Tropical radiosonde observations have been recorded as far back as the late 1940's at several locations. This relatively long record of in situ observations of temperature, moisture, and winds comprise a unique dataset that may be used for short-term climatological studies of tropospheric moisture. However, as this dataset was never designed as a climate-monitoring tool, there are many uncertainties that arise from both documented and undocumented changes in sensor type and sounding location.

Time series of tropical CAPE, which is highly sensitive to low-level moisture and plays an important role in the development of convection, reveal positive trends at nearly all tropical sounding stations with records extending back to at least 1978. The impact of changes in sensor type on these trends, however, is not clear. Changes in sensor type have been shown to impact both temperature and moisture profiles, both of which are used in computing CAPE. Furthermore, assessments of global temperature change and rainfall trend predictions are used in policy-making decisions world-wide, so it is imperative to determine the impacts of variable sensor types on the computed geophysical trend.

In this study, we attempt to ascertain the reliability of sounding-based CAPE trends by comparing times series of sounding-based precipitable water (to which CAPE is highly sensitive) and SSM/I-derived precipitable water estimates during the period 1988-1997. The comparison is initially limited to those stations with homogeneous sensor types throughout the period. Trends in sounding-derived lapse rate and GPCP precipitation are also examined. Initial results indicate that the trends in CAPE, lapse rate, SSM/I precipitable water, and GPCP precipitation are generally of the same sign for the stations studied, lending credibility to the sounding data when sensor type is held constant. In regions where there exists such consistency for a grouping of stations, we will repeat the analysis for an extended period and for stations without homogeneous sensor types throughout the period. If similar consistencies in trend sign remain, this may lead to greater confidence in the qualitative assessment of CAPE trends as determined by the sounding data.

Focused studies of biases in specific sonde types also exist which may assist in the evaluation. For example, 4-decade CAPE trends for several stations located in the West Pacific are positive, despite a documented underestimate of CAPE during the last several years of the record which resulted from a change in sensor type. For these stations, CAPE trends are likely greater than those based on the sounding record.