Running on the largest dedicated flood modelling computer grid in the world, it’s safe to say we’re pretty proud of JFlow.
The result of more than nine years of development by JBA Risk Management’s in-house hydraulic specialists, JFlow was and is the first hydrodynamic flood modelling software to repurpose gaming technology. It routes water from the hydrological model across a digital terrain model (DTM), solving the full shallow water equations.
JFlow was so innovative that it reached the finals of the UK’s most prestigious award in engineering (The Royal Academy of Engineering’s MacRobert award) and the technology behind it has been widely published in peer-reviewed literature.
There are currently around 300,000 stream processors in the grid.
Our response to the 2007 floods
Following the devastating floods that hit the UK in the summer of 2007, Sir Michael Pitt's Review identified the need for surface water maps. Having just completed our national-scale surface water maps, we were able to supply them to the Environment Agency almost immediately.
The Comprehensive Flood Map
Our Comprehensive Flood Map of Great Britain was the first flood map to include six flood types at high resolution: river, coastal, surface water, groundwater, canal failure and reservoir dam break. It is the most widely used flood map amongst UK insurers.
JBA Risk Management’s Stochastic Hazard Event Estimation Programme, JSheep, is statistical software for analysing the spatial and temporal patterns in the occurrences of extreme catastrophe events. The main output from the software is a set of highly customisable simulated hazard events which can be either plugged into a standard single or multi-peril probabilistic catastrophe model, or used independently to assess the probability of rare or unprecedented events, e.g. the joint probability of multiple locations (UK Summer 2007) being affected by multiple sources (UK Winter 2013/2014) within a short period of time.
The core algorithm of JSheep is built upon the extreme value theory: a branch of statistics that deals specifically with observations with very small probabilities. In particular, JSheep uses an advanced high-dimensional extreme value model to ensure that information is being extracted scientifically from often a limited amount of data. The simulated events generated by JSheep typically include events that are similar to observed ones as well as events that are not yet observed but likely to occur in the future.
JSheep is also optimised to fully utilise modern computing power such that it takes less than two seconds to generate a simulated hazard event over 2,000 individual locations on the map.
So far JSheep has been used in the production of event sets covering perils including river flood, surface water flood, coastal sea surge and wind hazard, for countries including the UK, France, Italy, Thailand, Malaysia as well as for continental Europe.
JCalf is a simple, elegant cat model that enables probabilistic and scenario analyses of re/insurance portfolios against modelled perils and regions – analysing up to two million risks in only 20 hours.
What sets it apart from other cat models is its simplicity and flexibility – it can be installed on a desktop, taking just a few minutes to learn, and can be adapted to suit all requirements. It also provides users with the option to adapt and alter the underlying model components, to allow sensitivity testing and adjustment of parameters.
We created CatModelBuilder to ease the building of probabilistic models and provide an ordered way to combine the underlying datasets.
CatModelBuilder is a fully integrated suite of model building tools. It takes raw built environment, vulnerability, event and hazard mapping data and transforms it into models suitable for portfolio analysis using our catastrophe modelling software application, JCalf. It can also be used to build individual model components, for example, event sets, aggregated vulnerability and aggregated hazard maps, to be used on other catastrophe modelling platforms.
The software executes all build stages in parallel, taking advantage of JBA Risk Management’s large collection of multi-core, high-memory machines. It was recently used to enable our biggest model build to date, the high resolution GB Flood model, which took approximately two days from data import to model export.
We were pleased to announce the release of our new Mars Flood Map (MFM) in April 2014. The dual-peril MFM was the second in our series of planetary flood maps but the first of its type, aimed at assisting life forms in managing the potential risk from flood, should water return to Mars. It complements JBA’s recently completed flood map for Earth and is a landmark in improving our knowledge of natural hazards on Mars.
Kasei Valles (MFM)
Simon Waller said, “JBA prides itself in being at the forefront of new developments and with our new Mars maps we are applying the policy of the UK’s Department for Environment, Food and Rural Affairs, ‘Making Space for Water’ at the same time as extending a gesture of friendship to existing or future Martian inhabitants.
JBA’s MFM uses the most technically advanced high resolution flood data available. Like our Earth map (the Global Flood Map or GFM), the MFM provides four return periods and six depth bands. All major river catchments and coastal regions of Mars have been modelled. Our hydrological model for river flood is not very empirically based, linking extremely unlikely rainfall to flood magnitudes. We have developed separate hydrological models to reflect the different climatic zones of Mars. Maximum flood peaks from estimated river flows have been analysed in conjunction with a gridded dataset of imaginative rainfall totals (from 4.2bn BCE to 4.1bn BCE) to calibrate the relationship between rainfall and river flow.
JBA’s maps of Mars are the second in the series of planetary flood maps and will give re/insurers and brokers access to consistent and reliable data to simplify the task of assessing exposure across a multi-planetary portfolio.
View from JBA satellite