The Use of GIS and Python Scripting for Processing and Mapping National and Regional Climate Data Simulations (invited)

In supporting the development of the 2013 National Climate Assessment report, we generated a large collection of climate maps to assist Assessment authors and other climate scientists in making better informed decisions about national and regional climate simulations. The use of Geographic Information Systems (GIS) software was critical for data management and processing, as well as mapping and visualization. In addition, Python scripting allowed for the automation of data processing both outside and within the GIS domain. The datasets of interest included annual mean simulations of multiple precipitation and temperature climate variables from 15 Climate Model Intercomparison Project phase 3 (CMIP3) models and 8 North American Regional Climate Change Assessment Program (NARCCAP) models. Seasonal mean simulations were also processed for some NARCCAP data.

Original datasets consisted of a combination of NetCDF data and non-GIS gridded text files. For the NARCCAP simulations, multi-model mean datasets were created for each corresponding climate variable by summing and averaging the data using Unidata's NetCDF Utilities via Python-scripted automation. Additional calculations were then performed to create simulated difference datasets for each variable by subtracting current climatology grids from future multi-model mean simulations. Within the GIS domain using ArcGIS Desktop software, other spatial geoprocessing steps on both vector and raster data included defining projections, re-projecting, vector point to raster grid interpolations, grid shifting, clipping, extracting, and data format conversions. GIS-based Python scripts (ArcPy) allowed for automated geoprocessing of several hundred datasets at the continental U.S. and 8 individual regional scales.

Processed datasets were then placed into appropriate cartographic displays to potentially be used as figures in the Assessment report. Careful attention was given to map and visual detail in order to display the information legibly, without map clutter, and at varying resolutions. This included effective data symbology classification schemes and color choice following the ColorBrewer cartographic standards. Multiple categories of modeled difference uncertainties were shown through cross hatching, mask overlays, and color. A final collection of approximately 150 cartographic figures displaying over 550 mapped climate datasets was the result of the effective use of GIS and Python scripting.