Convergence and new cell development

The reason why the thunderstorm systems evolve differently in the experiments can be traced back to the problem of what triggers convection. The initiation of convection by density currents, i.e., by gust fronts and sea breezes, is primarily a function of horizontal convergence.

In the experiments, the maximum low-level convergence4

is found at different positions at the gust front edge shortly after the gust front occurs. However, twelve minutes after the gust front forms, the convergence is largest at the NW edge of the cold pool in all experi- ments, except in EXP6 where large convergence in the NW is apparent already aftertgf+ 8

min (see Fig. 5.7c). This difference results from the faster overall evolution of the initial updraught and gust front in EXP6 compared to those in the other cases, which is caused by the larger temperature excess of the warm bubble that triggered convection. As the first new cells form about 20 min to 24 min after the gust front occurs, the mean convergence

∂u/∂x+∂v/∂y is calculated between t = tgf + 12 min and t = tgf + 20 min (see Table

5.1). This quantity should give information about how strong the low-level convergence

4

If the word “convergence” is used in the following, low-level convergence (z= 0.19 km) is meant – if not stated otherwise.

needs to be so that new convection can be triggered at the pre-existing cold pool edge. The convergence is largest in the cases where a large multicell complex develops (EXP5, basic experiment, EXP4), followed by the experiments where small multicell complexes form (EXP3, EXP2), while the experiments in which no new cell development occurs show the smallest surface convergence (EXP1, EXP6). Thus, the amount of convergence at the gust front needs to be larger than about 6 _×10−3

s−1

, on average, so that a new cell can be triggered. Thereby it is of great importance that strong convergence is present long enough time. If the convergence is large after the gust front appears, but decreases significantly some minutes later (see EXP1 and EXP6 in Fig. 5.7c), the upward motion declines – even before new cell development commenced.

The finding that strong convergence needs to be present over a sufficient time span is confirmed by observations. By studying the interaction of a SE-moving Darwin sea breeze front and a NW-moving gust front from a pre-existing thunderstorm, Keenan and Carbone (1992) noticed that a major storm did not develop until 25 minutes after the collision of the fronts. Then the new updraught formed 10 km downstream of the point of collision. The time lag between the collision of the fronts and the detection of the new cell was found also by Wilson and Schreiber (1986) who investigated mid-latitude convection at boundary-layer convergence lines. In their study, storms were initiated, on average, 18 min after a single front passed.

The reason for the different amounts of convergence at the cold pool edge in the ex- periments is, of course, related to the differences in the model settings. In three of the model runs with two sea breezes (basic experiment, EXP4, EXP5), the low-level wind in the cooler sea breeze environment creates a region of enhanced convergence when opposing the spreading gust front. Similar conditions are created by only one sea breeze (EXP2, EXP3), although, at later stages, the environment is less favourable for convection to form and the new cell development stops. In EXP1 (no sea breeze), the environmental winds of 14 November 2005 are not sufficient to provide an opposing flow to the spreading cold pool, which results in the demise of the initial cell and its gust front without triggering new updraughts. The question is why we do not see strong convergence and thus, new cell development in EXP6, even though there are two sea breezes creating the same en- vironmental conditions as in the basic experiment. The answer lies in the characteristics of the gust front, which opposes the environmental flow. The strength of the cold pool is determined by the strength of the downdraught. A measure for the latter is calculated by averaging wcb

min between the time of gust front occurrence, tgf, and tgf + 20 min. For the

basic experiment, the minimum vertical velocity wcb

min is −10.19 m s−

1

, while the down- draught in EXP6 is, with wcb

min =−7.74 m s−

1

, about 24% weaker. This difference in the downdraught strength is due to the larger temperature excess of the warm bubble in EXP6, which leads to a faster growth and decay of the cell, resulting in a weaker downdraught and gust front than in the basic experiment.

I examine now in more detail all the new cells listed in Table 5.1. In all but one experiment, the first new cell develops at, slightly behind, or ahead of the NW edge of the cold pool, at the location where the surface convergence is largest. However, in EXP4 (Wsb-experiment) the first new updraught forms at the western edge of the gust front, where the convergence is smaller than in the NW. This apparent inconsistency will be discussed in section 5.5.5, where the buoyancy and pressure gradient forces are studied.

Of the 17 subsequent cells studied, 76% of them form southwest of the previous up- draught. An example for this development is shown in Figs. 5.5 and 5.6. In Fig. 5.5a, only

5.5 Comparison of the experiments and discussion 89

the first cell which formed at the NW edge of the cold pool is apparent. Figures 5.5c and 5.6a show the second to fourth new cell which appear in the SW or south of the previous cell. The reason for this organisation is that the sea breezes, together with the low-and mid-levels winds, form an environment where the convergence is large to the SW of the previous cell. Thus, the line becomes oriented perpendicular to the low-level shear (see section 5.3.1). In two of the 17 cases, the convergence is not strongest at the location where the cells form. On the other hand, there are cases where, even though there is significant convergence, for example, at the northern edge of the cold pool in Fig. 5.5b, the new cell development is suppressed. The comparison of the distances between neighboring cells shows that a new updraught forms only at a distance larger than 5 km to the pre-existing ones. This organisation is because subsidence from previous cells suppresses convection in their surrounding (see section 5.5.5).

In summary, the regions where new cell development is possible can be predicted well by examining the low-level horizontal convergence in the model. However, in three cases (EXP4, first cell; basic experiment, third cell; EXP2, third cell), deep convection does not occur at locations where the convergence is at a maximum. Further, there are regions of weak ascent, such as at the northern and NNW edge of the gust front, but where convection is suppressed. Thus, even though the initiation of convection by density currents is primarily a function of horizontal convergence, there are several other factors that play a role. These factors were studied theoretically and numerically in the past decades, but often only in two dimensions. The theories and factors are discussed in the following sections, on the basis of all experiments (EXP1 to 7, basic experiment). It should be noted that the modelled multicell complexes and density currents show significant three- dimensional features, which could potentially limit the applicability of two-dimensional theories to interpreting overall system characteristics. Furthermore, the environmental shear profile of the 14 November 2005 case is neither idealised, nor restricted to low-levels. This makes it difficult to apply the results of existing studies to the observations of the Northeaster, and to the model results obtained here.