Implementation of inexpensive cloud-based forecasting systems
ABOUT THIS INITIATIVE
Cloud computing offers significant opportunity to rapidly deploy early warning hazard prediction capabilities to least-developed nations at a modest cost and in ways that can be tailored specifically for the unique needs of the diverse communities throughout the region. This initiative has focused on using cloud computing services to implement a regional forecasting solution (e.g., Weather Research and Forecasting: WRF) as an adaptable, on-demand, reliable forecast system for early warning and decision support system applications. The cloud-based forecast system is ideal for stakeholders looking for low-cost solutions without maintaining a high-performance computing system.
- NEED TO ADD
- NEED TO ADD
National Meteorological and Hydrological Services (NMHS’s) in the least-developed countries have the need for regional weather and hydrology forecast systems that provide timely information to support impact-based forecast operations in an effort to minimize losses from phenomena such as floods, storm surges, and wind damage. NMHSs in least-developed countries often lack infrastructure such as affordable and reliable commercial power, high-speed internet links develop and facilities to maintain a high-performance computing (HPC) system generate reliable forecasts. With the recent innovation of global cloud computing, implementing a sustainable regional forecast system for a NMHS has now become possible. The goal of this project is to develop a cloud-computing numerical weather prediction (NWP) framework that is designed to be adapted and implemented at NHMS’s. This framework than can extended to other hydrometeorological applications such as flood forecasting, tropical cyclone forecasting, and seasonal prediction.
What is Cloud Computing?
An innovative solution to these limitations at NHMS’s is to make use of global cloud computing centers. Cloud computing means that the computing resources required to run forecast system are located at remote, computing centers located in various locations globally (e.g., US, Europe, South America) that provide low-cost, sustainable, secure, HPC computing systems. A benefit of cloud computing is that commercial vendors such as Amazon Web Services (AWS), Penguin, and Microsoft Azure only charge when the computing services are used. Cloud computing centers provide a variety of services that include HPC for running complex NWP models (e.g., Weather Research and Forecasting (WRF): https://www.mmm.ucar.edu/weather-research-and-forecasting-model), web access capabilities, and data storage services in a seamless framework.
Benefits of Cloud Computing
Cloud computing offers significant opportunity to rapidly bring early warning hazard prediction capabilities to least-developed nations at a modest cost and in ways that can be tailored specifically for the unique needs of the diverse communities throughout the region. A main benefit to cloud computing services is the reduction of the cost of purchasing and maintaining a locally installed HPC system. An operational regional forecast modeling system (e.g., WRF) can require a HPC system that can cost upwards of several million dollars to purchase and similar amounts to maintain over the useful lifetime of the HPC systems (~5 years). Through economies of scale, commercial cloud computing vendors offer very substantial computing resources at costs that are a fraction of than the cost of purchasing and maintaining a dedicated HPC center. Since cloud computing is offered “on demand”, users only pay for the compute resources that they need.
[INSERT IMAGES HERE]