How grids help thwart Mother Nature

Hurricanes or typhoons, volcanic eruptions, heat waves, tornados, droughts, and landslides – the list of natural disasters is long, as is the time it takes to recover from their devastating impact. Every year, they leave a trail of destruction in their wake, resulting in countless injuries and fatalities worldwide and causing extensive socioeconomic damage.

Efforts in hazard mitigation, focusing on minimizing the vulnerability of disaster-prone areas and maximizing their preparedness, as well as the development and implementation of emergency response strategies, mainly depend on accurate predictions of when and where the disasters will occur. Running high resolution weather simulations, however, requires extensive geophysical information, global or local meteorological data, and a substantial amount of processing power, depending on the complexity of the calculations.

Now, researchers can run those simulations using grid computing and a standard web browser, thanks to the launch of the new gLite-based WRF (gWRF) web portal. The portal was developed by the Academia Sinica Grid Computing Centre (ASGC), in close collaboration with Academia Sinica’s Research Center for Environmental Change, and unveiled at the International Symposium on Grids and Clouds 2012.

"By making use of distributed grid resources, gWRF is capable of handling large simulation requests at the same time," said Eric Yen of ASGC.

In an effort to shield users from the technical complexity of grid environments, the new web portal serves as an alternative to the existing command-line interface and does not require prior installation.

"WRF is a powerful simulation package used for meteorological research. In order to facilitate collaborative research on regional extreme climate and weather simulations,
we have deployed the numerical weather simulation model WRF on the worldwide grid computing infrastructure," said Hsin-Yen Chen of ASGC.

Advancing meteorological and climatological research

gWRF draws on the Weather Research Forecasting model, one of today's most widely used atmospheric models. As a mesoscale weather simulation system, it provides a high degree of flexibility and can be easily adjusted to examine weather phenomena of different scales, ranging from less than one kilometer to hundreds of kilometers.

"Compared to its predecessor MM5 that has now been largely replaced, WRF offers a major advantage, as it features a number of optional modules and extensions that can be used for specific purposes, such as the simulation of wildland fires and dust storms," said atmospheric scientist Chuan-Yao Lin, who is based at the RCEC in Taipei.

In conjunction with modules such as WRF-Chem, an extension which adds chemical variables to the simulations, Lin and his colleagues have employed WRF for a broad range of purposes, including the monitoring of air quality and radiation emissions, the tracking of dust and biomass burning and as well as an impact evaluation of heat development in urban areas. Plans to integrate WRF-Chem into the web portal are also currently being developed.

Tracking typhoons

In 2009, WRF was used to compare historical meteorological data to typhoon Morakot, with a particular focus on the heavy rainfall and winds that occurred during this period of time. Morakot was an intense tropical cyclone that affected Japan and wreaked havoc on the Philippines, Taiwan, and China, killing more than 700 people overall. Within 100 hours, a record-breaking amount of rainfall hit Taiwan that resulted in devastating landslides and casualties, causing an estimated total damage of 3.3 billion US dollars (or 2.5 billion Euros, as of press time) in Taiwan alone.

In the case of typhoon Morakot, this unprecedented rainfall was caused by the convergence of the southerly component of the pre-existing strong southwesterly monsoonal flow and the northerly component of the typhoon circulation. Based on fine-scale WRF simulations, Lin and other scientists identified the convergence itself as the source of the heavy rainfall that was caused by its interaction with the topography; convective cells also existed within the typhoon’s main rainband.

Weather simulation systems, such as gWRF, can not only be employed to examine a past weather event but also to predict regional climate changes, based on a number of previously specified parameters. Even daily weather forecasts can be calculated by using gWRF if adequate processing power is provided.

Such simulations can help improve the accuracy of future projections, since some weather events, such as typhoons and hurricanes, show a significant tendency to follow certain paths and patterns. By modifying the simulation parameters, researchers can also easily estimate the effectiveness of various mitigation measures and how changes to the individual specifications might affect the development and impact of natural disasters.

Harnessing the power of global grids

gLite middleware enables WRF to run on the global grid infrastructure. Using gLite for system management allows for efficient job assignments, thus maximizing the benefits of distributed computing resources. Since WRF simulations generally require large amounts of processing power, the global grid infrastructure accelerates the simulations to a significant extent.

The current web interface is based on WRF version 3.1.1, with plans for further upgrades in the near future. Results can be downloaded either individually or as a package. The new web interface also facilitates the submission, retrieval, and monitoring of jobs, and provides debug information in real time.

As the new gWRF interface has been successfully implemented using EUAsia resources, users are only required to obtain a valid user certificate and become a member of the EUAsia Virtual Organization, which serves as a platform for all e-Science applications in Asia. The EUAsiaGrid was an EU-funded project, formed by four European and 13 Asia-Pacific partners, from 2008 to 2010. Coordinated by ASGC, which is in charge of the daily operations of all resource centers contributing to EUAsia, the resources are currently still available to support e-Science projects throughout Asia.

To improve the overall efficiency and allow for greater flexibility in selecting between the post-processing and visualization tools that are currently in use within the scientific community, users are required to handle the pre- and post-processing of the data on their local computers, particularly since they do not require extensive processing power.

With regard to the future development of gWRF, Yen said that they are committed to "continuously improving performance and enabling flexible visualization." Another goal is to collaborate with other climate and weather research communities on disaster mitigation and extend related e-Science activities to other grid infrastructures, said Chen.