The scientific objectives of the Mesoscale Alpine Programme were defined during the first MAP Workshop which was held on September 12-14, 1994, in Zurich. This workshop was attended by 77 participants representing 42 institutions from 13 countries. Scientists from 12 national weather services (Austria, Croatia, France, Germany, Great Britain, Hungary, Italy, Slovakia, Slovenia, Spain, Switzerland, USA) as well as from WMO and ECMWF were present. Hydrological components were integrated following a dedicated workshop on April 26, 1996, in Zurich, which was attended by about 30 scientists, among them hydrologists from Austria, Canada, France, Germany, Italy and Switzerland. A careful evaluation procedure resulted in the formulation of the following primary scientific objectives:
In addition to these primary scientific objectives, MAP supports other research activities. In particular, it supports efforts to establish a climatology of mesoscale weather systems in the Alpine region, and to facilitate the collection and exchange of so-called non-GTS data not operationally distributed through WMO channels. It is also expected that several research groups will contribute, and thus make optimal use of the instrumental set-up deployed for the foregoing primary scientific objectives. Activities related to boundary layer research and flow features at the tropopause level are particularly encouraged, since these will also provide important support on aspects of basic interest to the primary scientific objectives.
The effect of the Alps on the precipitation distribution is well attested and evident in all seasons (see for instance Fig. 2-1). In effect the orography creates specific patterns of ascending and descending air, which enhance or reduce precipitation. In the case of synoptic scale cyclonic activity, the orography enhances precipitation upstream, and in general induces a drying effect downstream. However, if conditions are favourable for convective developments, strong precipitation can occur even downstream. Convection in the Alps is quite common in situations of weak synoptic activity, as for example during the summer season when convection becomes the main source of precipitation. It is usually triggered via thermal circulations or other orographic effects, or may be prompted by upper-air anomalies advected into the area, but this mechanism is not so well understood.
Beside forming one of the most important aspects of the local climate, Alpine precipitation is also the source of specific natural hazards of significant economic importance. The majority of damages by natural disasters in the Alpine region is precipitation related. Severe flooding, hail streaks, land slides and avalanches cause loss of human lives and substantial material damage. Recent examples are provided by the severe flood episode of Piedmont (Italy, 5-7 November 1994), the flash floods of Brig (Switzerland, 23 September 1993), Vaison-la-Romaine (France, 23 September 1992, see Fig. 2-2), Valtellina (Italy, July 1987) and the so-called 'Papal front event' (southern Germany, 3 May 1987).
Figure 2-2. A comparison between observed
and simulated rainfall on 23 September 1992 (Vaison-la-Romaine
flash flood). The radar echo of 1130UTC has been converted into
instantaneous rainfall rates (kgm-2h-1) using the data of all
available raingauges as calibration (grey-scale steps at 4 and
24kgm-2h-1). Superposed is the predicted rainfall of a research
version of the Météo-France Péridot Model
(isolines), run here at 10km resolution, and accumulated for 1
hour (1100 to 1200UTC). This is the best forecast obtained during
subsequent studies on this case, yet it is not precise enough
to assess the extent of the damage in the area. (Courtesy of Météo
France)
A primary objective of MAP is therefore to improve the understanding and short term numerical prediction of Alpine precipitation, and particularly of the forementioned heavy precipitation and flooding events. This objective requires a drastic improvement of our understanding of the flow dynamics at several scales. Indeed, it is known that the dynamical effects of the Alps take place on a broad range of scales. For short term numerical weather prediction, the relevant scales range from the meso-a scale (e.g. lee cyclogenesis and modification of upper level troughs, see Tibaldi et al. 1990), through meso-b scale (e.g. modification of fronts by orography, see Hoinka et al. 1990; Buzzi and Alberoni 1992) to the meso-g scale (e.g. circulations in individual valleys, and inside individual clouds).
Although there exists some knowledge of the preferred synoptic-scale situations for severe precipitation in the Alpine region, little is known at present about the precise mechanisms that lead to the triggering, organisation, localisation, concentration and maintenance over many hours of these types of precipitating systems. Newly available observation systems (e.g. ground-based Doppler radars, and high-resolution numerical models) provide some evidence that episodes of strong precipitation are preceded by and/or associated with detectable mesoscale flow structures, such as upper-level short-wave troughs, convergence lines, shear lines, low-level jets, mesoscale vortices and thermally induced circulations. All these features become potentially more predictable as numerical models improve in resolution and accuracy, and incorporate more advanced representations of physical processes.
However, some basic knowledge is still missing on the precise structure of Alpine precipitating systems at the meso-b scale. Three issues need particular attention:
With the advent of airborne Doppler radars, and stratosphere-troposphere (ST) wind profilers, the instrumental capacity now exists to conduct a major field study and thereby to significantly improve our knowledge.
Beside the dynamical effects of orography, soil and surface properties determine the heat and moisture budgets which in turn affect the hydrological cycle. There is increasing evidence that the correct specification of the surface moisture budget is a key aspect for the numerical prediction of the rain intensity. Thus, substantial effort is required to improve the representation of land-use, soil properties, soil moisture and snow cover in mesoscale models of the Alpine region.
The experimental task related to the Primary Objective 1 of MAP is therefore to collect precise measurements of the structure and dynamics of precipitating systems over the Alps, at the smallest possible scale, together with their meso-b-scale environment, and their surface forcing. Both frontal and convective systems will be sampled. The atmospheric flow settings for intensive studies are of three types:
The resulting new and detailed observational data will serve to validate numerical models at the finest achievable scale. It is foreseen that both meso-b-scale and meso-g-scale models will be used for extensive case studies using the observations acquired during the field phase.
This intensive experimental and numerical effort will be complemented by a similar effort in the treatment of existing radar measurements, with the aim of producing quality controlled maps of precipitation at small spatial (<5km) and temporal (<3hours) scales.
Finally, the analysis of mesoscale dynamical features, and the diagnosis of momentum, heat and moisture budgets will greatly benefit from this effort, resulting in much improved conceptual models of the orographic influence on precipitation.
The MAP Primary Objective 1 builds upon the experience acquired in several European groups in the study of convective and frontal systems in the Alpine foreland. Examples include MATREP (Buzzi et al. 1991), the Front Experiment (Hoinka and Volkert 1992), CLEOPATRA (Haase-Straub et al. 1994), SETEX (Peristeri 1994), Grossversuch IV (Federer et al. 1986; Houze et al. 1993), as well as individual case studies (e.g. Meischner et al. 1991; Volkert et al. 1991; Senesi et al. 1994).
On the mesoscale the atmosphere and the underlying land surfaces represent a heavily coupled system (Eagleson 1978; Brubaker and Entekhabi 1995). On one hand, the moisture content and other soil properties determine the runoff production in response to atmospheric precipitation. On the other hand, the land surface provides fluxes of moisture, heat and momentum into the atmosphere that affect the atmospheric circulation. Both these factors are highly relevant to regional weather and flood forecasting and are substantially affected by complex topography. The task of coupling of hydrological models and numerical weather prediction models opens the possibility of connecting all the water fluxes in the land-atmosphere system. Within MAP pursuance of this highly desirable goal will directly contribute to the further development of flood forecasting systems and water-resource modelling, and also improve the assessment of the local and regional components of the latent and sensible heat input from land surfaces into the atmosphere, and thereby improve the basis for precipitation forecasting in the Alps.
Research areas in this field which require particular attention are (a) the integration of high-resolution radar and satellite data with classical rain gauge observations for precipitation measurements, (b) the improvement of the methodologies for the estimation of return periods of heavy precipitation events in ungauged sites where orographic effects are dominant, (c) the adaptation of soil-vegetation-atmosphere transfer schemes and surface-layer formulations to mountainous terrain, and (d) the integration of advanced hydrological and atmospheric numerical models to provide "coupled" runoff predictions. Further aspects related to these research areas are discussed in turn below.
A. Retrieval of rainfall estimates from radar data
Owing to the short response times (typically one to a few hours) of potentially dangerous watersheds located in mountainous regions, real-time monitoring of rainfall is highly essential, both for the assessment of the risks related to heavy precipitation events and for driving real-time hydrological models. The use of weather radar could substantially contribute to hydrological forecasting although the topography acts as a troubling factor (Joss and Waldvogel 1990, Andrieu et al. 1996). Considerable operational effort has been paid in recent years for the installation of modern radar systems in the Alps. Studies have been performed to better understand specific error sources. Newly developed correction algorithms were tested either against a few cases (e.g. Creutin et al. 1996, Bacchi et al. 1996) or in an operational context (e.g. Joss and Lee 1995). The results obtained are encouraging. Valuable quantitative information can be obtained from radar data (see Fig. 2-3) provided certain conditions are met. These are related to the choice of the radar site and the implementation of a high-rate volume scan strategy with the two-fold objective of maximizing the detection domain and characterizing the vertical structure of the atmosphere. In addition, refined physically-based processing methods are required to identify and eliminate ground clutter (here Doppler systems provide a base for significant advances), correct for partial beam blockage, and account for the vertical profile of reflectivity in order to obtain best estimates of quantitative precipitation at the ground.
Several additional problems need special consideration in the future: First, it is now recognized that the stability of the radar hardware components is essential and that an adjustment procedure based on rain gauge data may be used to compensate for an eventual absolute calibration error. Second, relevant relations between radar measurements (e.g. reflectivity, attenuation) and the rainrate are needed for radar data processing. Hence, the use of in-situ sensors (rain gauges, drop size distribution sensors) is necessary. Third, the merging of data from radar measurements at different altitudes and/or at different wavelengths, and the quantitative assessment of newly available radar parameters (e.g. differential reflectivity, Doppler velocity) are scientific tasks with promising potential.
B. Precipitation distribution
For the implementation of effective flood hazards control policies and the design of civil structures, an accurate knowledge of the probability of occurrence of extreme rainfall events with respect to their duration and spatial extent is required. For this purpose depth-area-duration relationships have been estimated (Grebner 1995, Lang et al. 1996) and different statistical models were applied in the Alps (Jensen 1986, Villi et al. 1987, Margoum et al. 1995, Watzinger 1996). Particular attention has been devoted to the statistics of extreme rainfall depths with a duration ranging from 1 to 24 hours, consistent with the response time of catchments with sizes ranging from some tens to some thousands square kilometres. Despite the numerous investigations, the spatial distribution of extreme precipitation in mountainous areas is still not very well known. Factors contributing to this gap are the bias in the distribution of rain gauge locations to the bottom of valleys, and the lack of sufficiently representative data records. The latter is particularly pertinent for short duration representative of the response time of small upstream basins. Physical considerations can be used to improve the understanding of orographic effects. For instance in the western part of the Alps, recent analyses show that micro-scale climatic effects are of increasing importance as time step increases from one hour to one day, while meso-scale climatic effects are more important as the time step decreases (Desurosne et al. 1995). This appears to be related to pronounced shielding effects induced by the Alpine barrier, as for instance in the upper Durance and Maurienne valleys in France, or in the Italian Alps (see Fig. 2-4). Such effects are also relevant for the precision of the underlying measurements, and associated microclimatic effects related to altitude, wind regime and exposure of the raingauge stations have been investigated in the Swiss Alps (Kirchhofer and Sevruk 1992).
Figure 2-3. Autumn precipitation in the Cevennes region (France). The figure presents time se
ries of hourly hyetographs for two potentially dangerous regions, namely the Gardon d'Ales
(325 km2) and the Gardon d'Anduze (550 km2) watersheds. Radar estimates (solid lines) are de
rived from 10 cm radar reflectivity. The data is corrected for the typical sources of error in moun
tainous areas such as ground clutter, beam blockage, and vertical profile of reflectivity. For cali
bration a single radar-raingauge adjustment factor for the whole event is used. The points and
the vertical bars represent the corresponding raingauge network estimates and their 80% confi
dence intervals obtained through a geostatistical approach. The good agreement between the two
time series is encouraging so as to the possibility of obtaining valuable quantitative information
from weather radar. Such information could be used both for the validation of rainfall estimates
provided by atmospheric models and as input into hydrological models. (Courtesy of H. Andrieu)
Figure 2-4. The rainfall depth with a dura
tion of 12 hours and a return period of
50 years as estimated from more than fifty
stations over the Central Italian Alps using a
log-normal scale-invariant statistical model.
The contours of the rainfall depth are super
imposed on a digital elevation model.
The central part of the Figure covers the Up
per Adda watershed (Valtellina), where a cat
astrophic flood occurred in July 1987. It ex
hibits a clear minimum that is induced by
the orographic shadowing of the Bernina and
the Orobie massifs, which are located, re
spectively, to the North and the South of the
valley. Hourly average rainfall depths de
crease with altitude in the investigated area
since intense short-duration convective
events are a dominant meteorological feature
in the lowland areas. However, the increase
of rainfall depths with duration is more pro
nounced in mountainous areas, probably as
a result of extended orography-induced pre
cipitation events related to the passage of
frontal systems. (Courtesy of B. Bacchi and
R. Ranzi)
Of scientific and practical importance is also the investigation of the scaling properties exhibited by precipitation fields in the space-time domain (Schertzer and Lovejoy 1987, Gupta and Waymire 1993) and in the statistics of their extremes (Burlando and Rosso 1996). How these properties are linked to thermodynamic variables (Pertica and Foufoula-Georgiou 1996) and to the rainfall-generating mechanisms in regions of complex topography is an issue worth to be explored.
C. Surface-Atmosphere Exchange Processes
As a result of both the Global Energy and Water cycle Experiment (GEWEX) and the Biospheric Aspects of the Hydrological Cycle (BAHC) research projects, it is now recognized that the correct formulation of soil-vegetation-atmosphere processes plays a key role in affecting both short-term meteorological forecasts and longer-term climatic projections (Betts et al. 1996). However, intercomparison projects of land-surface models (e.g. PILPS-GEWEX) show that the agreement between different formulations is sometimes small (Love and Henderson-Sellers 1994, PILPS 1994).
In high mountainous areas, the exchange of moisture and heat with the atmosphere is, in addition, highly influenced by the presence of snow cover. Radiation models that take into account the effect of topography (Dozier 1980, Ranzi and Rosso 1995) could be implemented to account for the distribution of radiative fluxes and albedo. Energy and moisture fluxes between the snow surface and the atmosphere (Lang 1981, Olyphant and Isard 1988) can be simulated with distributed models (Ranzi and Rosso 1991, Bloeschl et al. 1991), whereas snow-covered areas can be monitored by means of ground networks and remote sensing (Rosenthal and Dozier 1996).
Another key factor for the determination of the turbulent fluxes of sensible heat, latent heat and momentum is the efficiency of exchange processes between the surface (bare soil or rock, vegetated surface and snow fields) and the atmosphere. This efficiency is essentially determined through the turbulence structure of the lowest atmospheric layers and depends upon the nature of the underlying surface. Presently, the theoretical understanding of these processes is restricted to flat and horizontally homogeneous terrain for which the Monin-Obukhov similarity theory can be used (e.g. Kaimal and Finnigan 1994). In recent years, extensions to this theory have been proposed for either gently rolling terrain (e.g. Carruthers and Hunt 1990) or again idealized settings of inhomogeneity such as step changes in some surface property. Similarly, the turbulent exchange over tall vegetation (forests, crop fields) and within their canopy has been investigated (e.g. Kaimal and Finnigan 1994). However, relatively little is known at present about the turbulence structure over mountainous terrain and hence about the nature and efficiency of exchange processes. In this context, the following aspects require special attention: First, complex topography can induce a mean pressure gradient such that classical scaling variables may loose their physical significance. Second, the rough nature of the surface over the Alps leads to the formation of a roughness sublayer of non-negligible vertical extent (Raupach et al. 1991). Third, the horizontal inhomogeneity of the surface at scales smaller than that of mesoscale atmospheric models makes it very difficult to explicitly determine representative surface fluxes from a limited number of field observations or from the available parameters in a numerical model (e.g. Schmid and Buenzli 1995).
The above issues, and the fact that all these aspects of non-ideality occur at the same time and with still unknown relative importance, will have to influence the design of suitable experimental strategies and numerical experiments in order to proceed towards a better understanding of turbulent exchange processes over mountainous surfaces.
D. Hydrological processes and modelling
Mountainous terrain causes strong spatial variations in the hydrological processes and parameters, which pose particular problems to the assessment and modelling of the hydrological cycle and water balance. The knowledge about the spatial and temporal variability of precipitation, evapotranspiration and soil water storage is still poor, in particular in the higher Alpine regions. For example, recent research reveals vertical evaporation gradients quite different from what has been assumed up to now (Gurtz et al. 1996, Konzelman et al. 1996). More basic research and model development in the field of mountain climatology and hydrology needs to be activated with particular consideration of the scale problem (Lang and Schulla 1996).
The interaction of the gravitational field with the rough (geologically young) topography of the Alps produces complex patterns in the structure of the surface drainage system. Considerable efforts have been undertaken by the developers of lumped schemes to search for general laws that allow the synthesis of associated system variability and process nonlinearity (Dooge 1959, Rodriguez-Iturbe and Valdes 1979, Valdes et al. 1979, Tarboton et al. 1988, La Barbera and Rosso 1989, Rinaldo et al. 1991, 1995). In comparison relatively little such work has been devoted to distributed modelling schemes. The aim is to improve the description of processes at the elemental scale on the one hand (see Beven 1989), and to reduce the computational costs of running distributed models for mid- and large-scale catchments on the other hand. The hydraulic effects related to the locally-varying channel shape in complex drainage networks should also be addressed. Recent results (Orlandini and Rosso 1996) motivate further experimental and theoretical work with the goal to develop efficient parametrization schemes that describe the impact of the stream channel geometry in distributed catchment models.
Physically-based distributed models (Fig. 2-5) of runoff dynamics (Kite and Kouwen 1993, Wigmosta et al. 1994, 1996, Orlandini et al. 1996) offer some advantages and seem to be a potentially useful tool for the current project, since they make use of the full information content of spatially distributed data, and since they can be coupled with reasonable efforts with atmospheric models. Extended experience is available for a north Alpine river basin. With the support of GIS information about the land-surface characteristics, it was possible to determine all water flux processes for the Thur basin (1700km2). The models are grid based, either using the concept of aggregation into representative unit areas, or using regular grid cells with resolutions between 50 metres and some kilometres (Gurtz et al. 1996). Another successful example of distributed modelling is the recent work of Kouwen et al. (1995) in the 50'000km2 Columbia River watershed in British Columbia. Weather data generated by a simple boundary layer model or alternatively by a numerical weather prediction model were used as input to a gridded hydrological model (Kouwen et al. 1993). The hydrological model was designed to make maximum use of remotely sensed data, including detailed land-cover (see e.g. Donald et al. 1995) and meteorological data. The numerical weather prediction model used was the MC2 model (see Benoit et al. 1995), and the investigated periods included periods of intense precipitation. The experience gained in this study shows that the use of numerical weather prediction models to provide the necessary data for flood forecasting is a viable and promising approach.
Figure 2-5. Example of a numerically resolved catchment (Sieve catchment with an area of
about 840 km2). The resolution is 400x400 m, and shading shows cells with overland flow (white)
and channel flow (black). (Courtesy of Stefano Orlandini)
The understanding of stratified airflow past high topography has advanced considerably over the last 25 years. This progress has occurred due to the contributions from four areas of research: (1) theoretical studies, (2) numerical (and laboratory) simulation, (3) case-studies based on operational data, and (4) specially designed field programmes. The frontier of knowledge concerning mountain airflow and weather is related to the processes which have so far been omitted by the theoreticians and modellers, and by the spatial and temporal coverage and the accuracy not yet achieved with observations. Several of these poorly understood aspects are listed below:
Figure 2-6. Cross-section of equivalent
potential temperature from Verona to Munich (Brenner section)
during the Foehn of 8 November 1982, 1200UTC. The analysis is
based on radiosoundings (dashed lines), synoptic stations, and
upper-level aircraft measurements. The ground level refers to
a section along the valley floors, while the shaded area marks
the height of the surrounding mountains. (Adapted from Seibert
1990)
Three-dimensionality: There is a broad body of literature on theoretical and observational studies of two-dimensional aspects of flow past topography (cf. Fig. 2-6). However, real mountain flow is highly three-dimensional (cf. Fig. 2-7). Though a theme of observational and theoretical studies for a long time (e.g. Lammert 1920, Vergeiner 1978, Smith 1979), it is only in the last five years that a concerted effort has started to systematically investigate such flows. Well established results for two-dimensional geometry may require major modifications when a third dimension is added:
A better understanding of these three-dimensional aspects is important for a number of reasons:
Boundary-layer effects: Several studies have demonstrated that the planetary boundary layer significantly affects flow past topography, including the penetration of Foehn into valleys (Hoinkes 1950, Hoinka and Rösler 1987), the generation of low-level vortices (Thorpe et al. 1993), the stabilization of wakes (Grubisic et al. 1994), and the generation of surface gusts (Richard et al. 1989; Miller and Durran 1991). It has also been recognised that some of these processes are important on a planetary scale, and attempts are underway to parametrize low-level effects of subgrid-scale topography in large-scale and mesoscale numerical models (Mason 1991; Georgelin et al. 1994; Lott and Miller 1997). Some of these boundary layer issues are further addressed in Section 2.4.2.
Nonhydrostatic effects: The process of gravity wave breaking, the trapping of gravity waves by stable layers, as well as moist upslope convection are intrinsically nonhydrostatic. While several research models nowadays capture the effect of vertical acceleration airflow dynamics, we do not understand how this effect modifies the airflow separation from sharp terrain features, the horizontal distribution of momentum flux (cf. Keller 1994; Durran 1995) and the structure and location of wave breaking. The current programme comes at a time when the nonhydrostatic models are reaching an appropriate resolution to tackle these problems in a realistic and three-dimensional manner. In order to validate and improve these models, guidance from observations is imperative.
Non-stationarity: Until now, the assumption of stationarity has widely been used in order to gain understanding of the fundamental dynamical characteristics of gravity waves and Foehn-related phenomena. These phenomena are, however, intrinsically time-dependent, and new investigations will be required to improve our understanding of their evolution. The onset and especially the cessation of gravity waves and Foehn are poorly understood, and they are closely linked with the evolution of the synoptic-scale flow such as the progression of frontal systems past the Alps (cf. Egger and Hoinka 1992). Foehn can strengthen or weaken fronts (Heimann 1992a, b), and is usually terminated by the arrival of fronts. This interaction is poorly understood, and it represents one common forecasting problem in the Alpine countries.
Clouds and moisture: While the effect of water condensation has been included in simple two-dimensional airflow models (e.g. Smith and Lin 1982) and numerical simulations (e.g. Durran and Klemp 1982), and while some attempts have been made to document orographic clouds and precipitation (cf. Fliri 1983; Reinking and Boatman 1986; Lee 1986; Banta 1990), we are far from an adequate understanding of the interaction of clouds, precipitation and Foehn-like flows. Whether upstream precipitation enhances or weakens Foehn is unclear (Lilly and Durran 1983; Davies and Schär 1986). The relative contributions of moist ascent and net descent to Foehn warmth and dryness have not been determined climatologically although their importance is disputed (see Seibert 1990). Also, the role of windward evaporative cooling on airflow blocking and splitting is largely unknown. These issues are closely linked with the MAP scientific objectives relating to topographically influenced precipitation (see Section 2.1).
Coriolis cut-off: It has been known theoretically for some time that for scales of motion larger than 100 kilometres or so, the influence of the Coriolis force is to inhibit gravity wave phenomena (Queney 1948; see also Smith 1979; Durran 1990). Thus, airflow on the scale of the entire Alps is non-wavelike whereas airflows on the scale of individual peaks and valleys are wavelike. The role of this wave cut-off and the propagation of inertia gravity waves is significant for momentum transport and for flow over three-dimensional mountains (cf. Trüb and Davies 1994), but little is known about its role on Foehn, airflow splitting, clear air turbulence and wave momentum flux.
Approach
The Mesoscale Alpine Programme has been organised and designed to advance the boundaries of our knowledge and the forecasting capability in mesoscale mountain meteorology. With respect to airflow studies, the Alps are a suitable research area for this programme, because of rather than in spite of their complexity. They have already been well studied using earlier generations of models and instruments. Furthermore, the Alps are - already at current operational level - equipped with one of the densest networks of conventional surface and upper-air observing systems. The deployment of new techniques in the field phase in conjunction with the use of numerical models of the next generation, will enable the analysis of three-dimensional flow past topography, advance our knowledge of mesoscale dynamical processes in mountainous terrain, improve the forecasting of flow features in the Alpine region, and contribute to the understanding and simulation of global climate through improved representations of mountains in general circulation models.
In the context of numerical models in meteorology, the term validation is used in the sense of determining the conformity of numerical model simulations with the available observational data and knowledge. This in turn defines the fitness or suitability for the intended model application. According to Hollingsworth (1994) an elaborate process of cross-validation of model forecasts, data analyses, routine observations and field data is necessary to improve the forecast model's performance. The pressing need to validate high-resolution mesoscale models in a systematic fashion is acknowledged (Chouinard et al. 1994, Kuo 1993), but only a few detailed studies are available for regional limited area models with horizontal mesh size below 100km (e.g. Anthes et al. 1989).
In the early nineties, meso-b-scale numerical models covering the Alpine area became operational at several meteorological services. These high-resolution models aim explicitly at weather prediction (i.e. a correct spatio-temporal specification of cloud cover, rain areas, daily temperature extrema, etc.) rather than merely flow prediction (i.e. height and pressure patterns, wind fields; cf. Wergen 1994). Thus in addition to standard techniques (e.g. as compiled in Stanski et al. 1989) mesoscale model validation has to focus also on a detailed examination of physical parametrizations, as was claimed by Hollingsworth (1994) and exemplified for a gravity wave drag parametrization scheme by Bougeault et al. (1992).
Thus, important MAP tasks are: to systematically validate the performance of high-resolution models, to assess the quality of the regional forecasts in the Alps, and to stimulate model improvements.
Table 2-1 provides an overview of the typical mesh size and data storage interval of current operational models, along with typical station spacing and sampling intervals of different observational networks. Estimated values are included for the nonhydrostatic models that are currently under development and are expected to become operational during or shortly before the course of MAP. The spatio-temporal resolution of upper-air and surface observations lags significantly behind that of model data. Precipitation measurements have a high spatial resolution for daily sums, yet the data collection is organised nationally and not included in the GTS-network (see Section 2.4.1). Areal precipitation derived from the radar network is still not integrated into systematic validation schemes. Wind soundings with high temporal resolution from radar wind profilers are foreseen as providing important new information, even if only installed as isolated systems in the coming years.
The gap between the density of observed and forecasted data will significantly increase when the new generation of meso-g models with ten times smaller mesh size becomes available.
The first item necessitates the systematic collection of all routinely available data from within and outside the meteorological services, and the development of a validation strategy, which is intermediate between studies of single cases (e.g. Volkert et al. 1992; Pichler et al. 1993) and the operational calculation of skill scores for extended regions and periods of time (e.g. Majewski and Schrodin 1994). These tasks are to be started right at the beginning of MAP. Before the field phase data become available, diagnostic tools could be applied for the validation of thermodynamic processes during strong precipitation events (cf. Dorninger et al. 1992). The second item is to be tackled after the field phase of MAP. However, the processes to be evaluated (gravity wave parametrizations, dry and moist planetary boundary layer processes, cumulus parametrization schemes) have to be precisely defined early on in the programme so that sufficient parameters (e.g. heat and moisture fluxes) are observed during the special observing period of the field phase. It is foreseen that the observations from the field phase will serve in particular to validate the first meso-g-scale model performances in simulating Foehn and lee waves, breaking waves, intense convection and precipitation.
The data sets with enhanced resolution and a wider selection of
parameters are important for both initialisation and evaluation
of the numerical models. The Special Observing Period of MAP ought
to be long enough and cover a sufficiently large area to serve
both purposes.
SYNOP stations
Rain gauges (climatological and hydrol. networks)
Radar network (quantitative precipitation)
Wind profiler
30
15
5
-
1 or 3
12 or 24
0.25
0.25
Operational NWP Models 15
1 Upper-air network
250 6 or 12 nonhydrostatic models (under development)
1 -
Table 2-1. Typical spatial (horizontal
direction; km) and temporal (h) resolution of current and planned
operational NWP models and various observing systems.
Hydrological river basin models are used in several ways: (1) to derive a quantitative understanding of the different time-dependent processes in a river basin, and to determine the variations of the transfer and state variables; (2) to forecast important hydrological variables (runoff, soil moisture, snowpack, etc.) based on either observed and forecasted meteorological variables; (3) to assimilate observations from different sources into a spatially consistent time-dependent framework.
In order to run hydrological models, observed or predicted meteorological variables (i.e. precipitation, temperature, water vapour pressure, wind speed and radiation) are used as input. The spatial scale of most Alpine river basins is however such that in the past the resolution of both operational NWP models and/or observed meteorological data was often too low and thus incompatible with the needs of operational hydrological models. One particularly important objective of MAP is to improve the resolution and functionality of the models and thereby to allow for either one-way or fully interactive coupling of the meteorological and hydrological models. Data from the MAP General Observing Period (GOP) will be highly useful for the development, improvement and validation of relevant model components. Emerging "model chains" could efficiently be used for operational runoff and flood prediction purposes on the scale of individual river basins.
The combined use of high-resolution meteorological and hydrological models offers attractive advantages since it closes the water budget on the scale of individual valleys. For instance, precipitation is the most important input parameter for hydrological models, yet it is also very difficult to predict by atmospheric models, and also to observe in a spatially consistent manner. On the other hand, runoff can be measured quite accurately and may be used to validate model-derived rainfall amounts and precipitation-interpolation procedures. Such data could be used in two ways: First, coupled meteorological / hydrological model chains can directly be validated in terms of runoff predictions over selected river basins. Second, simulated precipitation of atmospheric models can be compared with rainfall amounts as derived over extended periods from the water balance in selected catchments. Since the water balance in the Alpine region is dominated by precipitation and runoff, the evapotranspiration term in the water balance can be estimated with sufficient accuracy.
Similar validation methodologies are recommended at larger scale by the WMO WCRP-Water project "Grid estimation of runoff data" (WMO 1994) which aims towards the preparation of monthly central European runoff data over a 30-year period with a grid resolution of 0.5 degree. The approach has also been used in the GEWEX Continental-scale International Project (GCIP) in the Mississippi River basin, the Baltic Sea Experiment (BALTEX 1995), and in the Columbia River Study (Kouwen et al. 1995). Essentially the same methodology has also been pioneered to validate surface-flux measurements in the Boreas Ecosystem-Atmosphere Study (BOREAS, see Sellers et al. 1994).
If hydrological models are driven by observed meteorological parameters, point observations must be interpolated into a high-resolution grid. This procedure is not an easy task in Alpine environment, in particular not for precipitation data. When calibrating hydrological models, interpolation errors may thus lead to distorted model parameters. Results from high resolution meteorological models together with information collected by radar and conventional observing systems during the MAP general observing period (GOP) could help to advance the understanding of the spatial distribution of precipitation in mountainous regions and hence to improve interpolation and calibration procedures for hydrological models.
MAP will therefore provide an ideal opportunity to carry out research on:
The choice of suitable river basins has to be done very carefully. Relevant criteria are:
The climate or climatic state of a region is defined as the mean state of the atmosphere which is the average behaviour of the regional-scale land-atmosphere system over relatively long periods (Landsberg 1967). This mean state is described by the statistics of
A complete regional climatology contains statistics of all three types itself, as well as statistics which describe the link between the different scales, from the local scale up to the synoptic scale. These statistics are derived by aid of methods of synoptic climatology (Barry and Perry 1973; Yarnal 1993).
Early work on Alpine climatology was published more than a century ago (for an overview see Fliri 1975). So far, conventional Alpine climatology - statistics of type 1 - consists of long-term time series of meteorological parameters at single stations and corresponding spatial distributions (e.g. Fliri 1975; Volkert 1985; Frei and Schär 1997). Recent examples of work on statistics of type 2 (e.g. Foehn) can be found in Seibert (1985), and on statistics of type 3 (e.g. synoptic-scale weather pattern, `Grosswetterlagen') in Hess and Brezowsky (1969). Further work deals with the statistical relation between synoptic-scale weather and meteorological parameters measured locally (Fliri 1984; Gerstengarbe et al. 1993, Cacciamani et al. 1995).
Nevertheless, at present a climatology of the Alpine region is far from being complete. Therefore, one MAP objective is to establish an Alpine climatology which contains a set of statistics describing the climatic states of the following mesoscale processes and features that are prominent in the Alpine region:
This regional Alpine climatology should contain statistics on the mesoscale atmospheric flow structures and their relation to both, smaller-scale local weather and larger-scale synoptic forcing. The statistics should describe the space-time variability, correlation and scale interacting characteristics of parameters and patterns on all scales relevant for the Alpine atmosphere.
The compilation of such a climatology should seek to optimize the use of existing observational data, beyond that transmitted operationally by WMO through the Global Telecommunication System (GTS). Currently all the Alpine countries operate dense non-GTS climatological networks (e.g. for precipitation). This data has been exploited with respect to the specific questions of the individual countries, but their use for the study of atmospheric processes over the whole of the Alps has suffered from the lack of an international, Alpine-scale database. By combining the various national data sets, a valuable archive of Alpine data could be compiled. Such an undertaking is time-consuming (see Hulme 1994), but is more readily attained within the frame of an international programme. MAP will aid the establishment of fruitful contacts with and between the various data supplying institutions, which in turn would accelerate the administrative procedures. Some effort should also be considered for homogenization of available routine and climatological data collected in various Alpine and close-Alpine countries.
It is also necessary to build up an integrated physiographical data set containing climatological information on soil and surface properties of the entire Alpine region including their daily, seasonal and annual variability. Satellite data (LANDSAT, NOAA etc.) would be very helpful in order to derive terrain height, vegetation and land-use data (Albertz 1991).
Clouds play an important role in the hydrological cycle. At present, the cloud distribution above the Alpine region is not well documented in its climatological variability. Remote sensing methods are a powerful tool for providing a climatology of clouds above the entire Alpine region in their monthly and seasonal variability (cf. Kästner and Kriebel 1994).
Additional effort is required to support the development of regionalization methods, i.e. upscaling and downscaling procedures. Firstly, scaling methods that can to a measure identify mesoscale processes from finer-scale observations would be of great utility (e.g. the regional generalization of very local rain-gauge measurements). At present, there is an obvious lack of methods presenting the point-to-region scale linkage, in particular for precipitation measurements. Secondly, downscaling procedures are required in order to obtain statistics of regional phenomena from long-term synoptic-scale analyses, e.g. the development of convection depending on larger-scale flow types. Finally, it is desirable to obtain a climatology of numerical forecasts of particular mesoscale situations, e.g. events favourable to wave breaking. This latter method should provide climatological information on phenomena which at present are not detectable by routine observational networks.
In summary, the desired set of statistics and the methods to be developed are a valuable contribution to climatology in general. Moreover, such an undertaking describes the present-day state of the climate of the Alpine region, and it is therefore fundamental for the estimate of potential future changes of the Alpine regional climate and their socio-economic impact (e.g. Kane et al. 1992). Moreover, this set of statistics does help the planning of the MAP field phase by supplying information for the selection of the location, timing and duration of the Special Observing Period.
The overall aim of this task is to investigate the structure and evolution of the planetary boundary layer (PBL) and its effect on Alpine weather systems.
The usual concept of a PBL comes from ideas developed over flat or homogeneous terrain, in which the layer is considered to be relatively uniform horizontally, but is evolving with time. The effects of turbulence, radiative flux divergence, advective effects, and cloudy convection are addressed in this context with various boundary layer theories (Stull 1988). In areas of complex terrain such as the Alps, further complications to PBL structure and evolution arise because of additional physical processes and special topographic effects (see Blumen 1990). The development of heated or cooled boundary layers on slopes instigates motion up or down the inclination, and in complex terrain it produces interacting slope, valley, and mountain-plain wind circulations. Terrain irregularities and variations in surface cover and ground moisture play a role in modifying the circulations at individual locations, as do mechanically driven circulations produced by interactions between the mountains and the prevailing flow. Because of these complications the Alpine boundary layer is poorly understood quantitatively, and its parametrization in existing numerical models, now based largely on flat terrain concepts, is expected to be a key area where advances in understanding can pay important dividends in improving forecasts of Alpine weather.
The PBL plays an important role in Alpine weather. It supplies the heat and moisture that generate clouds and precipitation (Banta 1990); it affects the development and penetration of the Foehn and other severe windstorms (Hoinkes 1950; Hoinka and Rösler 1987; Richard et al. 1989); it develops inversions that decouple airmasses within the topography from the winds aloft, allowing moisture, clouds, fogs and pollutants to build up within basins and valleys (Petkovsek 1992); and it generates the thermally driven wind systems (cf. Vergeiner and Dreiseitl 1987) that transport and disperse pollutants (ASCOT Investigators 1989; Whiteman 1990).
A research programme which combines field experimentation, numerical modelling, and theory is therefore envisioned to improve our understanding of the structure and evolution of the Alpine boundary layer and its representation in operational forecast models. This research programme will benefit from new research-grade numerical models that are able to treat Alpine topography on a scale heretofore unrealized, and new and improved observational technologies including atmospheric remote sensors (Neff 1990), air motion tracers, and airborne lidars and turbulence instrumentation.
Specific research topics that should be addressed in this topical area include:
These research topics will also be useful in addressing questions relating to thermally-driven wind systems within the Alpine topography (slope, valley, and mountain-plain wind systems), and effects of boundary layer processes on air pollution transport and dispersion within and across the Alpine region. For instance, several Alpine countries suffer from enhanced levels of pollutants during the summer and autumn season (cf. the period of the field campaign).
The research on the boundary layer topics will also benefit from the special observational and modelling resources of the MAP, and from the collaboration of research institutes and operational weather service components from many countries.
The Alpine region hosts a wide range of mesoscale flow phenomena that are essentially confined to the troposphere (see Section 2.1 and Section 2.2). In addition there are systems that are centred at tropopause elevations (e.g. tropopause folds). These upper-level features (ULFs) advect toward the region as deep (~4km), slender (~200km) and elongated (~1500km) intrusions of stratospheric air into the troposphere (cf. Danielsen 1968). Relative to the ambient atmosphere these bands are characterised by high values of potential vorticity and ozone and low values of humidity (Danielsen et al. 1987). Another key factor is that they are associated with, and accompanied by, a flow signal that although it decays with distance downward remains significant (~5-10ms-1) at low-levels, and thus the bands are an integral and important component of mid-latitude synoptic activity (see for instance Hoskins et al. 1985; Davis and Emanuel 1991). Furthermore they are the seat for significant stratosphere-to-troposphere exchange of atmospheric constituents (Danielsen 1968; Ebel et al. 1991; Hoerling et al. 1993; Lamarque and Hess 1994) and thereby they serve to modify key atmospheric radiative and chemical processes.
The occurrence of these ULFs over central and southern Europe is sufficiently prevalent for the potential vorticity imprint to be discernible in the monthly mean climatology at tropopause-level (Lau et al. 1981). In the Alpine region itself the low-level signal of the incoming ULFs is modified in two ways. Firstly, observational evidence (Appenzeller and Davies 1992) indicates that during their passage toward the Alps the bands often acquire a rich mesoscale structure as they break-up into a train of vortex-like disturbances, and secondly Alpine orography constitutes a potentially major perturbing effect. The bands often accompany the Alpine passage of surface fronts (Davies 1989) and there is considerable evidence linking ULFs to lee-cyclogenesis events (see for instance Bleck and Mattocks 1984; Lanzinger et al. 1990). There are also case study examples (Tafferner, personal communication) and some statistical-climatological information linking ULFs with isolated events of deep convection. There is also some evidence linking ULFs to incidences of anomalous high values of ozone at low elevations (Buzzi et al. 1984, 1985; Davies and Schüpbach 1994).
The general aims of this subprogramme are to obtain a better understanding of the structure and dynamics of the ULFs and to examine their influence upon atmospheric processes and weather systems in the Alpine region (cf. Fig. 2-8). The specific objectives are:
These objectives are closely tied to, and would serve as a means for testing, the following hypotheses regarding ULFs:
In addition to the Foehn, the Alps also induce a number of well known regional winds, such as the Bora (see Smith 1987; Jurcec 1989; Glasnovic and Jurcec 1990; Bajic 1991; Vucetic 1991), the Mistral (see Pettre 1982; Blondin and Bret 1986), the Bise in the Swiss Alpine foreland (cf. Wanner and Furger 1990), as well as an easterly low-level jet over the Po valley (see Tampieri et al. 1984; Buzzi and Alberoni 1992). The occurrence, intensity, and horizontal extension of these wind systems are all related to influence of orography. They merit further study in order to improve their practical forecasting and their representation in numerical models. Also, the total pressure drag of the Alps may be dominated by effects associated with one single regional wind system during certain episodes (Tutis and Ivancan-Picek 1991).
The studies of local wind systems will clearly benefit from a synergy with the primary objectives of MAP, as the basic physical mechanisms leading to their formation are comparable. In turn, these local wind systems contribute to the mesoscale environment, and must often be included in the forecasting of other features such as precipitation, convection and Alpine lee cyclogenesis.
