21st Conf. on Severe Local Storms and 19th Conf. on Weather Analysis and Forecasting/15th Conf. on Numerical Weather Prediction

Thursday, 15 August 2002: 4:45 PM
Uncertainty in numerical mesoscale modeling by the use of a topography defined via map projections
Marco A. Nuņez, Universidad Autonoma Metropolitana Iztapalapa, Mexico City, Mexico
Poster PDF (128.9 kB)
Computational mesoscale models such as RAMS, MM5, ARPS and HOTMAC, use conformal map projections to define the topography with respect to a tangent-plane XYZ coordinate system where Z is replaced by a terrain-following coordinate. The procedure consists in computing a point (X,Y) on the XY-plane via the map-projection of a point on the terrestrial sphere with geographic coordinates (l,f), and if H is the terrain elevation on (l,f), then it is assumed that the terrain elevation on (X,Y) is also H because the map projection takes into account the earth curvature. In this work it is shown that (i) the correct terrain elevation Z on (X,Y) cannot be computed because map projections do not consider the datum H to compute X and Y. (ii) The datum H is only an approximation of the correct elevation Z, and the approximation is valid on a horizontal domain D(L)=2LX2L with L bounded by 60 km when the terrain elevation data have an uncertainty of 30m. Applications of mesoscale models have used a domain D(L) with L=650, 882, 1665 km. In these cases the error of the terrain elevation reaches up 67, 120 and 450 km, respectively. A coordinate transformation to compute the correct terrain elevation from a Digital Elevation Model, is given. Analytic solutions of the governing equations show that the use of a map-projection topography in the definition of a terrain-following vertical coordinate generates a wrong atmospheric flow when L is large.

Supplementary URL: