AN EXAMINATION OF THE DIURNAL CYCLE IN OCEANIC TROPICAL RAINFALL USING SATELLITE AND IN-SITU DATA

Diurnal variations in tropical cold cloudiness are examined for the pe riod 1986-90 for each 2.5-degrees latitude-longitude area in the globa l Tropics. The fractional coverage of cold cloudiness, as determined f rom various IR brightness temperature thresholds, has been used as a p roxy for tropical convective precipitation, as direct observations of rainfall are unavailable for much of the earth, especially over the oc eans. Variations in fractional coverage of cold cloud for three differ ent temperature thresholds are examined: 235, 225, and 215 K. The resu lts of this study indicate that cold cloud is most frequently observed over land between 1800 and 2100 local time and is independent of the temperature threshold used. Over the tropical oceans, however, the tim e of maximum occurrence of cold cloud varies substantially with the te mperature threshold employed. Coldest cloud-top temperatures (< 215 K) are found to occur much earlier in the day than warmer cloud tops and peak between 0300 and 0600 local time, which is consistent with many earlier limited-area studies. This observation is further confirmed fr om precipitation intensity differences between morning and evening obs ervations from microwave satellite data. An interesting out-of-phase r elationship between oceanic and continental convection is also discuss ed. Ship reports of weather type from the Comprehensive Ocean Atmosphe re Data Set are examined as are hourly rainfall amounts from optical r ain gauges on moored buoys that were deployed for the Tropical Ocean G lobal Atmosphere Coupled Ocean-Atmosphere Response Experiment. Both of these data sources also indicate the preference for predawn oceanic h eavy rainfall and convective activity. A cursory examination of the di urnal variations in short-range (6 h) rainfall forecasts from the Nati onal Meteorological Center Medium-Range Forecast Model are compared wi th the satellite and in situ results. The daily variations of these fo recasts, which are made four times daily, indicate that the diurnal be havior of the model is in reasonable agreement with that of the satell ite and in situ observations.