Rebalancing the scales: using data-driven approaches for advancing water cycle representation in a globally resolved energy balance climate model (202

This project will develop a state-of-the-art global energy balance climate model with an improved and enhanced representation of the water cycle.

Particular focus will be on changes to global precipitation patterns and polar regions for past, present and future warming scenarios.

You will combine current energy balance modelling approaches, earth observation datasets, and apply statistical and machine learning techniques to advance the current state of hydrological representation within the new model.

The resulting changes to the model will be assessed through a number of experiments based on either current climate conditions and future warming scenarios.

Provides the opportunity to join the international activities within the Global Energy and Water Exchanges (GEWEX) community and the European Space Agency Water Vapour Climate Change Initiative.

Furthermore, you will collaborate with scientists from the UK Met Office and National Centre for Earth Observation with the opportunity to work on site through placements.

Energy balance climate models are a simplified representation of the Earth's climate system and are less detailed and comprehensive than complex general circulation models (GCMs).

Rather than try to attempt to resolve the dynamics of the climate system (e.g. large-scale wind and atmospheric circulation systems, ocean currents, atmosphere and ocean convective motion), they instead focus on the thermodynamics and energetics of the climate system.

A key advantage of this simplicity is that their underlying assumptions and equations have greater transparency, making it easier for researchers to trace and understand the factors influencing climate change. Energy balance models can run quickly, allowing for rapid experimentation and analysis.

This speed makes them helpful in exploring a wide range of climate-related questions without the computational burden of more complex models.Energy balance models are often used to estimate the Earth's climate sensitivity, a crucial parameter in understanding how the climate system responds to changes in greenhouse gas concentrations.

They provide a simple framework for studying the relationship between radiative forcing and temperature change.

For example, they can assess the temperature increase associated with different levels of CO2 emissions or perform sensitivity analyses, helping to identify the most influential factors affecting climate outcomes and guiding further research.With globally resolved energy balance models, sensitivity to changes in the hydrological cycle significantly impacts results as the model is now sensitive to the spatial distribution of water in all its phases.

For instance, sea ice extent or changes to regional rainfall will affect the model response to rising atmospheric CO2 levels. Figure 1 presents an example of the model run for 1850-2100 using a variety of possible atmospheric CO2 levels.

The top row of plots shows that while the model agrees well with current observations at the global scale, this breaks down when we focus on the polar regions, reducing confidence in projections beyond 2030. However, the results are spatially resolved to identify areas where biases may occur.

This PhD seeks to improve and enhance the representation of the water cycle in the model, making it a valuable tool for climate studies.

Initially, the student will take advantage of freely available globally resolved energy balance climate models to investigate similarities and differences in the inclusion and parameterisation of the water cycle and the resulting model sensitivities/biases.

The project's next phase will see the student working with state-of-the-art climate quality observations from satellites and ground-based measurements to develop data-driven methods to enhance and improve the water cycle representation in the Globally Resolved Energy Balance (GREB) model. These include data assimilation/Bayesian and machine learning approaches.