The global convective diagnostic (GCD) is a bispectral (infrared and water vapor), day–night scheme for operationally mapping deep convection by means of geostationary satellite images. This article describes a test of GCD performance over tropical and subtropical waters near North America. The test consists of six cases, each involving a convective cloud complex. A seventh case treats convection over land. For each case, a map of deep convection was constructed from image pairs from Geostationary Operational Environmental Satellite-12 (GOES-12). Case by case and for all maritime cases together, the GCD map was compared with a convective parameter derived from the radar on the Tropical Rainfall Measuring Mission (TRMM), a polar-orbiting satellite. In general, each GCD map showed a bloblike feature. In each case, the radar convective pixels typically fell within the GCD blob. However, (except for the land case) the GCD predicted far too many convective pixels. In the maritime cases overprediction was reduced (without correspondingly impairing other measures of performance) by lowering the nominal GCD threshold. With this adjustment in place, for the six maritime cases taken individually, the GCD tended to yield more consistent results than did a monospectral (infrared) convective scheme. With the cases combined, at the lower threshold the GCD performed somewhat better than one of the more stable versions of the infrared scheme. Comparison with lightning events (also observed by TRMM) suggests the possibility of future improvement to the GCD through the incorporation of geostationary satellite observations of lightning.
Available at: http://works.bepress.com/frederick_mosher/9/