Immersive storytelling and gaming
In this short article I describe how games can be used to help communicate research.
I describe how research I conducted through the Earth Arcade project that showed that the Flash Flood! VR activity helped people take an interest in flooding topics.
We explore how geoscientists, trained to interpret landscapes, are able to distinguish between real environments and those of the videogame, Zelda: Breath of the Wild.
Environmental modelling
Rainfall is difficult to measure and there are different ways to do it (each called a product). We show that the choice of rainfall product makes a big difference to amount of landscape change is simulated in our models.
Climate change will not only change the amount of rainfall to different area, it will change the way it is delivered. Here we show that it is important to simulate the changes in rainstorm structure for understanding future river change and flood risk.
An end-to-end forecasting system is a single framework containing multiple models simulating everything from weather, to river flows, to flood extents. We describe the learning from the NERC-FFIR programme and make recommendations for a successful end-to-end forecasting system.
We performed the first global sensitivity test on a landscape evolution model, highlighting the sediment transport law as the largest source of parameter uncertainty. We also describe a suite of new behavioural functions to assess model behaviour changes when using geomorphic models.
We highlight the importance of using high resolution data (sub-daily) when simulating water levels in estuaries. It was more important in smaller estuaries than larger ones.
We show that the way you apply rainfall data to a landscape evolution model can create large differences in the amount of erosion and sediment transport during simulations. We recommend that the highest resolution, both spatial and temporal, is used.
We compare simulated flood extents, predicted from a reduced complexity 2D hydraulic model and a simple bathtub model, to observed extents for storm surge events. The reduced complexity model produced the best results.
We developed a reduced complexity 2D hydraulic model for the Humber Estuary, UK, and showed that it performed well at simulating water levels during normal tidal conditions compared to observed water levels. It also performed well during storm surge conditions and showed promising results for predicting flood inundation extents.
Hydrological models contain values that can be tuned (parameters) to make sure simulated values match those observed in reality (calibration). My research assessed different ways to calibrate the model parameters when using ensemble rainfall inputs, finding that models calibrated against a whole ensemble performed best.
Thesis
Skinner, C. J., 2013. Ensemble-characterisation of satellite rainfall uncertainty and its impacts on the hydrological modelling of a sparsely gauged basin in Western Africa. P.hD Thesis, University of Hull, UK (Externally examined – Prof Paul Bates; Internally examined – Prof Dan Parsons)
In UK we can measure rainfall in high detail using dense networks of weather radar and raingauges. In sub-Saharan Africa these dense networks do not exist so satellite data is often used to fill gaps. However, satellite methods are not as accurate and the uncertainties can be large. My PhD research looked at how these uncertainties transfer to the hydrological models using satellite rainfall data.