1. Introduction
Mountainous areas contribute to the generation and modification of synoptic-scale and meso-scale atmospheric flows in various ways. For example, mountains deviate large-scale flows, excite orographic precipitation [
1], and exerting gravity wave drag (see [
2,
3] for overviews). Both weather and climate modeling require a correct representation of these processes on all scales [
4]. This correct representation has to take into account that mountains also significantly contribute to the modification of micrometeorological fluxes and energy budgets, because of the intimate interaction between boundary-layer structures and orographic features. Important international activities in the past (e.g., the Alpine Experiment (ALPEX) [
5], the Pyrenees Experiment (PYREX) [
6], the Southern Alps Experiment (SALPEX) [
7], and the Mesoscale Alpine Programme (MAP) [
8]) have addressed these issues. Especially the last one, MAP, has addressed atmospheric boundary layer issues in more detail [
9]. Another important project, which included experiments related to transport and exchange over complex terrain, was TRACT (Transport and air pollutants over complex terrain, [
10]). This latter project addressed the spectrum of processes which are relevant to describe the transport and turbulent diffusion in the lower atmosphere over complex terrain. Special emphasis was placed on the integration of field observations and mesoscale modeling. Since then, physical scales in time and space that can, in principle, be treated in observations and modeling have been getting continuously smaller [
11]. Surface and airborne remote sensing as well as more satellite programs provide not only better spatial resolution but also better temporal and spatial coverage [
12].
Specific mountainous processes which must be considered in more detail in future energy budget assessments and in climate and weather modeling should include:
Horizontal homogeneity which has been the paramount assumption in so many boundary-layer exchange studies in the past decades (see, e.g., [
16] or [
17]) should no longer be considered appropriate. Neglecting horizontal gradients in the governing equations, which had led to marvelously simple equations and relations in boundary-layer meteorology, is not viable for future mountainous atmospheric research.
New mountain campaigns have to aim at investigating processes contributing to transport and exchange processes between mountainous terrain, the boundary layer, and the free atmosphere at all scales with emphasis on multi-scale interactions. These interactions between relatively well-known processes at meso-scale or synoptic scale and local scales must be much better established so as to assess the relative importance of the different processes in order to decide which of them need to be parameterized in numerical weather prediction and climate models. Moreover, the type of parameterizations should take into model resolution and specific modeling goals. The investigation of the ‘small-scale end’ (i.e., boundary-layer processes such as turbulent exchange, local flows, budgets, scaling approaches, and boundary-layer structures in complex topography) is just at its beginning.
The interaction of different scales in the atmosphere with various types of topographic features is not yet fully understood. This hampers even the identification of where to begin investigating the interactions from small-scale to larger-scale processes in complex terrain [
18]. Some experimental and monitoring initiatives in recent years have already started to fill this gap. Among these are the American effort Mountain Terrain Atmospheric Modeling and Observations (MATERHORN) [
19], the Catalan/Spanish and French effort Cerdanya-2017 [
20,
21], and the German ecosystem exchange program Terrestrial Environmental Observatories (TERENO), with four sites of which three are in complex terrain [
22]. There had been even more detailed campaigns such as the Meteor Crater Experiment (METCRAX) in 2006 [
23], METCRAX II in 2013 [
24], and the Across Scales Experiment (ScaleX) at the alpine site of TERENO looking at very small scales in mountainous terrain [
25]. The ScaleX experiment investigated the conditions for land-air exchange fluxes in a small valley north of the Alps using a large variety of both ground-based and airborne in situ and remote sensing instruments. The backbone of micrometeorological, hydrological and ecosystem-atmosphere exchange instrumentation used by ScaleX was formed by the permanent environmental TERENO-pre-Alpine observatory, with stations distributed along an elevation gradient in the pre-Alpine region. One of the driving research questions for this experiment is to understand why the surface energy balance has yet to be truly closed [
26]. Further field campaigns over complex, mountainous terrain applying in-situ and remote sensing systems include: the Terrain-Induced Rotor Experiment (T-REX) in 2006 [
27], the Convective and Orographically-induced Precipitation Study (COPS) in 2007 [
28], the HYdrological cycle in the Mediterranean Experiment (HYMEX) in 2012 [
29], and Passy-2015 (Passy is a city in the French Alps) [
30,
31]. Long-term measurements are performed in the Inn Valley in the Alps within the i-Box project to study turbulent exchange processes [
32].
Moreover, climate simulations require taking into account significantly longer impact time scales (e.g., permanent and semi-permanent surface characteristics such as vegetation, snow, and ice) in local exchange processes than numerical weather prediction models. Parameterizing, for example, the surface water budget based on concepts adapted from those for flat terrain may not decisively ‘deteriorate the weather forecast’ (e.g., producing the wrong amount of rain in the wrong place and at the wrong time does not really harm the forecast for the next few hours), but it might systematically yield too much (or too little) evaporation. This could result in a particular location (not necessarily only the one concerned with the evaporation) becoming too dry (wet) over the years of simulation, thus altering the meso-scale flow patterns and feeding back to the large-scale flow. Also, application-oriented issues such as the siting of wind turbines (see, e.g., [
33] or [
34]), urban air quality (e.g., for Santiago de Chile [
35]), and air quality in mountainous terrain (e.g., the Salt Lake Valley in the United States [
36], the Arve valley [
30,
31] and the Adige valley [
37] in the Alps, and in the framework of the project ALPNAP the Inn valley [
38] and the Brenner cross-section in the Alps [
39]) need more high-resolution observations [
40].
Finally, large national meteorological and climate centers are moving from numerical weather prediction and/or climate scenario simulations to developing Earth-System Modeling capacities (see, e.g., [
41]). Earth system models include atmospheric constituents such as aerosols and trace gases and deal with vegetation feedback. Therefore, these models must account for the details of earth-atmosphere exchange over mountainous as well—or more generally-complex terrain. These efforts have to be substantiated by respective observational data.
The purpose of this contribution to a series of further papers overviewing the knowledge and challenges on mountain meteorology [
1,
11,
14,
15,
18,
42,
43] is to address the necessary observational techniques which have to be made available in order to achieve the relevant information of transport and exchange processes in complex terrain. The focus will be on radiation, kinetic energy, heat and moisture. Exchange of atmospheric trace substances will not be addressed, although some knowledge on turbulent heat and humidity fluxes can be transferred for this purpose. Therefore,
Section 2 will address the overall challenges, which influence observations in mountainous terrain.
Section 3 will then address surface in situ and surface-based remote sensing observations.
Section 4 describes challenges of airborne and space-borne observations. Conclusions and outlook will close this review.
2. Overall Challenges to Observations in Orographically Complex Terrain
Mountainous terrain influences static stability, dynamics, and thermodynamics of the atmosphere around and above it [
44,
45]. Strong horizontal inhomogeneity and considerable secondary diurnal circulations result in observations that are only representative of a very limited spatial and temporal extent [
46]. Static stability of air in valleys and close to mountains is modified by limited horizontal exchange, by radiative fluxes from elevated terrain, and by secondary circulations, which modify or even suppress turbulent vertical exchange. For instance, calm wintry conditions with a surface snow cover can lead to vertical layering in valleys (see
Figure 1 and [
47]). The flow dynamics are modified by flow over mountains and ridges, flow around mountains and flows through gaps, channels, and passes. Colder air masses can be trapped behind barriers. An extreme example of such local flows is the Laseyer wind in northeastern Switzerland which is not yet fully understood but is able to derail trains [
48]. The Laseyer-wind case study points to a need for very high horizontal resolution modeling in complex terrain (in this case better than 50 m). Similarly, the Passy-2015 case showed that a horizontal resolution of at least 100 m is needed [
49].
Modified thermodynamics may lead to suppressed or enhanced cloud and precipitation formation by thermally or dynamically forced vertical motion. Such vertical exchange processes have been part of the focus of the larger European research project Vertical Ozone Transport in the Alps (VOTALP) [
50]. Relevant secondary diurnal circulations are slope winds, valley winds, and meso-scale winds towards or away from larger mountain chains. Examples of measurements and model simulations are given in the literature for these type of winds in the Elqui valley in the Andes [
51,
52] and the Adige valley in the Alps [
53].
All the above-mentioned studies demonstrated strong spatial variability of nearly all atmospheric variables, which limits the spatial and temporal representativeness of data obtained by measurements (especially those describing larger spatial and temporal scales). For small-scale and micrometeorological variables and turbulence characteristics such as displacement heights, roughness lengths, spectral peaks, turbulent length scales, and profiles of turbulent intensities, the local surface properties exert considerably more influence than topographical complexity does [
54]. The representativeness of observations may be assessed by footprint models [
55], however, parameterized footprint models (e.g., [
56]) are only valid for homogeneous terrain. More sophisticated models (e.g., [
57]) require considerable computational resources [
58].
In contrast to flat, horizontally homogeneous terrain, the exchange and transport of moisture, heat, and mass are controlled not only by convection, but also by thermally induced mesoscale flows and synoptic scale flows [
15]. Besides measuring the turbulent fluxes at the surface, it is thus essential to observe local flows and advection with the mean flow as well. Minimal required observations should include profile measurements of mean quantities such as moisture and temperature at several locations (e.g., in and above valleys, above slopes, and around ridges) along with good knowledge of the three-dimensional flow structure with an appropriate resolution. This poses challenges not only to the measurement techniques and instruments, but also to the coordination and infrastructure. Finding measurement sites in mountainous terrain is usually a compromise between the “best” site from a scientific point of view and feasibility (e.g., sites must be accessible, power supply has to be assured, and landowner permission is needed).
5. Remaining Challenges, Gaps and Potentials
The possibilities to investigate turbulent and mesoscale exchange and transport processes over complex terrain have improved in the last decades; e.g., Doppler lidar systems have nearly closed the gap between small scale turbulence and the scale of convective cells and thermally driven circulations. Additionally, due to commercial application in wind energy assessments the costs of Doppler lidars have decreased significantly so that dual or even multi Doppler lidar applications are nowadays possible. Major challenges arise here from finding suitable measurement sites in complex terrain and by achieving the high necessary synchronization. In addition, even lower temporal and spatial averaging intervals and faster scan speeds have to be sought to fully resolve the spatial structure of turbulence. Nevertheless, Doppler lidars provide a large potential to capture the three-dimensional flow field over complex terrain, which has not been fully exploited up to now, while the three-dimensional structure of temperature and humidity is much harder to observe. So far, the only reliable option to get profiles of temperature and humidity with high vertical resolution is by use of in-situ measurements such as radiosondes, tethersondes, aircrafts and UAVs. However, these approaches suffer usually from low temporal resolution and limited sampling period. Profiles from microwave radiometers can provide complementary high temporal resolution and longer sampling period but (if used alone) suffer from relatively low vertical resolution and the inability in general to detect sharp gradients (e.g., boundary-layer capping inversions) at least above 1 km altitude. Although DIAL and Raman lidars can in principle provide temperature and/or humidity profiles with the necessary resolution, these systems are so far not commercially available, very expensive, and hard to maintain and operate. While analysis tools exist to interpolate data from surface stations in complex terrain [
162,
163], this is not possible above the ground (e.g., in the mixed layer) or of the atmospheric boundary-layer (ABL)) depth. To get highly resolved information in space and time on the temperature and humidity in the ABL and lower troposphere the usage of airborne platforms (UAVs, aircrafts) in combination with radiosondes/tethersondes and microwave radiometers is likely mandatory.
In particular, related to transport and exchange studies over complex terrain, the combination of different remote sensing and in situ instruments measuring different quantities is crucial to capture the relevant processes and process chains. For example, in order to study the processes leading to the evolution of moist convection over mountainous terrain, it is necessary to capture the energy exchange at the surface, continue via convection and thermally driven circulation in the ABL and also consider the exchange between the ABL and the free atmosphere. To study the evolution of moist convection over a mountain ridge, [
164] use radial velocity measurements from Doppler lidar and cloud radar. Near-surface flux measurements over mountainous terrain are already prone to major challenges related to slope angle and representativeness of the site. Getting full profiles of fluxes in the ABL and lower troposphere arises even more challenges. For flux calculations, it is necessary to measure e.g., humidity or heat and wind in the same volume of air at the same time. While flux profiles are possible with tethered platforms they are limited in maximum height and in the number of measurement levels [
165]. Flux profiles can also be obtained from stacked horizontal legs flown by aircraft and UAVs. However, this method is limited over mountainous terrain due to the requirement of long legs and the assumption of horizontal homogeneity along the leg. Remote sensing instruments like lidars have the potential to provide the required data. So far, only few studies present flux profiles from lidar measurements due to the large challenge of operating a Doppler lidar and DIAL or Raman lidar simultaneously. For example, [
166] obtain turbulent flux profiles from airborne DIAL and Doppler lidar measurements. Mobile integrated observation platforms combining different remote sensing and in-situ systems (e.g., KITcube [
164]), Atmospheric Radiation Measurement (ARM) mobile facility [
167], CNRM mobile facility (e.g., [
30]) can provide the required data and can be deployed in complex terrain for limited time periods during field campaigns. The data gathered from such integrated platforms can on the one hand be used for the evaluation of numerical weather prediction models and on the other hand used for data assimilation in Large Eddy simulation (LES) models. Over flat terrain, measurements are already combined with LES modeling within the LASSO project (LES ARM Symbiotic Simulation and Observation, [
168]). This allows achieving a self-consistent representation of the atmosphere around the measurement site, which can assist in a better understanding of the measurements and processes causing the observed conditions. Applying data assimilation to measurements over mountainous terrain is a very challenging task but has large potential to improve our understanding of transport and exchange processes in the future.
6. Conclusions and Outlook
Scanning and range-resolving remote sensing techniques together with multiple low-cost in situ sensors and in situ and remote instruments mounted on highly maneuverable airborne platforms are probably the best instrumentation to capture the full spatial and temporal variability of atmospheric fields in mountainous terrain. Due to the low spatial representativeness of measurement instruments in complex terrain, a larger number of instruments than in flat terrain is needed to provide a full coverage of a given area. Therefore, cost of instruments is an issue.
Furthermore, the dual or triple Doppler lidar technique will help measure wind profiles at positions of virtual towers [
169]. By this, wind profiles will be available from one set of instruments at several sites separated by horizontal distances of some kilometers. Application of the coplanar technique also will improve our understanding of mesoscale flow features over mountainous terrain. These techniques are important, because in complex terrain, the features of vertical profiles and flow features can vary rapidly from one site to the next, much more than in flat terrain. Further progress concerning turbulence information can also be expected from ongoing attempts testing and evaluating appropriate scanning algorithms [
170].
Another parameter, which is highly variable in space and time in the mountain ABL is humidity. As humidity is very important with respect to the initiation of moist convection, its distribution is urgently needed, e.g., for assimilation in Numerical Weather Prediction (NWP) models. Therefore, the development of affordable continuous boundary layer humidity profiling systems is pushed forward in the last years (e.g., [
171]).
But the small representativeness of measured atmospheric and surface data in complex terrain cannot meaningfully be overcome by pure technical solutions only (just more or better instruments per surface area). Sophisticated modeling techniques and elaborated algorithms have to be designed to be used interactively with improved measurements in order to provide reliable data which depict the full spectrum of atmospheric and surface processes taking place in complex terrain [
138]. Steps towards this direction are already being taken, i.e., by exploiting high-resolution modeling capability to bridge the gap from point measurements to larger scales. Inversely, some of the observational techniques could be used to bridge the scale gap between ensemble-averaging mesoscale modeling and case study-based microscale modeling. But also this remains an open issue in complex terrain as well as it has been so long for homogeneous terrain or urban areas (see [
40] and further references therein).
Crowdsourcing is also an alternative to explore how to obtain horizontally highly-resolving data [
172]. The long list of experiments given in the introduction of this paper shows that the issue of covering atmospheric conditions in complex terrain has already been recognized several decades ago. The fact that a perfect solution has not yet been found, points to the intrinsic difficulties of this issue.
Although the development of measurement instruments and strategies is presently ongoing at rather high speed, a solution for a full coverage of transport and exchange processes in the boundary layer over mountainous terrain does not seem to be very near. Therefore, it can be expected that new measurement techniques and data evaluation algorithms will continue to evolve in the future, which—hopefully—bring us a bit closer to a solution. The important role of complex terrain for exchange between the surface/near surface and the upper atmosphere in the climate system justifies every effort in this direction.