure and growth of the cyclones to be studied and how they relate to forecast errors of track and intensity, as well as how this is dependent on season and modes of variability in different regions of the Arctic. Observing system experiments will be performed using the ECMWF system (in collaboration with Dr Jonny Day at ECMWF) that modifies the number and types of observations assimilated that provide the forecast initial conditions, e.g. removing radiosondes, satellite data, or YOPP’s additional assimilated observations. The sensitivity and quality of the forecasts in terms of the location, intensity and structure of the cyclones will be explored for these different experiments, thus allowing the information content of these observations in relation to Arctic cyclones to be determined.
To discuss this PhD opportunity informally please contact Dr Kevin Hodges ([email protected]
Applicants should hold, or be predicted, a strong undergraduate degree (2:i UK honours degree or equivalent), or Masters (merit or distinction level), in a physical or mathematical science.)