Reply to the comment of Cai et al. on the paper “On the recent warming in the Murray-Darling Basin: Land surface interactions misunderstood” by Lockart et al.


[1] We thank Cai et al. [2010] (hereafter CCBJR) for their interest in our paper and their clarification of the limitations of the Murray Darling Basin (MDB)-averaged sunshine hour duration (SSH) time series. We agree that the averaging procedure used in LKF was not adjusted for latitude-dependency in the climatology. Such limitations of the MDB-averaged time series were explicitly acknowledged in our paper [Lockart et al., 2009, p. 2] with regard to the potential for spuriously improved correlation to Tmax, but not in the context of any potential trend in the residual time series. It is this omission that generated the comment by CCBJR.

[2] Our key focus was to demonstrate that Tmax is better explained by SSH (representing a proxy for solar radiation, the primary driver of air temperature evolution) than seasonally-averaged rainfall (which represents an indirect proxy of nonlinear soil moisture controls on land surface heat fluxes). Given the acknowledged limitations of the MDB SSH data in this context, this was instead achieved through correlation analysis of individual station data that are not subject to the issues of averaging and latitude-correction raised by CCBJR. Indeed, correlation analyses indicate that this result remains robust even after the de-trending of both averaged and individual SSH station data.

[3] Finally, note that even if an accurate SSH timeseries was available across the MDB, the post hoc attribution of trends in the residual series of univariate regression is speculative at best. Consequently, whilst accepting the criticisms of our averaging method as presented by CCBJR, we re-iterate that the “residual-error-trend” methodologies [Nicholls, 2003; Karoly and Braganza, 2005; Cai and Cowan, 2008; Cai et al., 2009] are inadequate for the purpose of climate attribution. Maximum temperatures are driven by multiple variables and complex non−linear land surface interactions not represented in such simplistic methodologies. Moreover, the subsequent claims that increased air temperature causes significantly enhanced evaporation are not supported by the analyses of Lockart et al. [2009], who show that increased antecedent air temperatures play a hydrologically insignificant role in elevating evapotranspiration, especially in comparison to variability in net radiation. Whilst we acknowledge the issues raised by CCBJR, these key conclusions remain robust and pertinent.