|
|
|
PART 2 : SECTION 1
Making judiciously chosen assumptions, the above discussion showed how the seasonal average winds over the tropical oceans can be inferred from the seasonal average SST. In reality, the atmosphere is much more complicated. In particular, we have not considered that the atmosphere is constantly changing from hour to hour and day to day. Is our simple model of the climate correct in its prediction of a strong SST impact in the tropics? We saw that the model at least to some degree appears to be consistent with the observed wind fields associated with two key patterns of SST variation in the tropical Pacific and tropical Atlantic (Fig. 1.6). A further way to explore this question is to use the most comprehensive models of the atmosphere that we have available known as the General Circulation Models (GCMs). These models do not seek a seasonal average mean solution for the atmosphere. Instead they simulate the weather every few minutes through a season. We then average the weather that they simulate in order to see the seasonal average conditions.
GCMs work by representing the atmosphere at every point in a three-dimensional grid. A typical grid might have a horizontal spacing of about 300km and about 20 vertical levels distributed through the key first 10km of the atmosphere, and a few more above that level. The models must be given an initial starting point for the atmosphere, that is, the temperature, wind speed etc at each grid point. For climate studies, the starting point is arbitrary. It can just be an example of the global weather pattern on any given day. We are going to be interested in the impact of a forcing (here the SST) on the average weather simulated by the model during a season. Equations in the model try to represent all the key processes that determine the evolution of the atmosphere (Fig. 1.1). For example, the pressure forces that are impacting the wind at a given grid-point. Based on the current prevailing situation, the model predicts the conditions expected up to a specified short time ahead. This is known as the time step of the model, and it may be around 10 minutes for a GCM. Once the prediction for 10 minutes ahead is made, the new prevailing situation is analysed e.g. the new balance of forces impacting the wind at each grid-point. Based on the new situation, a prediction is again made for 10 minutes ahead (see Fig. 2.1 for illustration of an example of the model stepping forward in time). This process is repeated until the model has stepped forward in time as far as needed. If we are making an experiment to study a 3-month season, then the model steps forward through 90 days. Since each day has 144 steps of 10 minutes, the total number of steps is 144x90 = 12,960.
|
|
|
|
|
Fig 1.1. Schematic of the Global Climate System
|
|
|
|
Fig 1.6a-c. Maps of Sea-Surface Temperature and Anomalies
|
|
|
|
Fig 1.6d-f. Maps of Sea-Surface Temperature and Anomalies
|
|
|
|
Fig 2.1. Schematic illustrating how a Numerical Climate Model steps forward in time
|
|