Seasonal Forecasting - Isis Project No 1042
Problem
Predicting weather conditions for longer than a day or a week has long been possible. There are aspects of a climate system, which vary on longer time scales, that can bias the probability of their occurrence.
Often the relationship between climate and impact is insufficiently accurate, leading to erroneous predictions. For example, changing levels of greenhouse gases can affect forecasts.
Background
Current climate observations are input into dynamical models of which there are many types, which are then run forward in time to produce a forecast. These types of models are based on a baseline climate that the model will return to once the short-term forecast is produced. This makes them unsuitable for seasonal forecasts.
Perfect ensembles, or weighted averages of analogous past climate systems, can be used for seasonal forecasting. Use of these ensembles has been restricted because it can take millions of years to observe the exact climate. Computer climate modelling speeds up the return time of the atmosphere by making assumptions based on past climates.
The Oxford Invention
A model based approach to seasonal forecasting using perfect and sub-ensembles. The model relies upon distributed computing to provide the necessary processing power and speed. The model is distributed to PCs, each one with different initial values, allowed to run over a period of time in to the future. The simulations that are consistent with recent observed climate change are used as the basis for ensemble forecasts of the future change.
Commercialisation Opportunity
Isis Innovation Limited is in a position to offer licences to this technology. The technology is the subject of a UK patent application.
Other Cleantech Projects
974 - Biological Fuel Cell
1103 - Combined Oxygen and Carbon Dioxide Sensor
1415 - Coastal Protection and Flood Defence Design
1489 - Improved Heating Control in Buildings
2871 - Smart Utility Meter
Request Further Information: Project Number 1042 - Seasonal Forecasting