Projections of future distribution and density of parasites and vectors are highly dependent on the accuracy of climate models. Over the last year, and particularly in recent months, it has been debated whether the projected future changes in climate might be wrong; -not wrong in the sense there will be no global warming, but how warming will evolve over time. The Earth is presently cooler than projected by most of the global climate models, and this has been picked up by news outlets like Daily Mail and The Economist. Figure 1 shows projected and observed warming for the last 30 and 15 years, and it is evident the observed warming is less than most of the models projected. Climate scientists have responded by pointing at the uncertainties in the role of aerosols, small particles reflecting sunlight, and interacting with clouds thus masking the radiative forcing caused by greenhouse constitutes. For example the lack of warming between the 1940s and 1970s has been explained by a substantial increase in the amount of aerosol in the atmosphere. Others have argued it does not make sense to make conclusions based on a 15 year period with no warming, and that at least 30 years is needed to robustly detect climate change. Increased heat content of the ocean has been suggested as yet another explanation, and so has changes in the net amount of incoming energy to the climate system.

Now, a new theory by Wyatt and Curry is claiming that the current pause in global warming could extend into the 2030s, and that the causal mechanism is not present in the current generation of climate models. According to reportingclimatescience.com the authors behind the new theory state that “Current climate models are overly damped and deterministic, focusing on the impacts of external forcing rather than simulating the natural internal variability associated with nonlinear interactions of the coupled atmosphere-ocean system”. It is not the first time this has been claimed.

In an article in Geophysical Research Letters, Emerging selection bias in large-scale climate change simulations, Kyle L. Swanson described how experiments might go wrong as we seek a desired solution. He hypothesised that a common wish to reproduce the recent warming in the Arctic, has led to less diversity among models with convergence towards some common solution. He considers that the current generation of model simulations are statistically inconsistent with the observed shifts in both the mean surface air temperature and the frequency of extreme monthly mean temperature events due to climate warming, despite a marked reduction in the spread of projected values that itself suggests convergence towards some common solution. This convergence indicates the possibility of a selection bias based upon warming rate. He illustrates this with an example of what happened after Robert A. Millikan’s original measurement of the charge of the electron: “Millikan’s original measurement was slightly erroneous due to the use of an incorrect value of the viscosity of air. In the decades following Millikan’s work and his subsequent Nobel Prize, other investigators empirically measured the electron charge. When they got a number that was too high above Millikan’s, they thought something must be wrong–and they would look for and find a reason why something might be wrong. When they got a number close to Millikan’s value they didn’t look so hard. And so they eliminated the numbers that were too far off, and did other things like that.”

The causes of the lack of warming the past 15 years (or so), and why global climate models do not see this pause, is interesting in itself. As interesting, is the debate about how we ended up in this situation; possibly a result of selection bias. This debate is relevant, independent of the branch of science or health research, and if a selection bias is present it has implications for projections of the spread of vector borne diseases.

Understanding how changes in local and global temperature may impact vectors and the diseases they carry is a field of intense interest with many important new models proposed in the past few years for malaria, dengue, and leishmaniasis, which are underpinned by existing models of global warming. Firstly, if climate models are suffering from selection bias, any studies using climate models to make projections about the future spread of vector borne diseases will inherit this bias. These projections will, due to the damped natural variability in climate models (they agree more than they should), underestimate the uncertainties relating climate change to vector borne diseases. Secondly, if it is correct that we will have a pause in global warming until the 2030s, as Wyatt and Curry suggest, projections about vector borne diseases during the next 20 years will also be wrong. An important implication of the study is that projections should be made for longer periods than 20 years as natural variability can be dominant at these time scales, although not visible in the current generation of climate models.