Observed variations of tropical convective available potential energy

Abstract

Data from selected tropical radiosonde stations (located between 25°S and 25°N) are used to compute multidecadal trends in convective available potential energy (CAPE). Positive trends slightly outnumber negative trends, with the greatest concentration of positive trends occurring in the western Pacific Ocean and the Caribbean Sea. Analysis shows that positive and negative CAPE trends are primarily driven by same‐signed trends in low‐level moisture, while lapse rate trends play a secondary role and may either increase or decrease the magnitude of the CAPE trends. Monthly CAPE anomalies are generally positively correlated with sounding‐derived precipitable water (PW) and lapse rate as well as with Special Sensor Microwave Imager (SSMI)‐derived PW for retrievals located over the sounding location. Sounding‐derived PW estimates for island and coastal locations were averaged and compared to tropical SSMI PW estimates, which are available only over the oceans. The two time series are poorly correlated, suggesting that the limited number of sounding stations considered herein does not adequately capture tropical‐wide variability. The relationship of CAPE changes to PW and lapse rate changes is examined on a variety of timescales for evidence of quasi‐equilibrium (QE) behavior, in which convection acts to maintain an approximately moist adiabatic temperature profile. Strict QE, in which convection maintains an exactly moist adiabatic temperature profile, is assumed in the convective parameterizations of many general circulation models. PW and lapse rate anomalies are negatively correlated on multidecadal timescales and on monthly timescales in heavily raining regions, in support of QE theory. However, no such negative correlation is seen over the course of the seasonal cycle or for monthly timescales in low‐rain areas. CAPE anomalies were compared to nearest‐grid‐point monthly precipitation anomalies derived from surface‐based rain gauges. Correlations are small and centered at about zero. Mean tropical rainfall estimates, based on a combination of satellite and surface gauge data, are uncorrelated with the mean tropical SSMI PW estimates.