09 February 1996

Summary of work done during the stay of Dr. C. Doswell

at the Universitat de les Illes Balears (SAB95-0136)


Diagnóstico y simulación numérica de conveccion severa

en el mediterraneo occidental

codigo UNESCO: 2509

Los principales objetivos de la investigatión son:

1. Diagnóstico numérico de ambos casos buscando elementos discriminadores.

2. Implementación de model numérico mescalar MM4 de Pennslvania State Univ-

NCAR en el ordenador de la UIB.

3. Simulación numérica con el MM4 anidado en el modelo del CEPPM.

4. Comparación de los resultados obtenidos con los modelos MM4 y SALSA.

This work began with Dr. Doswell's arrival on 25 September 1995 at UIB. The work has focused on three events included in the ANOMALIA project: 31 Jan - 5 Feb 1993 ("Algeria"), 08 Oct 1992 ("Menorca"), and 05 Nov 1994 ("Piedmont"). The study of the cases involved the two different lines described in the objectives above: diagnostic studies and numerical simulations. Although these research lines are quite different in their approaches, this should not be taken to mean that they are independent. In the work of studying heavy precipitation events in the western Mediterranean, we have employed the diagnostic and numerical simulation processes in an interactive way. In particular, the simulations have been very helpful in guiding the diagnostic studies.

Objective #1:

The diagnostic work constitutes the bulk of the effort during Dr. Doswell's stay, for reasons to be explained below. This work has proven to be interesting and challenging, especially in the sense that it has become quite obvious that no rigidly-specified diagnostic plan would be satisfactory for all cases. The basic physical processes that cause heavy precipitation have formed the basis for the diagnostic work, and it is clear from the cases that their evolutions are quite consistent with those physical principles. Simply put, heavy precipitation is associated with the rapid ascent of moisture-laden air. However, our studies have revealed that this basic process can come about in different ways. This is what necessitates a flexible, rather than rigid, approach to diagnosis of individual cases.

For the Algeria case, key aspects include:

1. A stagnant synoptic situation. The slow movement of the synoptic pattern enabled the gradual moistening and destabilization of the airmass over the Mediterranean at a time of the year when most airmasses are too dry and stable to produce heavy rainfalls.

2. Horizontal transport of the moist air westward to interact with the topographic features of the eastern Iberian Peninsula. This transport was enhanced by the development of a small cyclone off the Algerian coast, to the lee of the Atlas Mountains of north Africa.

3. Complex mesoscale structures embedded within the stagnant synoptic scale pattern, that modulate the low-level flow. The influences of these barely detectable mesoscale features are difficult to predict but they had a large impact on when and where the interaction of the low-level flow with the topography created the largest threat of heavy precipitation. Thus, day-to-day variations were substantial and difficult to predict.

4. It appears that the SALSA mesoscale model captured many of the important features of the event, including a persistent plume of upward motion along the eastern coast of the Iberian peninsula that was caused by the interaction of the low-level flow and the topography. This information was useful in developing a proper diagnosis of the causes for this upward motion and its associated observed "plume" of high relative humidity.

For the Menorca case, key aspects include:

1. A subsynoptic scale trough embedded in the complex synoptic structure. This trough was associated with a dry intrusion; the dry intrusion, in turn, appears to have altered the character of the convective system associated with it, when compared to the convective system well ahead of the dry air.

2. A complex evolution involving the initial development of two distinct convective systems on the boundary of a region of warm, moist air at low levels. The northern system remained quasistationary while the southern system moved northeastward just ahead of the trough described above. They eventually "merged" into a combined system that continued to move northeastward. Overall, the important convection persisted for more than 24 h.

3. A region of high helicity was also associated with the warm, moist air. This enhanced helicity (ground-relative and storm-relative values were both high) moved along with the southern convective system, and increased with time, making it quite likely that the convection in the southern system had a supercellular character (a conclusion strengthened by the occurrence of a tornado on Menorca). The evolution of the helicity in the SALSA model predictions was quite consistent with the observations. Study of the processes by which this helicity evolution occurred will be done in the future, to be g

4. Heavy precipitation (more than 100 mm in 24 h) fell in Valencia, apparently in association with non-convective moist upslope flow, and on Ibiza in association with the northern convective system, which was slow-moving and apparently had high precipitation efficiency.

For the Piedmont case, key aspects include:

1. The apparent non-convective nature of the heavy precipitation in northwestern Italy. This character is confirmed in the SALSA model simulations, which only showed limited "convective" precipitation. Although we cannot exclude the occurrence of some convective aspects to this event, they obviously are not the dominant process. The diagnostics validate this conclusion by showing the relative stability of the air at low levels flowing into the region.

2. The topography of the region. Although deep, moist convection was not an important process in this event, it supports the basic idea that heavy precipitation arises when moist air rises rapidly. In this case, the ascent is forced by the interaction between the low-level flow and the Alps. The fact that the rain did not fall in a few hours but rather persisted at moderate levels for most of the day is also evidence of the non-convective nature of this event.

The diagnostic study of these three cases is being summarized in a manuscript that is now in preparation and should be submitted soon after Dr. Doswell returns to the U.S. It is intended that this paper be sent to the journal Weather and Forecasting.

Objectives #2,3,4:

These objectives will be grouped under one heading, since they all pertain to the numerical simulation aspects of the study. After much effort and struggle, near the end of Dr. Doswell's stay, the Penn. St./NCAR mesoscale model known as MM4 has been ported successfully to the new Silicon Graphics Indy workstation here at UIB. Since this has been a difficult challenge and a successful run of the model is only occurring at the end of the stay, there are essentially no results upon which to report at this time. Certainly, objective #2 has been carried out successfully, but the simulations associated with objectives #3-4 have just begun.

Given this situation, it is perhaps useful to reflect on the reasons for relatively slow progress in this area. Although Dr. Doswell has used the results of numerical simulations from the MM4 model in his work, he is not a professional model developer. In retrospect, it would have been quite valuable to have had the services of a support scientist who was intimately familiar with the model and its operation. Issues of "portability" of large computer programs are complex and require the regular attention of an expert in several operating systems. Although the help from the UIB Centro de Cálculo was considerable, it is recommended that any future efforts to transport large computer programs from one platform to another include the resources for a computer specialist dedicated at least part-time to the project.

Having said this, it should be noted that the numerical simulation aspect of this work has had positive results. The SALSA model has been run successfully for all three of the ANOMALIA cases and its output, as noted above, has been invaluable in guiding the diagnostic studies. Moreover, the relative success of the SALSA simulations has provided the basis for a companion paper to the diagnostic studies, also in preparation for submission to Weather and Forecasting. An important result of these simulations is that they provide a basis for considerable optimism with regard to an eventual implementation of mesoscale numerical models in operational forecasting of hazardous weather like heavy precipitation and severe thunderstorms. Although by no means perfect, the SALSA simulations suggest that the output from the model might have been very useful guidance in operational forecasting.

Now that MM4 has been implemented on the new SGI Indy workstation, it should be possible to carry out the simulations and comparisons described in objectives #3-4. The MM4 model has two major components that differ in principle from the comparale components in SALSA: the initialization and the convective parameterization. We would like to assess the effectiveness of the two models with regard to these components. In particular, we intend to:

1. Run MM4 for the ANOMALIA cases already simulated by SALSA, in order to carry out the comparisons of the two different models.

2. Run MM4 with initial data that are the same as that used by SALSA.

3. Run SALSA with initial data that are the same as that use by MM4.

4. Run SALSA with the convective parameterization scheme (the so-called Kain-Fritsch parameterization) of MM4.

This last project is already well underway. The preliminary results indicate that perhaps the Kain-Fritsch scheme performs somewhat better than the scheme used in SALSA (developed by Kerry Emanuel of Massachusetts Institute of Technology and implemented in the model by the UIB's Grup de Meteorologica). If this turns out to be sustained by further work, we believe it to be due to the impact of model-resolved vertical motions in the Kain-Fritsch scheme.

Exchanging the initial conditions between the two schemes is an important component of the total project, since they differ so substantially in how they develop initial conditions. MM4 uses a very simple scheme of univariate objective analysis, whereas the SALSA model takes gridded data from the ECMWF and subjects it to considerable multivariate "balancing" in order to reduce the impact of initial dynamic imbalances. Depending on the outcome of this study, this work could have a big impact on how future operational mesoscale models are implemented in practice.

Moreover, the Meteorology Group of the Physics Department at UIB is considering moving their numerical studies to new model platforms. This comparison could have an influence on which models the group would choose to pursue for future work.

Finally, although these are not covered in the original objectives, Dr. Doswell has begun some very promising lines of research while here:

1. The dynamics of convection. This work will not be completed during his stay, but if it turns out to be as promising as the early investigations begun here suggest, it will go forward as a formal publication, perhaps to either the Journal of the Atmospheric Sciences or the Quarterly Journal of the Royal Meteorological Society. It is Dr. Doswell's intention to indicate in the acknowledgements of any publication related to this work that it was partially supported under this proposal, if that is deemed appropriate.

2. Diagnostic methods. The discussions associated with a paper submitted by C. Ramis and E. Tudurí to Weather and Forecasting have been very fruitful. It is anticipated that the methods described in that paper may be applied to some problems of interest with regard to severe storms in the United States. Moreover, the paper's method of calculating storm-relative helicity through a 3 km deep layer beginning with the lowest parcel exhibiting positive CAPE may turn out to be quite promising as a tool for understanding the development of "elevated" supercells (supercells above a shallow surface-based layer of cold, stable air). This work will be pursued upon Dr. Doswell's return to the United States.

3. Tornadoes in the western Mediterranean. Dr. Doswell has become quite interested in the meteorology of tornadoes and supercells in the western Mediterranean. An effort in collaboration with M. Gayà of the INM in Palma de Mallorca and C. Ramis to develop a synoptic climatology of tornadic events in the region has begun.