Yeah... Let's dive into the topic of latent heat release, in the context of clouds!
To build on Larry's one point, latent heat... Some wonder why the lift gets stronger near cloudbase (assuming the cloud is growing and not dying). The reason is, like Larry noted, is due to the mechanism of latent heat release of condensation. Latent is just a meteo term for hidden. As air (dry, not moist/saturated) rises (thermal, wave, fart, etc), it cools adiabatically at the dry rate of 9.8 c/km (5.4 f/1000') until the air condenses, and thus saturates, at the point where the air temperature meets the dewpoint temperature. Simply stated, the meetup altitude is where cloudbase exists (you can find this altitude on a Skew T plot). Once the air condenses, it releases its latent ("hidden") heat, which adds additional heat to the already rising thermal column, in an environment that is cold (at altitude and under the shade of a cloud). This latent heat that is released rises in this steeply lapsed environment causing even stronger, and more widespread lift. Not sure exactly about this part, but my theory is that well below cloudbase you have the lift rate of the thermal but when you get near cloudbase, where the air is starting to saturate (wispies are a good sign of this), the latent heat is released causing additional lift, at around the rate associated with the moist adiabatic lapse rate (which is variable due cloud microphysics). So basically, why lift can get so strong and thus Hoover-like just below a cloud, is that you have the lift associated with the newly released latent heat ADDED to the lift of the thermal. It's a summation game. Hypothetically, let's say you're having fun in a thermal that is rising at 800 fpm, and as you get closer to cloudbase, you encounter additional "new" lift of 1000 fpm due to latent heat release. You can see how this addition of the lift from latent heat can suck you up. Spiraling down in 800 fpm is no prob, but spiraling down in 1800 fpm is much more challenging. Further, the lift due to heat release is more widespread than the isolated thermal column width. So, even if you escape the band of 1800 fpm lift (thermal lift rate summed with the lift rate associated with the latent heat release), then you still have to deal with the larger-diameter under-cloud latent heat release lift (1000 fpm). This might be just enough to suck you in as you're trying to make a run for the edge.
Rich? Correct me here where I'm off.
, ) jake