Reply to Mölg et al. (2006)

In the previous post, the comment of Mölg et al. (2006) gave a number of clear critiques to the original paper of Taylor et al. (2006). As part of this peer review process, the response of Taylor et al. to this criticism was published in the same journal issue. As with before, the reply here will hopefully be stated in a concise manner.

The authors begin by accepting the uncertainty surrounding not only the influence of air temperature on glacial recession in the tropics, but the influence of atmospheric humidity as well. Perhaps most importantly, with respect to the comments of Mölg et al., the authors accept the uncertainty over the relationship between surface and higher elevation temperature trends in the tropical free troposphere. As well as all of this, they agree with the contention that surface energy balance (SEB) models provide the ideal platform to show climate and glacier relationships. However, the authors rebuke the criticism of Mölg et al. by stating that the lack of SEB climatic parameters was made clear in the original paper. Thus, Taylor et al. accept that a definitive understanding of climate-glacier interactions in the Rwenzori Mountains isn’t presently possible. However, the authors do believe that the evidence for air temperature being the primary influence on glacier dynamics is much greater than the available evidence for decreasing humidity (as theorised by Mölg et al).

The main criticism of the original paper was the validity of the observed air temperature trends at meteorological stations being reflective of trends higher up in the mid-troposphere. Whilst Mölg et al. believed that the original paper had simply ignored any inconsistencies, the authors contend that in fact there is mixed consensus on this issue and that the studies cited by Mölg et al. are unfairly selective. Indeed, whilst the critique cited mid-troposphere trends through the 20^th century as showing no warming (as opposed to the lower elevation measurements), the choice of start and end points alone can greatly alter the observed trend. As Taylor et al. state:

“...in the paper by Gaffen et al.  [2000] using MSU data in which at 500 hPa a cooling trend is detected between 1979 and 1997 but an overall warming trend occurs between 1960 and 1997.”

Further to this, more recent studies are mentioned which show the validity of the original paper:

“Nevertheless, recent studies that employ diurnal corrections to MSU observations between 1979 and 2003 [Mears and Wentz , 2005] and homogenized radiosonde data sets (HadAT2) between 1958 and 2002 [Thorne et al. , 2005], show that the middle troposphere warmed at a similar or slightly greater rate to the surface in the tropics [Fu and Johanson , 2005; Santer et al. , 2005]...”

The authors also criticise the use of reanalysed data by Mölg et al. to show a difference between surface air temperature and data in the mid-troposphere. According to Taylor et al. the use of NCEP data is controversial, given the widespread consensus that reanalysis data is unsuitable for trend analysis in climate change studies due to time-dependent errors. The authors use an alternative and perhaps less controversial dataset, the HadAT2 radiosonde data, to show that upper air temperature records closest to the Rwenzori Mountains show consistent warming from 1958 to 2005. As well as this, these upper air trends match the surface air temperature trends over the same period from another homogenised dataset (CRU TS 2.0).

Following this, the authors seek to discredit the observed trends of decreasing specific humidity by Mölg et al. The same NCEP reanalysis data was used to identify this trend in the mid-troposphere through the 20^th century. Apart from the already identified bias in the NCEP data, the reliability of the humidity readings is even more questionable according the Taylor et al. This is mostly due to humidity being a statistically derived parameter, and thus not suitable for trend analyses. The humidity data cited by Mölg et al. was in fact uncorrected, which means that any systemic dry biases remained unresolved. As well as this, a 20^th century decline in humidity is unsupported by the CRU TS 2.0 precipitation / water vapour datasets. Mölg et al. also argues that the observed East African glacial retreat was initiated by a vast reduction in moisture at the end of the 19^th century. This change in moisture is based on historical lake levels in East Africa. Taylor et al. argues that this drop in lake levels was actually the abrupt finish to a brief decade-long high lake stand, and nothing more. Indeed, a good comparison to this event was a 2.3m rise in the level of Lake Victoria between 1961 and 1964 which did briefly increase the regional humidity. However, this increased humidity gave a very small, year-long advance to the Rwenzori Mountain glaciers and does not support the humidity hypothesis of Mölg et al. which suggests continual glacial retreat in the latter half of the 20^th century. A final statement by Taylor et al. identifies another clear flaw in the humidity hypothesis:

“Even ignoring concerns regarding this evidence, the argument that these climate events are responsible for the expected demise of small, fast-responding glaciers that have persisted for at least 5000 years is improbable.”

In summary, the authors accept that both air temperature and air humidity are viable causes of glacier retreat in the tropics and are likely related in some way. There is also wide agreement that the meteorological data surrounding the Rwenzori Mountains is far from ideal. However, in the view of the original authors, there is more robust data supporting changes in air temperature as the primary cause of glacial retreat.

By identifying that glaciers in the tropics may well behave no differently to alpine glaciers elsewhere with regards to rising air temperature, it is important to further research this. As has been mentioned previously, these tropical glaciers often have great cultural significance as well as being key for the success of agriculture and survival in parts of East Africa. A more definitive account of how the Rwenzori Mountain glaciers responded to changes in atmospheric conditions in the past will help us to greater predict how long they may last. However, whilst further research may reveal that some of the glaciers could perhaps last longer than 20 years stated by Taylor et al., there remains little doubt that their days are numbered.