Higher resolution models should improve prediction by more realistically simulating small-scale processes.
Higher resolution models should improve prediction by more realistically simulating small-scale processes. They are also desirable by decision makers, who often want climate information tailored for their own locality. This workstream set out to assess how the resolution of models affects its ability to simulate key climate processes and metrics important for adaptation, for example: what are the benefits of increasing resolution; are there diminishing returns in benefit with increasing resolution; and as resolution increases, do technical problems arise?
To answer these questions, CSRP developed a new state-of-the-art regional climate model based on the HadGEM3 model. This was run for Africa at 5 different resolutions: 135km, 90km, 50km, 25km and 12km. The study found that, in general, increasing resolution improves the realism of the simulations over a range of climate statistics including the seasonal mean and variance, as well as daily events (as measured for example by the mean wet day intensity). Detailed case study experiments were conducted for a single season over East Africa to investigate further benefits obtained by very high (4km and 1km) resolution. It was found that these very high resolution integrations gave better representations of a seasonal rainfall maxima in some areas, such as Lake Victoria, suggesting potential benefits for adaptation decision making. However, the study also showed that the benefits are not universal; further analysis is needed to assess where the benefits of high resolution (for a decision maker) justify the added cost of such simulations. It also demonstrated that even very high resolution does not eliminate uncertainty and further work is needed on the interpretation of regional climate projections for decisions.
Last updated: 12 May 2016