Repository logo

Diurnal cycle of tropical deep convection examined using high space and time resolution satellite data


Infrared (IR) and visible (VIS) satellite data from GMS-4 with 5-km spatial and 1-hr temporal resolution was used to examine the diurnal cycle of deep convection over a sector of the tropical west Pacific warm pool (WP) bounded by 140°-180°E, 0°-20°N. Data were analyzed for 45 days of summer from 22 June 1994 - 5 August 1994 (JJA) and for 65 days of winter between 28 November 1994 – 31 January 1995 (NDJ). The synoptic backdrop for JJA was characterized by the monsoon trough, oriented northwest to southeast through the WP. Convection was largely focused along the trough. During NDJ, convection was concentrated within 5° latitude of the intertropical convergence zone (ITCZ) which was oriented east to west near the equator. December 1994 was characterized by an active phase of the intraseasonal oscillation (ISO) while January 1995 coincided with an inactive phase. Deep convective cloud was identified in IR imagery using brightness temperature (TBB) threshold techniques. Cloud forms associated with deep convection showed two distinct diurnal modes representing deep convection (TBB ≤ -60°C) and stratiform cirrus (-52°C ≤ TBB ≤ -23°C). Clouds with TBB warmer than -60°C and colder than -53°C comprised a mixed deep convection and cirrus anvil regime from the satellite's perspective with a diurnal cycle reflecting both modes of variability. The diurnal variation of cloud in these regimes was consistent for all time periods and for two tropical storms which occurred in the WP during December 1994. Based on these results and on previous studies, a -65°C cloud-top TBB threshold was chosen to isolate pixels containing active, deep convection. Spectral analysis of time series constructed from hourly cold cloud (≤ -65°C) pixel counts revealed a powerful diurnal cycle of deep convection significant at the 95% confidence level during JJA and NDJ. Composited hourly statistics of fractional areal cloud cover documented a 0500-0600 LST maximum with a 1500-1900 LST minimum of convection for both seasons. The ratio of maximum to minimum areal cold cloud coverage was greater than 2: I. A significant bi-diurnal cycle was evident in both JJA and January 1995. The bi-diurnal peak was strongest in the near-equatorial region during JJA. No semi-diurnal (spectral) peak occurred during either season. This suggests that semi-diurnal atmospheric tides do not strongly influence convection in the WP. Three objective analysis techniques were developed to analyze the relation of tropical cloud cluster structure to the daily spatial and temporal variation of deep convection. The first technique identified cold cloud intervals, called line clusters, in each image. These line clusters represented a characteristic horizontal dimension for cloud clusters of various sizes. Results showed that the diurnal cycle of convective rainfall with an early morning maximum was disproportionately dominated by the largest ~ 10% of clusters for each time period. While the number of large clusters increased only slightly throughout nocturnal hours, the area of cold cloud associated with these systems expanded dramatically. An algorithm called threshold initiation showed that all scales of organized, intensifying deep convection existed at all times of day and night. In addition, the early morning peak was composed primarily of building convection. Conditional recurrence probabilities of line clusters were computed at 24 and 48 hour intervals. Results for JJA and December 1994 revealed that when early morning convection occurred at any location, the same region contained convection the next morning nearly half the time. Convection was less likely at the 48 hour point. These results do not support diurnal theories based on sea surface heating, afternoon initiation of convection and nocturnal evolution of mesoscale convective systems. Findings indicate that the diurnal cycle of deep convective cloud is driven by the internal variation of large clusters. Clusters that exist into or form during the night, grow spatially larger and more intense. Some results support direct radiative forcing of clouds and large scale clear region radiative destabalization as possible contributors to diurnal convective variability. However, all findings are consistent with the work of Gray and colleagues that emphasizes the role of day-night variations in net tropospheric cooling in clear and longwave cooling in cloudy versus clear regions as an explanation of the observed daily variation of tropical convective rainfall.


September 1997.

Rights Access


Convection (Meteorology) -- Tropics
Clouds -- Diurnal variations -- Observations


Associated Publications