The climate is highly sensitive to water vapor changes in the upper troposphere where there is not such a high concentration of water vapor, and is relatively insensitive to changes in tropospheric water vapor in the boundary layer where there are high concentrations of water vapor, which can be seen in .
Thus, it is possible to have an increase in water vapor in the boundary layer, but still get a negative water vapor feedback, as shows.
A bunch of papers by Pielke, Douglass, Lindzen, Christy, Spencer etc. also show net negative feedback in the climate system.
For example, on average, the climate system radiates much more heat due to a radiative forcing than do models, suggesting a net-negative feedback, possibly confirming Lindzen's iris hypothesis:
Also see for such a diagnosis of negative climate feedback.
We explore the daily evolution of tropical intraseasonal
oscillations in satellite-observed tropospheric temperature,
precipitation, radiative fluxes, and cloud properties. The
warm/rainy phase of a composited average of fifteen
oscillations is accompanied by a net reduction in radiative
input into the ocean-atmosphere system, with longwave
heating anomalies transitioning to longwave cooling during
the rainy phase. The increase in longwave cooling is traced
to decreasing coverage by ice clouds, potentially supporting
Lindzen’s ‘‘infrared iris’’ hypothesis of climate stabilization.
These observations should be considered in the testing of
cloud parameterizations in climate models, which remain
sources of substantial uncertainty in global warming
prediction.