This report summarizes the work accomplished during the first phase of an investigation concerning methods of introducing digitized satellite imagery into short-range, objective forecasting operations. The data archive being assembled for this study is described, with particular attention given to the steps taken to maximize the accuracy of the satellite imagery. These steps included 'fine tuning' the navigation and selecting procedures for 'normalizing' the data by correcting for the effects of Lambertian and anisotropic scattering. Consistency of the data, spatial and temporal, was tested by analysis of ground reflectance during cloudless days, and a pilot test of the specification of single layers of clouds was conducted. Both of these tests gave encouraging results. An investigation of specifying precipitation rate, using just the visible reflectance and infrared temperature of the cloud top, also produced good results. Nomograms for the average rate during the hour following the satellite observation, as well as for the probability of observing more than 0.01 in. and 0.10 in. of precipitation, are illustrated. Two appendices present the geometrical and optical equations relevant to the investigation. (Author).

This report describes the development and testing of an objective technique to forecast cloudiness and precipitation through extrapolation of satellite imagery. By utilizing on objectively determined cloud-motion vector, the technique makes local forecasts of satellite parameters (brightness and IR temperature), with high temporal resolution, using simple linear extrapolation. Algorithms are then used to convert the satellite parameters to total cloud cover, probability of 1-hour precipitation, and presence of low, middle, and high clouds. The test program computed motion vectors and made forecasts out to 7 hours, in half-hour steps, at 30 locations. The program was tested on 12 spring and fall cases, using half-hourly GOES imagery. For periods beyond 2 hours, forecasts of cloud cover and precipitation were markedly better than persistence, which deficiencies in specification hindered short-period performance. Forecasts of cloud layers were worse than persistence due to inadequate specification algorithms. The net results were quite encouraging, and further refinements and developments are planned.

Bi-spectral satellite thresholds for precipitation specification are explored with visual and infrared satellite data collocated with Service-A hourly observations for 137 surface stations in the southeastern United States. The data span the month of August 1979 and total 70,623 observations, including 538 daylight precipitation observations. The distributional and statistical differences of four satellite resolution sizes ranging from 484 to 2025 nmi2 are explored and determined to be significant in the representation of weather conditions. Precipitation and no-precipitation data can be statistically differentiated with the visual and infrared mean and standard deviation values. For overcast ceiling reports, a simple linear bi-spectral threshold based on a 50% probability of precipitation is defined as extending from albedo 1.00 to 0.60 with associated cloud top temperatures 290K and 210K, respectively. For overcast and broken ceiling reports, and albedo greater than 0.80 specifies a 50% probability of precipitation. (Author).

This study examined the capability of current geostationary satellite sensors in resolving and monitoring important mesoscale rain areas. Shallow mesoscale precipitation patterns were analyzed with radar and recording rain gage data associated with mid-latitude cyclones in New England. These cyclones were characterized by relatively little stratiform cloudiness in the upper troposphere. Radar thresholding was used to determine the spatial and temporal scales of small mesoscale areas (SMSAs), large mesoscale areas (LMSAs), and major echoes. Radar rainfall estimates were used to determine the relative amount of rainfall contributed by the mesoscale features. Radar analysis of the three cases showed that the spatial scale of major echoes, LMSAs, and SMSAs were large enough for detection by GOES sensors. Major echoes and the LMSAs contained within them were found to have lifetimes that are long relative to the half-hour interval between GOES transmissions. The duration of some SMSAs and all cells, however, was on the same order as the time interval between GOES transmissions. In addition, radar rainfall estimates showed that the major echoes, LMSAs, and SMSAs were the major contributors to mesoscale rainfall. For the cases examined in this study, GOES sensors were not capable of distinguishing the individual mesoscale precipitation areas detected by radar. Therefore, GOES sensors do not appear adequate for developing satellite rainfall estimation techniques to specify shallow mesoscale precipitation.

A previously developed advection forecast technique was modified to include data extracted from satellite imagery. A forecast experiment was then conducted using a data base gathered at AFGL during March 1984. This experiment was designed to test the usefulness of : (a) 3-hour forecast updates, (b) a biquadratic interpolation, and (c) cloud and precipitation information from satellite imagery. The test results confirmed earlier tests in that advection using space-averaged 500-mb winds produced the best overall scores and that in general the scores for 1 - 15 hours were better than persistence. The age of the advection flow (3, 6 or 9 hours old) did not affect forecast score, making updates useful. The biquadratic interpolation procedure produced better fits to observation than bilinear and appears to have improved forecasts. There was but a small benefit from adding satellite information to surface observations when forecasting cloud cover and hourly precipitation. the difficulties of trying to forecast even 30 to 50 percent of the time-change variance suggest that alternative approaches such as mesoscale modeling will be needed for accurate, reliable short-range forecasts.