A Numerical Weather Prediction-Based Infrastructure for Tropical Meteorology Research and Operations in Brunei

Being able to improve the understanding of the societal and economic impact of severe weather events on a regional scale and concern for how they may evolve as the climate of the earth changes is a motivation for investment in additional, localized research. As an example, working with IBM, the Universiti Brunei Darussalam (UBD) has established a Centre for Regional Climate-Weather Modelling focused on tropical meteorology. Given Brunei's location on the northern coast of Borneo, issues related to rainforests are of keen interest as well as renewable energy and climate change impacts on agriculture and flooding in the region, and overall concern for sustainability. This international collaboration from several organizations within IBM, working with UBD has two key components to establish a foundation to support both research and operations, namely computational resources and modelling capabilities. Given the importance of modelling, the first component and prerequisite is to have appropriate computational resources. To provide both a unique capability in the region and a system appropriate to specific computational tasks as well as being scalable and energy efficient, UBD selected an IBM Blue Gene/P (BG/P) to support the Centre.

The second component focuses on specific modelling capabilities, starting with those that are relevant for short-term weather impacts on public safety, agriculture and energy. This builds upon the on-going work at IBM Research connecting the business implications of these issues to weather models, dubbed “Deep Thunder”. In particular, it is the ability to predict specific events or combination of weather conditions with sufficient spatial and temporal precision, and lead time coupled to the operational impacts to enable proactive allocation and deployment of resources (people and equipment) to mitigate the effects of severe weather.

The first step in this collaboration was the adaptation of Deep Thunder to the region to support operational weather-sensitive business operations in Brunei on the BG/P, which serves as a core for broader efforts in both climate modelling and flood forecasting. Given the geography of Brunei, such capabilities have significant challenges. In addition to its tropical setting along the coast of the South China Sea, the mountainous areas in the east, and the complex terrain and rainforests of Borneo must be considered. Of particular interest is the so-called Borneo Vortex, which occurs during the northern hemisphere winter when cold fronts from Siberia blow across the South China Sea and interact with the Equatorial trough, which is modulated by the Arctic Oscillation and the El-Niño-Southern Oscillation.

Current state-of-the-art numerical weather prediction (NWP) codes operating at the meso-gamma-scale have been shown to have potential in predicting specific events to address some of the aforementioned applications. However, they have not been widely applied to this region and these issues. To begin, the WRF-ARW (version 3.2.1) community NWP model was adapted for use in Brunei. A number of model configurations were developed to generate many numerical experiments as part of retrospective analysis of recent impactful precipitation events. They included horizontal resolution in the 1 to 1.5 km range for Brunei and the surrounding area, and as high as 3km resolution for all of Borneo. To address the influence of the complex terrain, various vertical resolutions were evaluated as well as adjustment to the nesting configuration (i.e., three or four two-way nests) to avoid numerical instabilities. The model orography was developed from altimetry data at 90m resolution available from the NASA Shuttle Radar Topography Mission. Validation was limited because of the paucity of data available in the region. Surface observations provided by agencies of the Brunei Government have been used along with remote sensing data from agencies outside of Brunei. The experiments were run as hindcasts, and therefore, used data at 0.5 degree resolution from the NOAA Global Forecasting System for initial conditions and lateral boundaries. The configuration has parameterization and selection of physics options appropriate for the range of geography in the region. It included the use of a sophisticated, double-moment, 6-class, explicit cloud microphysics scheme. In addition, scaling experiments were done to maximize the efficiency of the model configuration on the BG/P system at UBD with the goal of enabling an operational forecasting environment. The results led to a three-way nested configuration, which includes a 4.5km nest covering all of Borneo with a large 1.5km nest covering Brunei and the surrounding area. To address the orographic influence of the complex terrain, 45 vertical levels were established with typically the lowest ten being within the boundary layer. This configuration was placed into operations in November 2011, producing a single 48-hour forecast per day, initialized at 00 UTC.

To provide information on potential flooding, a mathematical model was implemented simulating surface flow and water accumulation using a locally conservative approach by employing the shallow water equations. Such gravity-driven flow uses detailed topographic data along with precipitation estimates generated by the aforementioned WRF-ARW configuration to determine if a site could receive a surface runoff that leads to flooding. The flood inundation model is based upon a finite volume method and uses an explicit time integration scheme. The model is second-order accurate in space and time and is aimed to make simulated flood flows (of the order of a few meters) over large spatial domains. The model employs a two-dimensional structured cartesian mesh for capturing local topographic effects. The frictional and infiltration properties for the study area are derived from land use and soil type data. Since the objective in this work is to model the flooding due to intense precipitation, evapotranspiration is currently not included in the model.

We will outline the research objectives of this work in Brunei and the challenges of enabling it. We will discuss some of the scientific and computational results to date, and lessons that were learned. Since this is work in progress, we will present its current status and our plans for future work.