Cooperative Institute for Research in Environmental Sciences

A winning proposal for the Innovative Research Program, 2007:

Should We Care about a Variable and Noisy Sun?

Investigator: Tom Chase

I. Objective

A. Background

Figure 1 shows the observed time series of the solar constant. It is apparent that there is a great deal of natural variability and structure in the solar energy reaching the top of earth's atmosphere including a quasi-sinusoidal sunspot cycle at approximately 11 years with higher frequency variability superimposed. This higher frequency variability itself has structure with higher amplitudes at the crest of the sunspot cycle and less in the troughs. There is also a great deal of random variability (noise) imposed on the more ordered fluctuations which is evident in the observation that no single fluctuation is exactly alike. The maximum amplitude of these variations is approximately 6W/m² which can be compared to the current radiative forcing due to anthropogenic greenhouse gasses of approximately 2.3 W/m² .

Fig.1

Interestingly, even the most complex climate models still use a constant measure of the amount of solar radiation reaching the top of the atmosphere which is an average of the above time series often put at about 1366 W/m². This is for two reasons: the first is that it has been assumed that the variability and noise in the solar constant would average out over time, it is not until recently that the field of non-linear geophysics has developed an appreciation of the role of stochastic processes, the second was that climate models were very expensive to integrate and the large ensemble simulations necessary to characterize the effect of this variability were too costly to attempt.

Recently, there has been evidence from theory and from examples from simple non-linear models that variable and stochastic processes have a strong influence on the climate statistics and need to be more completely included in model simulations (Penland, 2006). For example, the addition of simple white noise to a model of ENSO has been shown to greatly amplify model variability (Wunsch, 1999)

My objective here is to test the hypothesis that variability and noise in the solar constant will significantly affect climate model statistics both in the average state and in the variance of circulation patterns such as the North Atlantic Oscillation (NAO).

B. Proposed methodology

I propose to use the Planet Simulator, a well-known, low resolution, 3-dimensional General circulation Model (GCM) developed at the University of Hamburg to examine noise and variability in its solar constant on the simulated atmospheric circulation. It is computationally inexpensive but includes all key elements of a full General Circulation Model including all feedback processes such as ice albedo and water vapor feedbacks. It includes enough complexity to reasonably reproduce the observed atmospheric circulation and precipitation patterns and is simple enough to be useful for examining large ensembles of simulations which is vital for bracketing the range of model variability. For example, a 125 year simulation can be run overnight on our local linux machine.

I will examine this question in three steps. First, I have already added a white noise generator (equal power at all frequencies) to the solar constant and present preliminary results below. I will then introduce an idealized red noise generator (higher power at lower frequencies – similar to the sun) for comparison. Finally, I will generate a long synthetic time series with the full observed structure of the solar constant to compare with the first two sets of simulations.

C. Preliminary results

In order to test this concept I ran a 100 year control experiment with a solar constant of 1366 W/m², a second experiment where I added white noise with amplitude of approximately 6 W/m² added to the solar constant (approximately the amplitude observed), and a third experiment where the amplitude of the noise was approximately 4 W/m² (approximately the amplitude of the largest observed high frequency variability). Figure 2 shows results from these simulations in the northern hemisphere sea level pressure field anomalies for the 6 W/m² NOISE simulation. Of great interest are the 95% significant SLP anomalous dipole in the North Pole and a negative anomaly in Europe: apparently the noise has affected the model simulation of the North Atlantic Oscillation which is of great significance to Northern Hemisphere climate and climate change. Other simulations with lower levels of NOISE had similar SLP anomalies related to the NAO indicating a robust response.

Figure 2

II. Importance

While this is an atmospheric model experiment it is also fundamentally an experiment in non-linear geophysics which spans and will affect all the earth sciences. Theoretically noise should change the characteristics of non-linear model simulations in all disciplines. Our initial results indicate that adding idealized noise can affect major modes of atmospheric circulation significantly. However, it has been previously untested in a systematic way allowing an unchanging solar constant to continue force present climate change simulations.