5.5 A Python Toolkit for Earth Science Research and Education

Tuesday, 24 January 2017: 5:00 PM
Conference Center: Chelan 5 (Washington State Convention Center )
Andrew T. White, Univ. of Alabama, Huntsville, AL; and B. Freitag, U. S. Nair, and A. P. Biazar

Process studies for earth science research often require analysis of multiple datasets including numerical model outputs, in situ observations, satellite and ground based remote sensing datasets.  Evaluation of numerical model performance through comparison against co-located observations is an important step in process studies using numerical models.  Fusion of numerical model outputs and observations are often needed for process studies.  Numerical model outputs and observations are also valuable for inquiry based learning in a classroom setting.  However, analysis of diverse datasets involved in earth science research requires considerable programming expertise.  Started as part of the class project in graduate level numerical modeling course at University of Alabama in Huntsville (UAH), we have developed a python package designed for research and education utilizing a variety of earth science datasets.  The python package is currently being used to aid in the analysis of satellite cloud assimilation research conducted at (UAH).          

The GUI toolkit consists of a series of objects that could be directly utilized to access and co-locate different earth science datasets by experienced programmers.  A GUI interface on PyQT and matplotlib libraries, provides user friendly access for beginners in utilizing the analysis capabilities of these objects.  The collection at present includes objects for accessing and manipulating WRF, CMAQ, MERRA, NCEP reanalysis, geostationary satellite, radar datasets and meteorological station location.  Whenever possible, OpenDap access is utilized.  Objects are designed to allow users to easily incorporate computations for variables derived from existing data fields. 

The GUI interface allows for displaying slices of three-dimensional datasets along any of the orthogonal planes and also along user specified curtains.  Other features include profile and time series plots at a specified location and Hovmöller diagrams.  The interface also provides capabilities for marking polygonal selection regions on the plot and computing statistics for the sampled regions.  We are in the process of adding additional datasets and capabilities to the toolkit.  Datasets that are being added include OLAM, RAMS, TRMM, GPM, CloudSat, Calipso, MODIS land product datasets, CCMP and storm track data.  Processing capabilities that are being added include phenomenon detection and segmentation capabilities.

We have designed the GUI with the objective of sustaining this tool kit through community participation.  Objects are provided with methods that easily allow users to incorporate their own custom fields.  The GUI object is also designed so that users could easily develop their own plots and analysis which could be incorporated into the GUI.  This toolkit has greatly enabled graduate research of the developers and is expected to assist researchers studying a varied of earth science topics from meteorology to air quality.  In the classroom, this tool provides easy access to over four decades of atmospheric data and its utility inquiry based learning is being evaluated.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner