Monday, 8 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
A regional impact of climate change has been assessed using various models that simulate the biophysical processes within an ecosystem. Regional climate scenario data including the Coordinated Regional Climate Downscaling Experiment (CORDEX) data have been open to the public through web database, e.g., the Earth System Grid Federation (ESGF; https://esgf.llnl.gov/). A data tool for the CORDEX data would be useful to minimize the efforts for processing the gridded climate data that have different spatial and temporal properties. The objectives of this study were to develop the CORDEX Data Support Library (CDSL), which would aid preparation of weather input data for different ecosystem models. Grid Analysis and Display System (GrADS) was used as the backbone of the CDSL. The CDSL was designed to parse metadata of spatio-temporal properties, to create ensemble datasets with weights, and to support high performance computing. This poster presents a use case in which the CDSL was used to load the CORDEX data from six domains including West Asia, East Asia, Europe, North America, South America, and Africa in order to calculate reference evapotranspiration (ET0) in R. ET0 was calculated for 25 years (1981 – 2005) using the FAO 56 formula. A small number of functions implemented in the CDSL was used to parse metadata of CORDEX file based on the internal database where records of spatial and temporal properties of each domain have been stored and to read the CORDEX datasets from different domains. When the option for high performance computing was turned on, the wall clock time to read the CORDEX data reduced by 25 % using 16 cores. These results indicated that the CDSL help process the CORDEX data for simulation of ecosystem functioning using different models. This would merit further studies to extend the functionality of the CDSL in order to support other data formats such as GeoTIFF.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner