J6.3
Satellite imagery and virtual globe cloud layer simulation from NWP model fields

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 21 January 2010: 9:00 AM
B218 (GWCC)
John F. Galantowicz, AER, Lexington, MA; and J. J. Holdzkom, T. Nehrkorn, R. P. D'Entremont, and S. Lowe

Environment representations are used in a variety of simulation-based training exercises and sensor performance studies in which standardized meteorological product suites must be realistic, time- and space-correlated and physically consistent. Here we focus on simulation of satellite imagery and virtual globe (VG) cloud layer visualizations from four-dimensional numerical weather prediction (NWP) model representations using physics-based radiative transfer modeling. Satellite imagery products are designed to closely mimic operational imagery from a number of sources, including characteristic background coloring, various file formats (NITF, PNG, GeoTIFF), and embedded tags (e.g., geospatial, time). The VG cloud layer visualizations are derived from the same core physics but are rescaled to better simulate the more subjective point-of-view of the human eye and to accommodate the uniform brightness typical of VG base imagery.

In this presentation, we describe techniques for interpreting NWP model fields, diagnosing atmospheric cloud water path profiles, and generating imagery in visible, water vapor, and thermal infrared bands. Specification of the cloud liquid water profile is critical for realistic imagery generation, which depends on cloud height, opacity, and temperature. Approximate cloud water diagnostic methods are necessary to enforce consistency between imagery and other model products (e.g., total cloud cover and rainfall) produced along with imagery in a simulated weather products suite. Radiances are simulated by a combination of first-principles radiative transfer and interpolation against pre-calculated products, which speeds imagery production for non-window bands. VG cloud layer visualizations are simulated for placement (via KML) at prescribed VG altitudes (e.g., 2, 4, and 8 km), with two imagery layers placed at each altitude to provide separate cloud top and bottom views. Image layer transparency is set to mimic atmospheric layer and total column transmittances and a partially transparent gray-scale terrain overlay is also produced to mimic the effects of cloud shadowing on the VG base imagery.