Natural Variability, Attribution and Climate Models #5

The Challenge of Understanding "Natural Variability"

How do we know that a change in the climate, for example, rainfall trends in recent decades, is due to human activity like burning fossil fuels? How do we know it’s not natural variability?

This is the question we’ve started to look at in the first four parts of this series.

When it comes to attribution there are 1000s of papers considering the attribution of various trends in climate metrics to human activity (primarily burning fossil fuels).

Here’s what the 6th assessment report (AR6) says about attribution in plain English:

We need to make some assumptions: observed changes are due to a simple addition of forced changes (e.g effects from more CO2 in the atmosphere) and natural variability; and we can work out natural variability

Another way to write the second part:

If we don’t understand natural variability then our assessment could well be wrong.

The text of the report from p. 429 & 430 is in the notes below.

One of their references is Jara Imbers and co-authors from 2014:

A key input in the procedure of fitting this multiple regression model is an estimate of the internal variability of the climate system, against which the statistical significance of anthropogenic and natural signals must be compared. Hence, an accurate depiction of this variability is crucial for the robustness of the results..

..Typically, long GCM control simulations are employed for this purpose. This is such a key step in the process of detecting and attributing climate change that, in fact, for some authors (e.g., Huybers and Curry 2006), the debate surrounding global warming centers on the uncertainties in the structure and magnitude of the internal variability of the climate system.

..The first challenge is to choose an adequate stochastic representation for the internal variability. The difficulties finding the appropriate stochastic model are due to the uncertainties in characterizing internal variability from the observational record, which, as discussed before, is contaminated by the external forcings and is too short relative to the long time scales potentially relevant to the current climate variability

In simple terms, we don’t have high quality global data going back a long time before we started burning fossil fuels, so we have a difficult challenge.

They note that various efforts to estimate internal variability looked at GCM simulations over long time periods. We’ll probably have a look at some of those in future articles.

We remark that our goal is to explore the sensitivity of the detection and attribution statistics to the representation of internal variability. Therefore, the main assumptions of detection and attribution of climate change, namely, that the forced responses can be linearly superimposed on internal variability and that there are no interactions between forced and unforced variability, are assumed to be valid.

Science often proceeds like this. Make some assumptions and see what results you get.

The IPCC very likely statement that anthropogenic emissions are affecting the climate system is based on the statistical detection and attribution methodology, which in turn is strongly dependent on the characterization of internal climate variability as simulated by GCMs. The understanding of the internal climate variability has been identified as one of the hardest geophysical problems of the twenty-first century (e.g., Ghil 2001). One of the barriers we face to advance our understanding is the lack of long enough reliable observational records.

[Emphasis added]

Their approach is to try two simple models of internal variability (AR1, or first order autoregressive and FD, or fractional differencing). We’ll have a look at details in the next article as it gets a little technical and some readers might not be so interested.

---

Notes

Some extracts from the Methods section of Chapter 3 of AR6 on Attribution, p. 429 & 430. Emphasis added:

Regression-based methods, also known as fingerprinting methods, have been widely used for detection of climate change and attribution of the change to different external drivers. Initially, these methods were applied to detect changes in global surface temperature, and were then extended to other climate variables at different time and spatial scales.

These approaches are based on multivariate linear regression and assume that the observed change consists of a linear combination of externally forced signals plus internal variability, which generally holds for large-scale variables. The regressors are the expected space–time response patterns to different climate forcings (fingerprints), and the residuals represent internal variability. Fingerprints are usually estimated from climate model simulations following spatial and temporal averaging.

A regression coefficient which is significantly greater than zero implies that a detectable change is identified in the observations..

..A signal can be spuriously detected due to too-small noise, and hence simulated internal variability needs to be evaluated with care. Model-simulated variability is typically checked through comparing modelled variance from unforced simulations with the observed residual variance using a standard residual consistency test, or an improved one. Imbers et al. (2014) tested the sensitivity of detection and attribution results to different representations of internal variability associated with short-memory and long-memory processes. Their results supported the robustness of previous detection and attribution statements for the global mean temperature change but they also recommended the use of a wider variety of robustness tests.

---

References

Eyring, V., N.P. Gillett, K.M. Achuta Rao, R. Barimalala, M. Barreiro Parrillo, N. Bellouin, C. Cassou, P.J. Durack, Y. Kosaka, S. McGregor, S. Min, O. Morgenstern, and Y. Sun, 2021: Human Influence on the Climate System. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change

Sensitivity of Climate Change Detection and Attribution to the Characterization of Internal Climate Variability, Jara Imbers et al, Journal of Climate (2014)