Abstract
Errors in operational forecasts produced by high-speed electronic computers can be classed broadly into two categories: (1) those resulting from inadequacies of the dynamic model, and (2) those resulting from poor specification of the initial fields. Many regions of the Northern Hemisphere, particularly oceanic areas, are poorly observed in terms of conventional meteorological data, especially upper-air data. The SINAP (Satellite Input to Numerical Analysis and Prediction) Project at the Weather Bureau's Meteorological Satellite Laboratory has been working to develop techniques for incorporating information derived from satellite cloud pictures into the operational numerical analysis in data-sparse areas.
Trial reanalyses of the National Meteorological Center (NMC) 500-mb. stream function analysis, or its Laplacian, were performed for data-sparse areas of the central and eastern Pacific Ocean using an analysis modification technique consisting of two steps: (1) inferring features of the flow pattern or of the field of large-scale vertical motion from an interpretation of the TIROS-viewed cloud patterns, and (2) modifying the 500-mb. analyses to produce an appropriate vorticity advection field. Underlying this method are certain simplifying assumptions about the relation of the cloud field to the vertical motion field on the one hand, and of the vertical motion to the vorticity advection on the other.
Application of the method and the results obtained are illustrated for one case. Thirty-six-hr. barotropic forecasts were run from both the original NMC analysis and the SINAP modified analysis and then compared with the verifying chart. Verification statistics, such as the root mean square (RMS) error of the stream values and of the vector geostrophic wind, are presented for the case illustrated and for five additional cases. Significant reductions in forecast error were achieved in most cases, the overall average reduction in the RMS error of the wind being 5.4 percent.
