Since there has been so much interest in the topic of the âbutterfly effectâ?, a weblog on climate prediction with respect to its sensitivity to initial conditions is warranted.
The answer to the question posed on todayâs weblog, of course, is YES.
With respect to weather prediction, the importance of initial conditions is universally accepted. As just one example, we can refer to their importance in hurricane track forecasts, where the size of the initial hurricane vortex, its initial motion, and its intensity each matter in terms of its subsequent motion. These are large enough perturbations to upscale (unlike a butterflyâs flapping wings!). Weather also exhibits chaotic behavior such as when slight differences in large-scale flow patterns can determine whether baroclinic cyclogenesis occurs or not.
For climate prediction, however, the existence of two definitions of âclimateâ? complicates the discussion. The term âclimateâ? has been used to mean long-term weather statistics, but also the coupled atmosphere, hydrosphere, lithosphere, and biosphere (see the weblog posting for July 29th entitled âWhat is climate changeâ?).
The use of long-term weather statistics to mean âclimateâ?, however, is an atmospheric-centric view. Weather statistics, as the definition for âclimateâ? has traditionally been limited to physical variables such as temperature and precipitation, but not even to atmospheric chemical composition (see the AMS definition of âclimateâ?).
The distinction is important. With the atmospheric-centric view, the ocean, land, and continental ice are often treated as boundaries that are prescribed. This places a constraint on the âclimateâ? prediction since the interactions with these surfaces are reduced or even ignored. With the more inclusive definition of climate, there are interfacial, nonlinear fluxes between the atmosphere, oceans, land, and continental ice. That is there are no true boundaries.
This subject is discussed in my essay – Pielke, R.A., 1998: Climate prediction as an initial value problem. Bull. Amer. Meteor. Soc., 79, 2743-2746. In that essay, I concluded that âas a result of the variety of significant ocean-atmosphere-land surface interactions, model-based forecasts of future climate should be viewed as sensitivity analyses rather than as reliable predictions.â?
A specific example on a seasonal time scale of the sensitivity of a climate prediction to the initial soil moisture content (i.e., a non-atmospheric variable) as its affects growing-season weather is presented in Pielke Sr., R.A., G.E. Liston, J.L. Eastman, L. Lu, and M. Coughenour, 1999: Seasonal weather prediction as an initial value problem. J. Geophys. Res., 104, 19463-19479. In this paper, we concluded
ââ¦that the seasonal evolution of weather is dependent on the initial soil moisture and landscape specification. Coupling this model to a land-surface model, the soil distribution and landscape are shown to cause a significant nonlinear interaction between the vegetation growth and precipitation. These results demonstrate that seasonal weather prediction is an initial value problem. Moreover, on seasonal and longer term timescales the surface characteristics such as soil moisture, leaf area index, and landcover type must be treated as dynamically evolving dependent variables, instead of prescribed variables.â?
See also Lu, L., R.A. Pielke, G.E. Liston, W.J. Parton, D. Ojima, and M. Hartman, 2001: Implementation of a two-way interactive atmospheric and ecological model and its application to the central United States. J. Climate, 14, 900-919.
What is missing from the IPCC and US National Assessments is the recognition that climate is not atmospheric-centric (or even physical ocean-atmosphere centric), but as involving significantly the other components of the climate system as both forcings and feedbacks. Ocean plankton distributions, fresh water river and sediment discharge into the oceans, and land-cover/land-use are just a few examples of climate variables that need to be initialized in the non-atmospheric components and involve interfacial, nonlinear fluxes, but whose importance has been ignored or understated.
When we learn of âprojectionsâ?, âforecastsâ?, and âpredictionsâ? of climate decades into the future (e.g., see âNo Winter by 2105? New Study Offers Grim Forecast for the U.S. ), we should first assess whether the suite of model simulations that were used to create the envelope of predicted future climate has included the spectrum of the initial climate conditions which must include the non-atmospheric components. If they have not (which is the case for all existing modeling studies of this type), the value of such studies are as sensitivity experiments, and should not be presented, as the National Geographic has done, as forecasts.