Feature - Predicting the almost unpredictable
|
||
|
|
||
Weather forecasters do not always get their predictions right, especially when it comes to "severe weather events" that pose a hazard to life and property. Western Europe is no stranger to such very rare but extremely dangerous weather, as demonstrated by the 'deep freeze' that the UK experienced in January earlier this year. With temperatures dropping to -21C (-6 degrees F), the UK encountered its coldest weather front in 30 years. The country's Federation of Small Businesses said that by the time all the economic costs of the three-week cold spell were added up, they topped 1.2 billion pounds - the equivalent of 1.4 billion euros, or $1.8 billion dollars (US), as of the exchange rates at press time. Now, at the opposite end of Europe, Armenian scientists have shown that such severe weather can be more accurately predicted, using the combined power of grid computer processing and methodologies used to predict climate change. The problem facing the Armenian Hydrometeorological and Monitoring Service was challenging. Due to Armenia's complex geography and diverse terrain, the country's hills and mountains can greatly influence storm fronts, making thunderstorms, strong winds and flash floods difficult to predict. (Hydrometeorology - the study of water and energy transfer between the land surface and the lower atmosphere - is also the science used to predict climate change.) In addition, most prediction methods use Numerical Weather Prediction (NWP) maps which highlight 'synoptic' scale (large-scale analysis of an area of more than 1,000 km) weather patterns in low-quality resolution. Armenia is located on the edge of these maps, thus making it difficult to obtain whole and complete views of the weather fronts surrounding the country. Also, the prediction model doesn't allow for the integration of satellite data - important for providing real-time, top-view, high-resolution images over large geographical areas. What was needed was a weather analysis model at the 'mesoscale,' encompassing an area of five to hundreds of kilometers. At this level, finer-grain studies can be made, allowing for research into events such as thunderstorms. |
||
Enter WRF The solution? WRF (Weather Research Forecast) - a new computing model developed by scientists using the processing resources of SEE-grid, the South-Eastern European grid. The method works by integrating satellite imagery with the observed data from four hydrometeorological stations in Armenia. As a result, lead time and accuracy of severe weather prediction has been significantly improved, as storm fronts can now be analyzed in all directions in and around the country. The calculation takes about three hours, and the prediction can forecast a severe weather event up to three days in advance of it occuring. As for accuracy, tests of this new model have shown that by combining grid processing with hydrometeorological data, weather forecasts generally matched actual results one out of four times, a much better ratio than previously. (In some cases there were discrepancies due to lack of data from isolated mountain areas.) Hrachya Astsatryan, head of the HPC laboratory at the National Academy of Sciences of the Republic of Armenia, said: "The collaboration between technology and science is crucial to implement new distributed computing infrastructures for operational weather forecasting and research." This is a step closer in providing more accurate weather forecasting, mitigating future economic losses and most importantly, allowing forecasters to get their predictions right about the most unpredictable of weather events. -Adrian Giordani, for iSGTW |
Our Underwriters
Thank to you our underwriters, who have supported us since the transition from International Science Grid This Week (iSGTW) into Science Node in 2015. We are incredibly grateful.
