Water vapor and cloud feedbacks determine the sensitivity of the climate system to an external perturbation, affect the mean climate, and control the amplitude of natural variability. Thus validating and understanding water vapor and cloud feedbacks in climate models are of critical importance. Here we compare the water vapor and cloud feedbacks from the two most recent versions of the atmospheric component of the NCAR CCSM--the NCAR CAM1 and CAM2--with those from observations over the equatorial Pacific cold-tongue.

The results show that in both the CAM1 and CAM2, the water vapor feedback is considerably larger than that from ERBE observations. Even larger errors are found in the cloud feedbacks, in particular, the feedback from the short-wave forcing of clouds. While observations show a strong negative feedback from the short-wave forcing of clouds (-7.8 Wm^-2), the feedback from the short-wave forcing of clouds in CAM2 is positive (+3.4 Wm^-2K^-1). The sign of the feedback from the short-wave forcing in CAM1 is correct, but the strength is less than half of the observed value (-3.0Wm^-2K^-1). Consequently, the net atmospheric feedback in both CAM1 and CAM2 over the cold-tongue region are strongly positive (+5.1 Wm^-2K^-1 in CAM1 and +6.7Wm^-2K^-1 in CAM2) while in the real atmosphere it is strongly negative(-6.4Wm^-2K^-1).

Further examination of the humidity fields over the cold-tongue reveals an excessive response of the upper level humidity in the models to a surface warming. The upper cloud cover response in the models to a surface warming is also substantially larger than that from the ISCCP observations. The optical depth of clouds in the model also appears to have large errors.

The consequences of the errors in the cloud and water vapor feedbacks upon the simulations of the coupled climate is explored using a linear feedback model as well as the NCAR Pacific Basin Model coupled with an empirical atmosphere. The numerical results suggest that the discrepancy in the water vapor and cloud feedbacks may have contributed significantly to the cold bias in the equatorial Pacific in the NCAR CCSM.