Thursday, 14 January 2016
There are massive amounts of weather data generated in real-time on a daily basis. Archives of predicted and observed data can easily become enormous. Both researchers and end-users of weather information require tools that can perform rapid analysis and easily share these data. For example, our research has recently involved providing the Indiana Department of Transportation (INDOT) with hourly, gridded information regarding the severity of winter weather over multiple seasons. The end-user in this case was not familiar with data formats that are most commonly used in meteorology, such as GRIB, which is the WMO standard for binary data on regularly spaced grids. While there are several useful tools available for processing GRIB files, processing large amounts of GRIB files remains a challenge. There is no standard file naming convention, and obtaining access to long archive of forecast data is difficult. Scripts for processing forecast data must be rewritten for each new dataset that is included for analysis. Common analysis functions such as accumulation, aggregation over arbitrary spatial and time domains, visualization, and animation were also desirable for this project. As a result, we developed a database and other tools that provide services to store and search through meteorological data. This service can provide users with an interface that is simple to interact with and quick to develop applications against, a storage facility that is capable of scaling with the user's data, a system that imports their data as quickly as it is produced, is capable of storing the data within a manageable amount of resources, and provides access to the data through web based services.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner