Sphere / Spheroid spatial accuracies

Numerical Weather Models treat the Earth as a perfect sphere, whereas, most geographic data are based upon a spheroidal Earth. When numeric model output is overlaid over the Earth surface characteristics, significant location error can be introduced if the proper conversions are not taken into account. The assumption that all positional coordinates are equivalent introduces positional errors of up to 21 km at mid-latitude locations. Figure 1 illustrates the positional shift when positions from a sphere model are plotted directly onto a spheroid. Positional error may be introduced at an earlier phase of numeric weather modeling if the input data locations for the models, such as elevation, land cover, etc., are also based on a spheroid but brought into a sphere without any transformation taking place. This research seeks to understand how WRF's treatment of Earth as a perfect sphere (rather than an ellipsoid) -- including the manner in which it ingests topography and land use datasets -- affects the accuracy of the weather simulations.  The project participants are testing whether translating the input parameter information such as elevation and land use from a spheroidal based earth (geodetic) to a sphere based earth (geocentric) will improve positional accuracy of the model results. Current experiments over Houston, TX and Denver, CO are conducted under the NASA-funded SIMMER project. 


Jennifer Boehnert (NCAR/RAL)
Michael Barlage (NCAR/RAL)
Andy Monaghan (NCAR/RAL)
Richard Koehler (NOAA/NWS)
Olga Wilhelmi (NCAR/RAL)


NASA and NCAR base. NCAR is sponsored by the National Science Foundation