Impact of dataset choice on calculations of the short‐term cloud feedback

Abstract

AbstractDessler [2010, hereafter D10] estimated the magnitude of the cloud feedback in response to short‐term climate variations and concluded that it was likely positive, with an average magnitude of +0.50 ± 0.75 W/m2/K. This paper investigates the sensitivity of D10's results to the choice of clear‐sky top‐of‐atmosphere flux (ΔRclear‐sky), surface temperature (ΔTs), and reanalysis data sets. Most of the alternative ΔRclear‐sky data sets produce cloud feedbacks that are close to D10, differing by 0.2–0.3 W/m2/K. An exception is the Terra SSF1deg ΔRclear‐sky product, which produces an overall negative cloud feedback. However, a critical examination of those data leads us to conclude that that result is due to problems in the Terra ΔRclear‐sky arising from issues with cloud clearing prior to July 2001. Eliminating the problematic early portion yields a cloud feedback in good agreement with D10. We also present an alternative calculation of the cloud feedback that does not require an estimate of ΔRclear‐sky, and this calculation also produces a positive cloud feedback in agreement with D10. The various ΔTs data sets produce cloud feedbacks that differ by as much as 0.8 W/m2/K. The choice of reanalysis, used as a source of ΔRclear‐sky or as adjustments for the cloud radiative forcing, has a small impact on the inferred cloud feedback. Overall, these results confirm the robustness of D10's estimate of a likely positive feedback.