Diurnal Winds in the Himalayan Kali Gandaki Valley. Part III: Remotely Piloted Aircraft Soundings

Joseph Egger Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Joseph Egger in
Current site
Google Scholar
PubMed
Close
,
Sapta Bajrachaya Department of Hydrology and Meteorology, Ministry of Science and Technology, Kathmandu, Nepal

Search for other papers by Sapta Bajrachaya in
Current site
Google Scholar
PubMed
Close
,
Richard Heinrich Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Richard Heinrich in
Current site
Google Scholar
PubMed
Close
,
Philip Kolb Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Philip Kolb in
Current site
Google Scholar
PubMed
Close
,
Stephan Lämmlein Fachbereich Maschinenbau, Fachhochschule Regensburg, Regensburg, Germany

Search for other papers by Stephan Lämmlein in
Current site
Google Scholar
PubMed
Close
,
Mario Mech Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Mario Mech in
Current site
Google Scholar
PubMed
Close
,
Joachim Reuder Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Joachim Reuder in
Current site
Google Scholar
PubMed
Close
,
Wolfgang Schäper Astrium, Friedrichshafen, Germany

Search for other papers by Wolfgang Schäper in
Current site
Google Scholar
PubMed
Close
,
Pancha Shakya Department of Hydrology and Meteorology, Ministry of Science and Technology, Kathmandu, Nepal

Search for other papers by Pancha Shakya in
Current site
Google Scholar
PubMed
Close
,
Jan Schween Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Jan Schween in
Current site
Google Scholar
PubMed
Close
, and
Hilbert Wendt Meteorologisches Institut, Universität München, Munich, Germany

Search for other papers by Hilbert Wendt in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

In 1998 a field campaign has been conducted in the north–south-oriented Kali Gandaki valley in Nepal to explore the structure of its extreme valley wind system. Piloted ballon (pibal) observations were made to map the strong upvalley winds as well as the weak nocturnal flows (Part I). The stratification of the valley atmosphere was not explored. In Part II of this multipart paper, numerical simulations are presented that successfully simulate most of the wind observations. Moreover, the model results suggest that the vigorous upvalley winds can be seen as supercritical flow induced by contractions of the valley. Here, the results of a further campaign are reported where remotely piloted airplanes were used to obtain vertical profiles of temperature and humidity up to heights of ∼2000 m above the ground. Such profiles are needed for an understanding of the flow dynamics in the valley and for a validation of the model results. This technique is novel in some respects and turned out to be highly reliable even under extreme conditions. In addition four automatic stations were installed along the valley's axis. Winds were observed via pibal ascents. These data complement the wind data of 1998 so that the diurnal wind system of the Kali Gandaki valley is now documented reasonably well.

It is found that the fully developed upvalley flow is confined to a turbulent layer that tends to be neutrally stratified throughout the domain of observations. The stratification above this layer is stable. A capping inversion is encountered occasionally. This finding excludes explanations of the strong winds in terms of hydraulic theories that rely on the presence of strong inversions. Pairs of simultaneous ascents separated by 5–10 km along the valley axis reveal a remarkable variability induced by the topography and, perhaps, by an instability of the flow. The analysis of the surface data as well as that of the soundings shows that the flow above the neutral layer affects the surface pressure distribution and, therefore, the acceleration of the extreme upvalley winds.

Corresponding author address: Joseph Egger, Meteorologisches Institut der Universität München, Theresienstr. 37, 80333 München, Germany. Email: j.egger@lrz.uni-muenchen.de

Abstract

In 1998 a field campaign has been conducted in the north–south-oriented Kali Gandaki valley in Nepal to explore the structure of its extreme valley wind system. Piloted ballon (pibal) observations were made to map the strong upvalley winds as well as the weak nocturnal flows (Part I). The stratification of the valley atmosphere was not explored. In Part II of this multipart paper, numerical simulations are presented that successfully simulate most of the wind observations. Moreover, the model results suggest that the vigorous upvalley winds can be seen as supercritical flow induced by contractions of the valley. Here, the results of a further campaign are reported where remotely piloted airplanes were used to obtain vertical profiles of temperature and humidity up to heights of ∼2000 m above the ground. Such profiles are needed for an understanding of the flow dynamics in the valley and for a validation of the model results. This technique is novel in some respects and turned out to be highly reliable even under extreme conditions. In addition four automatic stations were installed along the valley's axis. Winds were observed via pibal ascents. These data complement the wind data of 1998 so that the diurnal wind system of the Kali Gandaki valley is now documented reasonably well.

It is found that the fully developed upvalley flow is confined to a turbulent layer that tends to be neutrally stratified throughout the domain of observations. The stratification above this layer is stable. A capping inversion is encountered occasionally. This finding excludes explanations of the strong winds in terms of hydraulic theories that rely on the presence of strong inversions. Pairs of simultaneous ascents separated by 5–10 km along the valley axis reveal a remarkable variability induced by the topography and, perhaps, by an instability of the flow. The analysis of the surface data as well as that of the soundings shows that the flow above the neutral layer affects the surface pressure distribution and, therefore, the acceleration of the extreme upvalley winds.

Corresponding author address: Joseph Egger, Meteorologisches Institut der Universität München, Theresienstr. 37, 80333 München, Germany. Email: j.egger@lrz.uni-muenchen.de

1. Introduction

The Kali Gandaki valley in Nepal stands out both because of its extreme geometry and the intensity of the diurnal upvalley winds. The Kali Gandaki River originates near the town Lo Manthang (see Fig. 1) and flows southward through the Mustang basin. It cuts through the Himalayan barrier between the villages of Marpha and Ghasa forming there one of the deepest valleys on Earth. Farther south, the river rushes down into a gorge to reach the lower parts of Nepal at an altitude of ∼1000 m above MSL. The Mustang basin extends from Marpha to Lo Manthang. It is confined by the towering mountain chains to the east and west and by the Himalayas in the south. A mountain pass leads to the Tibetan Plateau about 20 km to the northeast of Lo Manthang.

Before 1998, scattered information was available indicating that the diurnal wind system of the valley exhibits rather strong upvalley winds (∼20 m s−1) between Marpha and Kagbeni (see Egger et al. 2000, hereafter KG1, for details and relevant literature). This upvalley wind is called Lomar by the locals (“southerly wind”; we change here the spelling “Lhomar” as used in KG1 and Zängl et al. (2001, hereafter KG2) to “Lomar” to be consistent with the spelling of Lo Manthang where lo refers also to the south). Nocturnal downvalley winds appeared to be weak.

In fall 1998 the Meteorological Institute of the University of Munich and the Department of Hydrology and Meteorology in Kathmandu conducted a joint field campaign in order to explore the structure of this wind system in detail (KG1). The following conclusions of KG1 are based on about 100 double-pilot ascents performed at eight locations covering the distance from Lete to Lo Manthang.

  1. The upvalley winds start near the surface before noon. The layer of strong winds with typical velocities of ∼15 m s−1 grows over about 1 h to a depth of ∼1000–1500 m.

  2. The breakdown of the upvalley wind regime after sunset begins close to the ground. The upvalley flow ceases before midnight.

  3. Nocturnal downvalley flows are quite weak.

  4. Upvalley winds are less powerful both in the so-called exit region between Chuksang and Lo Manthang and in the entrance region (Ghasa–Tukuche) than in the core region (Marpha–Kagbeni).

Stimulated by these results, Zängl et al. (2001) performed numerical simulations of the Kali Gandaki wind system using the Pennsylvania State University–National Center for Atmospheric Research fifth-generation Mesoscale Model (MM5). The total flow domain of the simulations included the Tibetan Plateau as the dominant topographic feature of the region. Five nests were needed to resolve the core region reasonably well with a grid size of 800 m. There are 38 levels in the vertical with a maximum resolution of 100 m near the ground. The initial state is in thermal wind balance with the meridional temperature gradient. The level of no winds is chosen such that the atmosphere is almost at rest in the Kali Gandaki valley. The model calculations were quite successful in that nearly all the observed features of the wind field were reproduced in a reference run. Sensitivity experiments were carried out in order to elucidate the mechanisms driving the valley wind system including precipitation [see also Barros et al. (2000) for a recent precipitation analysis of the area].

It is a key result of KG2 that a rather stable layer with strong upvalley winds forms during the day in the entrance and, in particular, in the core region (see Fig. 2). This layer of 1000–1500-m thickness is found upstream of the widening of the valley near Marpha. The isentropes descend between Marpha and Jomsom. The flow accelerates in the descending branch to attain a maximum speed of 23 m s−1 near Jomsom. The layer of rapid flows stays close to the ground farther to the north and ascends toward Tibet with slightly reduced flow speeds. Note, however, that inflow of moderate speed occurs above this layer up to a height of ∼5000 m. The stability of the Lomar layer increases from Tukuche to Jomsom. The Brunt–Väisälä frequency is N ∼ 1.2 × 10−2 s−1 in Marpha and 1.6 × 10−2 in Jomsom where the isentrope θ = 315 K and the ground are chosen as reference surfaces (θ equals potential temperature).

The flow pattern in Fig. 2 is reminiscent of the results of laboratory and hydraulic model studies of flows through lateral contractions. For example Arakawa [(1969); see also Baines (1995) for an updated outline of the theory] analyzed single-layer flow through a valley of variable width where the Froude number F = U(gH)−1/2 is the key parameter (U, flow speed; g′, reduced gravity; H, depth of the flow). There exists a class of solutions where F = 1 at the narrowest point and where F > 1 downstream. This type of flow is similar to that in Fig. 2 where event indications of a hydraulic jump are seen downstream of the maximum contraction near Marpha. Such models have been invoked by Pettre (1982), Jackson and Steyn (1994a, b), Pan and Smith (1999), and others to explain strong wind storms in valleys and gaps, Structures as displayed in Fig. 2 can also be found in two-layer models (Armi 1986; Baines 1995) and in continuously stratified flows underneath a free surface representing an inversion (Armi and Williams 1993).

In principle, hydraulic theory cannot be applied to thermally driven flows simply because sources and sinks of heat are not included. Here, however, the situation is somewhat different. As explained in more detail in KG2, the diurnal heating of the Mustang basin generates a pressure difference between, say, Jomsom and the “free” atmosphere to the south of Ghasa. The strong winds can be seen as a response to this pressure gradient in much the same way as gap winds respond to imposed large-scale pressure fields. Moreover, Fig. 2 suggests that the Lomar is separated by a rather stable top layer from the atmosphere above. All this indicates that an application of hydraulic theory might help to better understand both the observations and the model results. In turn, information on the thermal structure of the valley atmosphere is needed in order to better understand the vigorous Kali Gandaki valley winds. Such information was not provided by the campaign of 1998. Standard instrumentation for vertical soundings [see e.g., Clements et al. (1989) for the design of a valley flow experiment] like radiosondes is impractical in these remote areas where the gas needed to fill the balloons has to be carried up by porters. Tethersondes cannot be used for the same reason. Moreover, the winds are too violent. However, battery-powered model airplanes with remote control (remotely piloted vehicles, RPV) are highly suitable for this purpose. They are light, their energy demand is low, and they can be carried by porters to almost any starting position in the Mustang region. Their maximum ascent height of ∼2000 m is sufficient to penetrate the layer of strong winds. Such planes were used during the field campaign in 2001 to be described in this paper (18 February–27 March). Double theodolite pilot balloon observations were carried out simultaneously. In addition, an array of surface stations was installed. It was the main goal of this effort to collect information on the stratification of the upvalley flow both in the narrow part of the valley and in the Mustang basin and to verify in this way the model results of KG2 as displayed in Fig. 2.

The observations described in KG1 were made in the fall. By choosing February and March we wished to learn more about the seasonal variability of the Kali Gandaki valley wind system. The mean wind speeds in February and March 1990 in Kagbeni are shown in Fig. 3. These curves are fairly similar to those for September and October (see Fig. 2 of KG1) except that the winds are slightly less vigorous than in the fall. Note that Fig. 2 represents equinoctial conditions in the simulations of KG2 that do not take into account the observed climatological mean flow. Therefore, a validation of Fig. 2 is possible in March as well as in September. The climatological flow has a southerly component at upper levels in the fall while westerlies prevail in the spring (e.g., Ramage 1971). Throughout the campaign the flow at 500 hPa was generally westerly although many perturbations moved over the area from the west.

This paper is organized as follows. The equipment is described in section 2. The surface observations are presented in section 3, the soundings in section 4. A discussion is presented in section 5, and concluding remarks are given in section 6.

2. Instrumentation

a. Airplanes

The idea to use model airplanes as a carrier of instruments in meteorology is not new. Corresponding experiments were conducted successfully in the 1970s (Konrad et al. 1970; M. Reinhardt 1997, personal communication). More recent activities are described in Renno and Williams (1995), Chilson et al. (1999), and Stephens et al. (2000). The new airplanes were designed by W. Schäper (WS) and built by WS and S. Lämmlein in cooperation with Modellbau Ulrich and Blue Airlines. They can be flown up to heights of at least 5000 m above sea level in highly turbulent wind fields. The prototype Kali is shown in Fig. 4a. The plane has a length of 1.29 m and a wing span of 2.10 m. The total mass in 3 kg. Flight velocities are in the range of 10–40 m s−1. The optimum climb rate for highest altitude is ∼5 m s−1. The propeller is driven by an electromotor (Hacker HBR 50S26). Power is provided by 14–16 rechargeable NiMH cells. To ease control of the planes in turbulent flow, the planes are equipped with gyro systems for stabilization around the roll axis. Special binoculars have been developed by Firma Zeiss so that the plane can be followed visually up to heights of ∼2000 m above the ground (Fig. 4b). To increase color contrast, blue blocker sunglasses, also made by Zeiss, were used. The pilot controls the plane by a Robbe radio gear control. Nets are used for landing in difficult terrain and in strong winds. A sounding is normally completed within 15 min. The flight path during ascent is chosen to maximize the ascent height; that is, the pilots try to exploit slope winds and thermals. Descent is performed so as to equal time spans for ascent and descent. The planes were steered by several of the authors (P. Kolb, S. Lämmlein, and WS).

The planes carry sensors for an observing system designed and built by Ingenieurbüro Würtenberger consisting of a miniaturized datalogger and specially adapted sensors for pressure, temperature, and humidity. The pressure sensor is based on the Motorola MPX 2100. Its signal is amplified and compensated for changes of temperature. The resolution is ∼1 hPa. Humidity is recorded via the HIH-3605-B Honeywell sensor with 0.6% resolution. The response time is ∼5 s. Temperature is obtained from a LM50 C National semiconductor sensor. The resolution is ∼0.2 K and the response time is ≤1 s. Ascents for an adiabatic lapse rate must be expected to be 0.3°C warmer than descents because of the related hysteresis effect. Three wind generators provided the electric power to recharge the batteries of the planes and all portable computers. One plane out of 11 was lost during the campaign, while another 1 was damaged but could be repaired. It is impossible to fly the planes during the night; therefore, we concentrated on the daytime flow evolution during this campaign.

b. Permanent stations

Four permanent stations were installed. They contained instrumentation for temperature, wet-bulb temperature, pressure, and wind speed and direction. These instruments were mounted on a 3-m mast. Power was provided by a solar panel. Global radiation was recorded at one station (Jomsom). Observations are made with a time step of Dt = 120 s.

The surface stations were positioned in Kagbeni on the roof of a house, in Jomsom close to the airstrip, in Marpha in an open field near the river, and in Tukuche again on the roof of a house (Fig. 1). It would have been preferable to have all stations located close to the axis of the valley. The stations in Marpha and Jomsom satisfied this requirement quite well, and that in Kagbeni came close as the house chosen (Hotel Niligiri) is exposed to the full force of the wind. On the other hand, the village of Tukuche is sheltered to some extent and so were our instruments. Wind speeds at this location underestimate the wind strength found in the riverbed (see KG1). The stations were in operation as follows: kagbeni, 25 February–23 March; Jomsom, 22 February–26 March; Marpha, 23 February–21 March; Tukuche, 23 February–21 March.

c. Pilot balloons

Two theodolites were used to track a helium-filled balloon. The system is identical to that described in KG1. Observations were made in Jomsom, Dhumpha, Chuksang, Tangye, and Lo Manthang (Fig. 1).

3. Surface observations

The surface observations are presented first because they contain detailed information on the diurnal cycle of flow conditions in the valley. This information helps in the interpretation of the soundings. The days of 19 and 20 March are selected to demonstrate the main characteristics of the time series in Kagbeni and Marpha (Figs. 5 and 6). Marpha is located in the narrow part of the valley, which opens toward the Mustang basin just north of Marpha (see Fig. 1). According to Fig. 2, Marpha is situated upstream of the region of flow acceleration while Kagbeni is downstream. The wind speeds recorded in Marpha were moderate, with maximum speeds of ∼7 m s−1. There were weak northeasterlies on 19 March until the Lomar set in and continued until midnight. The next day was essentially calm until 0900 LST when the Lomar set in again. The global radiation data of Jomsom (not shown) indicate that the afternoon of 19 March was partly cloudy. Visibility was unusually low on the morning of 20 March. Clouds covered the sky at ∼1300 LST and rain fell for a few hours. Thus, 20 March is a day with perturbations in the afternoon. Temperature and the wet-bulb temperature were rising relatively late on both days because of the shadow cast by the Annapurna massif. Moisture decreased in the morning of 19 March as is typical at this station. It increased after the onset of the Lomar, which appeared to bring moister air up the valley. However, the difference between temperature and wet-bulb temperature was quite large during the day. Relative humidities were ∼35%.

The diurnal pressure oscillations are represented clearly in the surface pressure time series. According to Dai and Wang (1999) both the diurnal and the semidiurnal tidal oscillation reach maximum values of 0.6–0.8 hPa in Nepal, the semidiurnal one being slightly larger. By and large, our observations (see Table 1) are in agreement with this finding. However, as can seen from Table 1, the amplitude of the diurnal (semidiurnal) pressure oscillation increased from 0.6 (0.8) hPa in Tukuche to 1.2 (1.1) hPa in Kagbeni. This must be a local effect that is presumably related to the strong upvalley winds. In Figs. 5 and 6 there is a pressure maximum about 2 h after sunrise and another one before midnight. The minimum in the afternoon is more pronounced than that early in the day.

The Lomar set in at Kagbeni later on both days and was more powerful with maximum speeds of 18 m s−1. The weak winds in the morning of 20 March were mainly from the south. There was, however, a distinct onset of Lomar on the day as well. Temperatures in Kagbeni were quite similar to those in Marpha. The diurnal and semidiurnal oscillations of temperature were about the same at all stations (Table 1). The evaluation of the related phases shows that temperature maxima occur at about the same time (1310–1325 LST).

The data for Jomsom (not shown) were similar to those in Kagbeni except that the Lomar commences earlier but with reduced intensity. Maximum wind speeds were ∼12 m s−1 on both days. Those in Tukuche were ∼8 m s−1. The increase of the intensity of the valley winds from Tukuche to Kagbeni is also reflected in the diurnal and semidiurnal variations of the wind speed in Table 1.

Both the delay of the onset of the upvalley wind regime in Kagbeni with respect to Marpha and the reduction of the diurnal pressure variation in Marpha with respect to Kagbeni were found almost every day. Table 2 gives the mean delay between all pairs of available stations. Here, we define the onset time as that moment where the upvalley wind reaches half the maximum strength of that day. This definition avoids ambiguities due to the existence of weak upvalley winds before the onset of the strong winds. It is clear from Table 2 that the upvalley wind regime moves from Marpha to Jomsom and farther up to Kagbeni. Typical speeds of propagation are 5 m s−1. The variability of the delays from day to day is large. For example, the maximum delay between Kagbeni and Marpha is 116 min while lomar was recorded a few minutes earlier in Kagbeni on 7 March. Data from Tukuche are not included because of the general weakness of the winds there, which makes it sometimes difficult to determine an onset time.

The winds are driven by pressure gradients. In principle, pressure gradients can be evaluated by reducing all pressure to the height of Jomsom. However, the barometers used at the permanent stations differed in the range of ∼1 hPa. This deviation is too large for a pressure reduction to be useful. Instead we simply argue that the nocturnal winds are extremely weak. Therefore, there is no appreciable pressure gradient force in the Mustang basin at night. The pressure decrease during the day is different at the various stations and the station with the largest decrease is the one with the lowest pressure. The mean value of the pressure decrease from the morning maximum to the minimum in the afternoon is given in Table 3 for each station (first row). The decrease is smallest in Tukuche and largest in Kagbeni. This yields a mean pressure difference between Kagbeni and Jomsom of 0.6 hPa in the afternoon. The mean pressure difference between Marpha and Jomsom is 0.5 hPa. The reference simulation in KG2, where the afternoon pressure minimum is located slightly north of Jomsom, is broadly consistent with this finding.

As has been mentioned, the daily ranges of temperature are about the same in Marpha, Jomsom, and Kagbeni. This implies that the enhanced pressure changes in Kagbeni are due to dynamical processes on top or above the boundary layer. Pressure difference between various stations can be explained by variations of the depth of the upvalley wind layer if there is a pronounced inversion on top of this layer. As will be shown in the next section, such an inversion exists only occasionally. One may, however, argue that the observed widening and intensification of the flow between Marpha and Kagbeni is possible only if air descends from above into the Lomar layer. The related warming and pressure decrease appears to be largest in Kagbeni.

The northward delay of the onset of strong winds as quantified in Table 2 is seen in the diurnal and semidiurnal wind oscillations as well. The wind maximum in Kagbeni as descried by these oscillations lags that in Marpha by 30 min and that in Tukuche by 35 min. The related pressure extrema are delayed in an opposite sense. The pressure maximum in the morning occurs in Kagbeni at 0817 LST, in Jomsom at 0830, in Marpha at 0850, and in Tukuche as late as 0947. The pressure minimum in the afternoon is recorded in Kagbeni at 1525 LST, in Jomsom 1530, in Marpha at 1535, and at 1635 in Tukuche. This shows again that the evolution of pressure gradients between the various stations is hardly linked to the temperature of the Lomar layer. Differential heating from the ground between, say, Tukuche and Kagbeni is presumably not important in generating the strong upvalley winds.

The covariance functions for all time series have been evaluated as well. Of course, the diurnal and semidiurnal signals are dominant. However, long timescales prevail even after these periods are removed. The first zero crossing of autocorrelations of single station data is found in the case for lags of 4–6 h. Delays are the most interesting features to be extracted from these functions. For example, Fig. 7 shows the correlation of temperature and wind in Marpha for positive lags. The wind velocities peak at a lag of about 25 min. Thus, the reaction of the winds in the narrow part of the valley to changes of the temperature is quite fast. On the other hand, the winds have cooling impact on the temperature (not shown). The correlation of temperature and wind in Jomsom reaches a maximum after ∼60 min, that is, considerably later than in Marpha (Fig. 7). This difference may be due to the fact that Marpha is located in a valley where the flow is more constrained.

4. Vertical structure

As has been mentioned, the airplane soundings were performed to explore the stratification of the valley atmosphere during the day and to validate the model results of KG2 as presented in Fig. 2. The model predicts a deep inflow layer for Marpha and shallow and rather stably stratified flow in Jomsom.

By far the largest number of ascents were made at airport in Jomsom (15–17, 19–21, 23–25 March), that is, in the core region of the Kali Gandaki wind system. A selection of these results will be presented first. Parallel ascents upstream of Jomsom were made in Dhumpha (21 March) and Marpha (16 and 17 March). The only soundings in the entrance region took place near Tukuche (23–25 March). Further parallel ascents in Eklobati (see Fig. 1; 19 and 20 March) provide information on the stratification downstream of Jomsom. In addition, the flow in the exit region was investigated in Tangye on 27 and 28 February, in Lo Manthang (3, 4, and 6 March), and in Chuksang (10 and 11 March). It was intended to obtain a reasonably complete dataset of temperature and moisture profiles throughout the Mustang basin during the day by undertaking this rather strenuous part of the campaign.

a. Core region

1) Early morning

Only one ascent was made early in the morning in Jomsom (Fig. 8; 25 March). Flight operations at the airport usually began before 0700 LST and were terminated before noon. Model plane ascents were not permitted during that time. Visual control is impossible before sunrise, leaving little time for early aircraft observations. A ridge was moving toward the Mustag area on that day but gradients at 500 hPa were weak. Light northeasterly winds prevailed throughout the first 2 km (not shown). The morning atmosphere was stably stratified at least up to the maximum ascent height of 1450 m above the ground (N ∼ 1.3 × 10−2 s−1). The air was slightly cooler during descent. The loop during descent close to the ground reflects, of course, a flight maneuvre before landing. The temperature profile in Fig. 8 deviates substantially from those found quite often early in the morning in valleys (Brehm and Freytag 1982; McKee and O'Neal 1989; Whiteman 1990) where an inversion extends from the ground to a height of a few hundred meters. This difference is presumable due to the absence of nocturnal downvalley flow in Jomsom.

The moisture decrease linearly with height. Ascent and descent values are almost identical. The absence of a clearly visible hysteresis effect in the moisture profile suggests that the difference of the potential temperature between ascent and descent are real because the response time of the temperature sensors is shorter than that of the humidity sensors.

2) Upvalley flow

A fairly complete set of ascents has been obtained on 19 March and will be presented in detail below. A warm ridge was centered over Tibet on that day.

The upvalley wind regime was just beginning to establish itself at 1100 LST (Fig. 9a). Rather weak northeasterlies are found up to a height of 2000 m above a shallow layer of upvalley flow. Convection appears to erode the stable layer established during the night so that an almost neutral layer of ∼300 m depth is seen above a shallow superadiabatic layer. Higher up, the stratification is stable (N ∼ 10−2 s−1). Such profiles are common in valleys late in the morning (e.g., Whiteman 1990). Ascent and descent temperatures differ by 1–2 K close to the ground presumably because of additional heating before the start. The moisture is well mixed up to a height of ∼700 m and is constant above z = 1400 m. The θ profile in Eklobati is similar in shape to that in Jomsom but the air was warmer by about 1 K and also drier. One hour later (1200 LST; Fig. 9b), the upvalley wind was quite strong in the lowest 250 m of the valley atmosphere. A layer of weaker upvalley winds extended at least up to a height of 1300 m. The velocity spike at z = 1300 m is presumably real because tracking problems tend to occur only during the first minutes of an ascent when the balloon is close to the observers. This profile of velocities is reminiscent of those reported in KG1 (Figs. 10 and 11 of KG1) for the early stages of the Lomar. In general, ascent temperatures are higher than those found in descent. The systematic temperature differences of more than 2 K as recorded up to heights of 1200 m are so large when compared to our estimate of hysteresis effects that we speculate that these structures are linked to deep eddies. Rapid and strong variations of the vertical motion of the RPV have also been reported by the pilots. The mountains along the valley axis are quite rugged and have heights of 1000–2000 m with respect to the valley bottom. It is conceivable that vortices are generated by the interaction of the upvalley flows with these obstacles. Moisture appears to be affected by the eddies as well. Humidity is higher during descent in contrast to what one would expect for a hysteresis effect. All in all, the onset of the Lomar led to a decrease of the temperature in the lowest 250 m, as one would expect for thermally driven flows. However, the temperature rose in the layer above the strong winds. The pronounced increase of moisture within 1 h up to a height of 700 m must be due to advection.

Given a flow speed of 10 m s−1 it takes just 10 min to advect air from Jomsom to Eklobati. Indeed, temperatures in Eklobati decreased in the lowest 500 m so that Eklobati was no longer warmer than Jomsom. There is some similarity is both profiles, but many details do not coincide. Advective moistening is seen in Eklobati as well.

A well-mixed layer of 1000-m depth was capped by a pronounced inversion at 1300 LST in Jomsom (Fig. 9c). A stable layer was found higher up. The strong winds were confined to the neutral layer. The moisture was well mixed in the neutral layer and the moisture content was strongly reduced above the inversion. As compared to Fig. 9b, the potential temperature is now lower below the inversion and about the same above. The close proximity of ascent and decent in Fig. 9c is surprising given the large flow speeds in the Lomar layer. Moreover, the noisiness of the wind speed profile indicates that the flow was highly turbulent. The θ profile in Eklobati at 1300 LST is quite similar to that in Jomsom but there is warming above the lomar layer. Warming from above is common in valleys during the day (e.g., Brehm and Freytag 1982) and is thought to be due to the downward branch of the cross-valley circulation with slope winds forming the upward branch. Given the large width of the Mustang basin it is doubtful if this explanation is sufficient in this case. It is likely that upvalley flow descending above the Lomar layer contributes to the warming (see also Fig. 2).

The neutral layer in Eklobati was not as deep as in Jomsom. Therefore, we have here an example of a downward sloping inversion where hydraulic modelling may be appropriate. The potential temperature “jumps” by about 3 K on top of the neutral layer so that the flow is supercritical with F ∼ 1.5 in Jomsom. The related pressure difference between Jomsom and Eklobati is 5–7 Pa.

One hour later, the air above 1000 m was cooler and drier in Jomsom and the inversion disappeared (Fig. 9d). There is no obvious reason for this process. It is astonishing that the inversion still existed in Eklobati where a slight cooling aloft was accompanied by a moistening. The q curves in Eklobati differ greatly between ascent and descent. This difference is not due to an observational error. The plane carried two moisture sensors that gave almost exactly the same result.

The Lomar layer reached a depth of at least 1500 m at 1500 LST in Jomsom (not shown). Ascent and descent differ again substantially at that time so that one cannot assign a clear structure to this turbulent boundary layer. Towering cumulonimbus clouds evolved in the afternoon in the south above the mountains bordering the kali Gandaki valley. This feature is typical of radiation days with clear skies in the morning. Like on many other days, the clouds moved northward and showers were recorded in Jomsom. Thus, this last flight of the day documents a late stage of the Lomar where the valley flow was affected by convective cells. The parallel ascent in Eklobati was not successful.

So far we have found that capping inversions may occur but do not appear necessary for the Lomar to develop. The Lomar layer is neutrally stratified in contrast to the model results. Both warming and cooling are seen to occur above the Lomar layer. This suggests that conditions above the Lomar layer proper have an impact on the evolution of the upvalley flows in agreement with the numerical simulations.

On 21 March piloted-balloon (pibal) observations and related airplane ascents were made in Dhumpha at the mouth of the Langpoghyun valley (see also Fig. 1). This valley ascends along the slopes of the Annapurna massif. One expects maximum descent near Dhumpha according to Fig. 2. The sounding at 1100 LST reveals an inversion at z ∼ 1200 m that descends until noon by 200 m due to an increase of the temperature aloft. The parallel soundings in Jomsom do not reveal inversions but the temperature aloft is increasing as well. The flow is neutrally stratified in the Lomar layer.

In 1998, tree deformations were mapped in the Langpoghyun valley (see Fig. 18 of KG1). It was found that strong upvalley flows must prevail in this valley. These winds form a branch of the Kali Gandaki wind system but are oriented almost normal to the standard flow direction of the lomar. On 21 March, such winds were observed. The track of a balloon released at 1026 LST at the axis of this side valley is depicted in Fig. 10. The balloon followed this valley at a velocity of ∼10 m s−1. It should be said, however, that the flow above the balloon in Fig. 10 was southerly. At the moment, the cause of these rapid northeastward accelerations is unknown.

Ascents in Marpha were made in unfavorable conditions. The upper-level flow was southwesterly on both days and appears to have induced strong downward motion above Marpha. It was frustratingly difficult for the pilots to gain height. The ascents are, therefore, not presented.

b. Entrance region

The only soundings in the entrance region of the Lomar were made near Tukuche. Examples are given in Fig. 11. The pre-lomar θ profile at 1130 LST in Tukuche is quite similar to that in Jomsom but temperatures are slightly lower in the lowest 750 m than in Jomsom. Note that the stratification is neutral even before the onset of the Lomar. At 1330 LST the potential temperature in Tukuche is essentially constant in the lowest 500 m. The stratification is weakly stable above. The air is clearly warmer above the Lomar layer in Jomsom than in Tukuche. Moisture is well mixed throughout the ascent in Tukuche but decreases slightly above the Lomar layer in Jomsom. All this indicates that there is descent between Tukuche and Jomsom. It is difficult to compare Figs. 2 and 10 because there are no wind observations available in Tukuche so that the depth of the Lomar layer cannot be estimated. The Brunt–Väisälä frequency is N ∼ 5 × 10−3 s−1 when evaluated for the total profile in Tukuche at 1330 LST. This is less than in Fig. 2. The warming at upper levels in Jomsom with respect to Tukuche is found also in the numerical simulations.

c. Exit region

Soundings in the exit region were carried out in Chuksang, Tangye, and Lo Manthang. Wind intensities in Chuksang were lower than in Jomsom in agreement with what had been found in 1998 (KG1). The Lomar layer was neutrally stratified without a capping inversion. The village of Tangye is located to the east of the Kali Gandaki River. An escarpment overlooking the main valley was selected for the observations. Until that time, no flow data had been collected in the eastern part of the upper Mustang basin. Observations were made on 27 and 28 February under a prevailing southwesterly upper-level flow ahead of a trough at 500 hPa. Upvalley winds were of moderate strength on 27 February. The next day extremely strong upvalley winds were observed in the presence of strong convective activity in the south. Figure 12 shows a situation where at least the lower part of the Lomar layer is well mixed with maximum velocities of ∼20 m s−1. This establishes that vivid upvalley winds occur in the eastern part of the Mustang basin. The balloons could not be tracked for long enough nor did the planes reach sufficiently large altitudes as needed for an estimate of the depth of the lomar layer. The dryness of the flow indicates that the air, say, in Kagbeni, is not simply advected up to Tangye. Dry air from higher levels must be mixed into the Lomar flow.

It was hoped that strong katabatic winds or cold air outflows from the Tibetan Plateau would be centered in Lo Manthang. Conditions late in winter tend to be favorable for such flows. However, the observations were disappointing in that respect. The morning observations of 3, 4, and 6 March revealed that the stratification was quite stable close to the ground (N ∼ 0.015 s−1). On 3 March upvalley winds were quite vigorous and the stratification was neutral at least up to heights of ∼1000 m. No inversion was found. An interesting situation occurred on 6 March with a deep layer of stably stratified northerlies on a rather cold morning. The 500-hPa map shows that a small ridge moved over the area. The stable layer disappeared quite rapidly (ascent at 0945 LST; not shown) but is was not before late in the afternoon that southerlies were observed. The Lomar layer of a depth of ∼1250 m was well mixed with maximum flow speeds of 8 m s−1. No inversion was detected. The northerly adverse flow conditions were almost too dominant for the Lomar to reach Lo Manthang on that day.

All in all, it became clear that the basic characteristics of the Lomar layer, namely neutral stratification and the absence of an inversion, are found also in the upper part of the Mustang basin.

5. Discussion

We have to conclude on the basis of the soundings that the strong winds in the Kali Gandaki valley cannot be explained on the basis of hydraulic flow theory, and we have, therefore, to look for other ways to understand the observations. As a first step let us fit equations to the surface data in order to obtain information on the dynamics of the Lomar. For example, we may select the stations Kagbeni and Marpha to adopt the equation of motion along the valley axis to the available surface data of wind and pressure. With
i1520-0493-130-8-2042-e51
(t = nDt; Dt = 120 s; subscripts k, m for Kagbeni and Marpha; p, pressure; u, along-valley wind component; u > 0 for upvalley flow), we have a simple equation to be tested that asserts that changes of the along-valley wind speed uk in Kagbeni are caused by the pressure gradient force estimated by using the pressure data from Kagbeni and Marpha. There is also a damping term. It is understood in (5.1) that all terms are deviations from the time mean. Nickus and Vergeiner (1984) applied this equation to observations in the Inn valley. Of course, (5.1) is a rather simple approximation to the full equation of motion:
i1520-0493-130-8-2042-e52
where y is the cross-valley coordinate; υ, the cross-valley flow; w, the vertical velocity; and f, the Coriolis parameter. When using (5.1), we assume that cross-valley advection, vertical advection, and rotation effects are negligible. This omission is presumably justified with respect to rotation and cross-valley winds. It is less obvious if vertical advection is unimportant. Unfortunately, no data are available to check this assumption. In principle, a term ∼[(unk)2 − (unm)2] could have been added on the right of (5.1) in order to include along-valley advection. It would have been inconsistent, however, to include horizontal advection when vertical advection is excluded.
We wish to introduce another more empirical equation in combination with (5.1). As outlined above, there is some evidence that descent and the related warming aloft cause the pressure to fall deeper in Kagbeni in the afternoon than in Jomsom or Marpha (see Table 2). This descent is presumably linked to the acceleration and widening of the Lomar jet toward the north. If so, the wind velocity in Kagbeni provides a gross measure of this effect. We assume that
i1520-0493-130-8-2042-e53
captures this situation. Strong upvalley winds would lead to an increase of the pressure difference pmpk, provided γ > 0. A damping term is added on the right. Note, that (5.3) is counterintuitive. One could argue that strong winds advect mass to Kagbeni. This would reduce the pressure difference and, then, γ should be negative. Of course, (5.1) and (5.3) are simply the equations of a regressive model of first order relating the variables uk and (pmpk) to their changes in time.

The coefficients αδ in (5.1) and (5.3) are determined by a least square fit using all observations. The resulting coefficients are given in Table 4 for three pairs of stations. With α = 0.32 × 10−4 (m s−1 Pa−1) we obtain about half the correct value ρ−1o Dx−1km ∼ 0.6 × 10−4 for the pair Kagbeni–Marpha where Dxkm is the distance of the villages. This means that the true accelerations due to the pressure gradients are partly canceled by an effect that is not represented properly in (5.1). Advection of momentum from above is presumably a good candidate. The damping term β is negative and implies a damping time |β|−1 ∼ 40 min. Nickus and Vergeiner (1984) obtained a damping time of 30 min for the Inn valley.

The coefficient γ is positive with γ ∼ 6 × 10−5 (Pa m−1). The pressure difference between Marpha and Kagbeni increases by about 2 Pa within 1 h if the anomalous wind speed is 10 m s−1. This is a relatively weak effect. The damping is quite strong with a damping time of |δ|−1 ∼ 20 min. The system (5.1) and (5.3) is, of course, stable in the sense that any initial perturbation would be damped out in an integration given the coefficients in Table 4. Without damping, however, the system is unstable with an e-folding time of (αγ)−1/2 = 6.4 h. The related growth is slow but not negligible.

It is surprising that α is smaller for the Kagbeni–Jomsom pair than for Kagbeni–Marpha despite the fact that Jomsom is closer to Kagbeni than Marpha. The cancellation of the accelerations due to the pressure gradient by unrepresented effects must even be larger in Jomsom than in Kagbeni. The damping time is longer and γ has just half the value of that of the Kagbeni–Marpha pair. On the other hand, the coefficients attain the largest values for the Jomsom–Marpha pair. The e-folding time (αγ)−1/2 is just 3.6 h in this case. This indicates that the Jomsom–Marpha section is the most important one for the evolution of the upvalley flow. The interaction of the Lomar with the layers above is strongest there. However, this interaction extends clearly to Kagbeni. This conclusion is supported by the frequent observations of warming above the Lomar layer.

We wish to address here one further issue that came up quite often during the presentation of the soundings. There is good evidence that the strong Lomar flows are turbulent. Profiles during ascent and descent differ sometimes so strongly that it is quite unlikely that this difference is due to the choice of the flight track (see Figs. 9 and 11). As has been mentioned, this turbulence may be generated by the orographic obstacles to the flow. There is, however, also the possibility that the configuration of Lomar is inherently unstable. We may perceive the flow in an idealized manner as a two-layer system where the strong winds are confined to a mostly neutral layer. The air is stably stratified above this layer and almost at rest. Sometimes there is an inversion, more often there is none. We have here, therefore, a variation of the classic Kelvin–Helmholtz instability problem (e.g., Drazin and Reid 1981). As outlined in the appendix, the stability in the upper layer is only weakly damping. The discontinuity of the upvalley wind at the upper boundary of the Lomar layer supports instability at all wavelengths L = 2π/k (k is the along-valley wavenumber) provided an inversion is absent. However, growth rates increase ∼k so that short waves are preferred. It is conceivable that such instabilities play a role in generating the deviations in Figs. 9 and 11.

Finally, let us comment on the relationship of the numerical simulations of KG2 to the observations. Many aspects of the wind profiles obtained in our campaign are compatible with those presented in KG2. The onset of the upvalley winds close to the surface and the following buildup of the upvalley wind layer as found in KG2 were observed this time, as in 1998 (KG1). Moreover, the model results agree with the observations of strong upvalley winds in Tangye (see Fig. 6 of KG1) and Lo Manthang. It is, however, our impression that the turbulence parameterization in MM5 underestimates the generation of turbulence in strong flows as in Fig. 2. It is presumably for that reason that the model predicts a stably stratified Lomar layer in contrast to the observations. Pronounced gravity wave features have been found in KG2 that where generated by the ridges near Marpha (see Fig. 8 of KG2). It was suggested in KG2 that these waves are important for the generation of the strong upvalley winds. This speculation can be ruled out on the basis of our observations. Internal gravity waves cannot be excited in neutrally stratified flow.

6. Concluding remarks

The following summarizing statements can be made on the basis of our observations. The period of weak winds extends normally from late in the evening till 0900–1100 LST. Although weak downvalley flow prevails during that time, weak upvalley flow occurs quite often. The rather limited number of soundings in the morning suggests that the atmosphere is stably stratified at sunrise but that there is no inversion. After sunrise a convective layer of a few hundred meters depth evolves with a shallow superadiabatic layer at the ground.

The upvalley winds set in close to the ground. A well-mixed upvalley wind layer is established within about an hour with a depth of 1000–1500 m. Sometimes an inversion is found on top of this layer, but more often the neutral layer is simply topped by a stable layer. The fully developed upvalley flow is presumably unstable. These flows are found at all locations from Tukuche up to Lo Manthang.

The transition to the upvalley regime occurs first in Marpha and Tukuche and moves with speeds of ∼5 m s−1 upward to Kagbeni and farther on to Lo Manthang.

As outlined in KG1 and KG2, the upvalley winds are generated primarily by the heating of the Mustang basin before noon. Low pressure is established there with respect to the atmosphere to the south of Lete at the same height as the basin. The flow driven by this pressure gradient has to pass the narrow part of the valley where the Kali Gandaki River cuts through the Himalayas. The dynamically most interesting part of the flow is that near and to the northeast of Marpha where the highest wind speeds are found. The suggestion of KG2, that some kind of supercriticality is involved in generating these high velocities, is not supported by the soundings. However, the soundings support and even extend the suggestion of KG2 that descent above the Lomar is an important part of the dynamics of the Kali Gandaki wind system. Although the soundings provided information on the stratification of these upper layers, the wind observations via pibals are too inaccurate at large heights to tell us much about the flow conditions up there. Therefore, important aspects of the dynamics of the Kali Gandaki wind regimes have to await further clarification.

It is an important side result of the recent field campaign in the Kali Gandaki valley that soundings up to heights of ∼2000 m above the ground can be made by use of RPV even under extreme conditions. The vertical resolution of the resulting profiles is quite good as is the quality of the observations. This sounding system is quite mobile and vertical profiles can be obtained almost anywhere. It is a drawback of this method that highly skilled pilots are needed at least under the extreme conditions of the Kali Gandaki valley.

Acknowledgments

The campaign could not have been conducted without the financial support by Deutsche Forschungsgemeinschaft. A great number of people supported the project in many ways. We wish to express our gratitude to Robbe Modellsport for help with respect to the remote control, to Hacker Antriebstechnik for help with respect to the plane engines, to Ingenieurbüro Würtenberger for support with respect to the sensors, to Zeiss for its generosity with respect to the development of the optical control systems, to Simprop Electronic for advice on the implementation of the motor, to H. Müller of Modellbau Ulrich and Blue Airlines for help in building the plane, to K. Budion for advice with respect to batteries, and to Optik Schadow. We are grateful to G. Zängl for comments and to L. Gantner for providing information on the synoptic situation. The comments by the referees helped substantially to improve the presentations of the results.

REFERENCES

  • Arakawa, S., 1969: Climatological and dynamical studies on the local strong winds, mainly in Hokkaido. Japan Geophys. Mag., 34 , 349425.

    • Search Google Scholar
    • Export Citation
  • Armi, L., 1986: The hydraulics of two flowing layers of different densities. J. Fluid Mech., 163 , 2758.

  • Armi, L., and R. Williams, 1993: The hydraulics of a stratified fluid flowing through a contraction. J. Fluid Mech., 251 , 355375.

  • Baines, P., 1995: Topographic Effects in Stratified Flows. Cambridge Monographs on Mechanics, Cambridge University Press, 482 pp.

  • Barros, A., M. Joshi, J. Putkonen, and D. Burbank, 2000: A study of the 1999 monsoon rainfall in a mountainous region in central Nepal using TRMM products and rain gauge observations. Geophys. Res. Lett., 27 , 36833686.

    • Search Google Scholar
    • Export Citation
  • Brehm, M., and C. Freytag, 1982: Erosion of the night-time thermal circulation in an Alpine valley. Arch. Meteor. Geophys. Bioklimatol., B31 , 331352.

    • Search Google Scholar
    • Export Citation
  • Chilson, Ph, P. Johansson, M. Johnsson, R. Moses, J. Stanojev, Th Hedquist, A. Niva, and R. Scheifele, 1999: Ripan: A remotely controlled aircraft project for tropospheric and stratosphere research. Proc. 14th ESA Symp. on European Rocket and Balloon Programs, Potsdam, Germany, ESA SP-437, 111–116.

    • Search Google Scholar
    • Export Citation
  • Clements, W., J. Archuleta, and P. Gudiksen, 1989: Experimental designs of the 1984 ASCOT field study. J. Appl. Meteor., 28 , 405413.

  • Dai, A., and J. Wang, 1999: Diurnal and semidiurnal tides in global surface pressure fields. J. Atmos. Sci., 56 , 38743891.

  • Drazin, Ph, and W. Reid, 1981: Hydrodynamic Stability. Cambridge Monographs on Mechanics and Applied Mathematics, Cambridge University Press, 525 pp.

    • Search Google Scholar
    • Export Citation
  • Egger, J., S. Bajrachaya, U. Egger, R. Heinrich, J. Reuder, P. Shakya, H. Wendt, and V. Wirth, 2000: Diurnal winds in the Himalayan Kali Gandaki valley. Part I: Observations. Mon. Wea. Rev., 128 , 11061122.

    • Search Google Scholar
    • Export Citation
  • Jackson, P., and D. Steyn, 1994a: Gap winds in a fjord. Part I: Observations and numerical simulation. Mon. Wea. Rev., 122 , 26452665.

    • Search Google Scholar
    • Export Citation
  • Jackson, P., . 1994b: Gap winds in a fjord. Part II. Hydraulic analog simulations. Mon. Wea. Rev., 122 , 26662676.

  • Konrad, T., M. Hill, J. Rowland, and J. Meyer, 1970: A small radio-controlled aircraft as a platform for meteorological sensors. APL Tech. Dig., 10 , 1119.

    • Search Google Scholar
    • Export Citation
  • McKee, Th, and R. O'Neil, 1989: The role of valley geometry and energy budget in the formation of valley wind. J. Appl. Meteor., 28 , 445456.

    • Search Google Scholar
    • Export Citation
  • Nickus, U., and I. Vergeiner, 1984: The thermal structure of the Inn valley atmosphere. Arch. Meteor. Geophys. Bioklimatol., A33 , 199215.

    • Search Google Scholar
    • Export Citation
  • Pan, F., and R. Smith, 1999: Gap winds and wakes: SAR observations and numerical simulations. J. Atmos. Sci., 56 , 905923.

  • Pettre, P., 1982: On the problem of violent valley winds. J. Atmos. Sci., 39 , 542554.

  • Ramage, C., 1971: Monsoon Meteorology. International Geophysics Series, Vol. 15, Academic Press, 296 pp.

  • Renno, N., and E. Williams, 1995: Quasi-Lagrangian measurements in convective boundary layer plumes and their implications for the calculation of CAPE. Mon. Wea. Rev., 123 , 27332742.

    • Search Google Scholar
    • Export Citation
  • Stephens, G., and Coauthors. . 2000: The Department of Energy's Atmospheric Radiation Measurement (ARM) unmanned aerospace vehicle (UAV) program. Bull. Amer. Meteor. Soc., 81 , 29152937.

    • Search Google Scholar
    • Export Citation
  • Whiteman, C., 1990: Observations of thermally developed wind systems in mountainous terrain. Atmospheric Processes over Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 5–42.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., J. Egger, and V. Wirth, 2001: Diurnal winds in the Himalayan Kali Gandaki valley. Part II: Modeling. Mon. Wea. Rev., 129 , 10621078.

    • Search Google Scholar
    • Export Citation

APPENDIX

Instability of Two-Layer Flow

We consider a two-layer atmosphere. The lower layer of depth H represents the lomar layer with constant mean wind U1 and vanishing Brunt–Väisälä frequency N21 = 0. A second layer with vanishing wind U2 = 0 and stable stratification N22 > 0 is assumed to extend to infinity above this layer. Both layers are separated by a material surface. There is no density jump. We assume perturbations of the form exp[ik(xct)] and solve the well-known wave equation
i1520-0493-130-8-2042-ea1
[i = 1, 2; ŵi(z) Fourier coefficient of vertical velocity], under the boundary conditions ŵ1 = 0 at z = 0, 1 = 2 at z = H, and ŵ2 = 0 at infinity. Moreover, the kinematic condition
ŵ1cU1ŵ2c
relates the vertical velocities at z = H. With ŵ1 ∼ sinh(kz) and ŵ2 ∼ exp(−nz), one obtains the polynomial
i1520-0493-130-8-2042-ea3
There are two real roots. One of them violates the upper boundary condition and must be excluded. The other one is cU1/2 for L ≤ 4 H. The complex pair is cU1(1 ± i)/2 except for L > 4 H, where |c| is somewhat smaller. In essence we recover the result of the Kelvin–Helmholtz problem.

Fig. 1.
Fig. 1.

Map of the Kali Gandaki valley: airplane ascents, dots; permanent surface stations, crosses with circles; villages and sites mentioned in the text, crosses. Pibal ascents were made in Jomsom, Dhumpha, Chuksang, Tangye, and Lo Manthang. LK is Langpoghyun Kola. Height contours, solid (m); contour interval is 500 m. Horizontal distances as indicated at the axes (km). The map is based on topographic data with a resolution of 30″ × 30″. These data have been interpolated to a 1 km × 1 km grid. See also Fig. 1 of KG1. Dashed, Kali Gandaki and LK

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 2.
Fig. 2.

Isentropes (contour interval, 1 K; solid) and wind (vectors) in a section along the Kali Gandaki valley as obtained in the reference run REF of KG2 in the afternoon (t = 15 h). Shading: light, wind speeds 10–15 m s−1; medium, 15–20 m s−1; dark, >20 m s−1. See also Fig. 7a of KG2. The bold letters mark the locations of Tukuche, Marpha, Jomsom, and Kagbeni. Height is above MSL. The narrowest point of the valley is located near Marpha

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 3.
Fig. 3.

Monthly mean values of the hourly mean wind speed U (m s−1) as observed in Kagbeni in Feb–Mar 1990 at a height of 9 m

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 4.
Fig. 4.

(a) Prototype Kali. (b) The pilot is wearing the special binoculars needed to follow the plane at heights of more than 1000 m above the ground

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 5.
Fig. 5.

Time series of wind speed υ (m s−1) (bold) and direction (dots), temperature T (°C) and wet-bulb temperature Tw (°C) (dashed), specific humidity q (g kg−1), and surface pressure p (hPa) in Marpha for 19 and 20 Mar

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 6.
Fig. 6.

The same as Fig. 5 but for Kagbeni

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 7.
Fig. 7.

Correlation of the temperature and the wind velocity in Marpha (bold) and Kagbeni (dashed) for positive lags; i.e., wind perturbations are shifted by τ (h) with respect to temperature. Diurnal and semidiurnal cycles are removed

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 8.
Fig. 8.

Potential temperature θ (K) as a function of height during ascent (solid) and descent (broken) at 0620 LST 25 Mar in Jomsom. Also given is the specific humidity q (g kg−1) during ascent and descent (triangles). The vertical coordinate z is defined such that z = 0 at Jomsom airport at 2751 m above MSL

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 9.
Fig. 9.

Potential temperature θ (K) and specific humidity q (g kg−1) as a function of height during ascent and descent on 19 Mar in Jomsom and Eklobati. Also given is the wind speed (bold) and the wind direction (dots) as obtained at the base in Jomsom. As can be seen from Fig. 1 upvalley winds occur for 180° ≤ dir ≤ 240°: (a) 1100, (b) 1200, (c) 1300, and (d) 1400 LST. Wind velocities too noisy to be shown. Jomsom (Eklobati) is at 2751 m (2810 m) above MSL. Jomsom airport: z = 0 m

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 9.
Fig. 10.
Fig. 10.

Track (dots) of a balloon released at 1026 LST 21 Mar near Dhumpha. Bold shows height lines (m); T1T2 shows the baseline. The numbers along the track give the height of the balloon above the starting position; y (x) axis pointing toward the north (east)

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 11.
Fig. 11.

Potential temperature θ (K) and specific humidity q (g kg−1) during ascent (bold) and descent (broken) in Jomsom and Tukuche on 24 Mar. Also given are wind speed (bold) and direction (crosses) as observed at the Jomsom baseline: (a) 1130 and (b) 1330 LST. Jomsom (Tukuche start) is at 2751 m (2670 m) above MSL. Jomsom airport: z = 0 m

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Fig. 12.
Fig. 12.

Potential temperature and specific humidity during ascent (bold) and descent (broken) at 1200 LST 28 Feb in Tangye (3600 m above MSL). Also given are wind speed and direction; z = 0 at the observing site

Citation: Monthly Weather Review 130, 8; 10.1175/1520-0493(2002)130<2042:DWITHK>2.0.CO;2

Table 1.

Amplitudes of the diurnal (first entry) and semidiurnal (second entry) oscillations of pressure, temperature, and wind velocity at the four permanent stations

Table 1.
Table 2.

Mean delay (min) of the onset of the valley wind regime at the station in the left column with respect to the station in the top row. In parentheses, number of days with negative delays, total number of days, and propagation velocity in m s−1

Table 2.
Table 3.

Mean ratio of the pressure decrease between the maximum in the morning and the minimum in the afternoon at the station in the left column and that in the upper row. The second entry in the first row is the mean pressure decrease (hPa) at the station indicated on top

Table 3.
Table 4.

Coefficients α, β, γ, and δ as determined from the station data by a least square fit of (5.1) and (5.3) for the station pair given in the first row

Table 4.
Save
  • Arakawa, S., 1969: Climatological and dynamical studies on the local strong winds, mainly in Hokkaido. Japan Geophys. Mag., 34 , 349425.

    • Search Google Scholar
    • Export Citation
  • Armi, L., 1986: The hydraulics of two flowing layers of different densities. J. Fluid Mech., 163 , 2758.

  • Armi, L., and R. Williams, 1993: The hydraulics of a stratified fluid flowing through a contraction. J. Fluid Mech., 251 , 355375.

  • Baines, P., 1995: Topographic Effects in Stratified Flows. Cambridge Monographs on Mechanics, Cambridge University Press, 482 pp.

  • Barros, A., M. Joshi, J. Putkonen, and D. Burbank, 2000: A study of the 1999 monsoon rainfall in a mountainous region in central Nepal using TRMM products and rain gauge observations. Geophys. Res. Lett., 27 , 36833686.

    • Search Google Scholar
    • Export Citation
  • Brehm, M., and C. Freytag, 1982: Erosion of the night-time thermal circulation in an Alpine valley. Arch. Meteor. Geophys. Bioklimatol., B31 , 331352.

    • Search Google Scholar
    • Export Citation
  • Chilson, Ph, P. Johansson, M. Johnsson, R. Moses, J. Stanojev, Th Hedquist, A. Niva, and R. Scheifele, 1999: Ripan: A remotely controlled aircraft project for tropospheric and stratosphere research. Proc. 14th ESA Symp. on European Rocket and Balloon Programs, Potsdam, Germany, ESA SP-437, 111–116.

    • Search Google Scholar
    • Export Citation
  • Clements, W., J. Archuleta, and P. Gudiksen, 1989: Experimental designs of the 1984 ASCOT field study. J. Appl. Meteor., 28 , 405413.

  • Dai, A., and J. Wang, 1999: Diurnal and semidiurnal tides in global surface pressure fields. J. Atmos. Sci., 56 , 38743891.

  • Drazin, Ph, and W. Reid, 1981: Hydrodynamic Stability. Cambridge Monographs on Mechanics and Applied Mathematics, Cambridge University Press, 525 pp.

    • Search Google Scholar
    • Export Citation
  • Egger, J., S. Bajrachaya, U. Egger, R. Heinrich, J. Reuder, P. Shakya, H. Wendt, and V. Wirth, 2000: Diurnal winds in the Himalayan Kali Gandaki valley. Part I: Observations. Mon. Wea. Rev., 128 , 11061122.

    • Search Google Scholar
    • Export Citation
  • Jackson, P., and D. Steyn, 1994a: Gap winds in a fjord. Part I: Observations and numerical simulation. Mon. Wea. Rev., 122 , 26452665.

    • Search Google Scholar
    • Export Citation
  • Jackson, P., . 1994b: Gap winds in a fjord. Part II. Hydraulic analog simulations. Mon. Wea. Rev., 122 , 26662676.

  • Konrad, T., M. Hill, J. Rowland, and J. Meyer, 1970: A small radio-controlled aircraft as a platform for meteorological sensors. APL Tech. Dig., 10 , 1119.

    • Search Google Scholar
    • Export Citation
  • McKee, Th, and R. O'Neil, 1989: The role of valley geometry and energy budget in the formation of valley wind. J. Appl. Meteor., 28 , 445456.

    • Search Google Scholar
    • Export Citation
  • Nickus, U., and I. Vergeiner, 1984: The thermal structure of the Inn valley atmosphere. Arch. Meteor. Geophys. Bioklimatol., A33 , 199215.

    • Search Google Scholar
    • Export Citation
  • Pan, F., and R. Smith, 1999: Gap winds and wakes: SAR observations and numerical simulations. J. Atmos. Sci., 56 , 905923.

  • Pettre, P., 1982: On the problem of violent valley winds. J. Atmos. Sci., 39 , 542554.

  • Ramage, C., 1971: Monsoon Meteorology. International Geophysics Series, Vol. 15, Academic Press, 296 pp.

  • Renno, N., and E. Williams, 1995: Quasi-Lagrangian measurements in convective boundary layer plumes and their implications for the calculation of CAPE. Mon. Wea. Rev., 123 , 27332742.

    • Search Google Scholar
    • Export Citation
  • Stephens, G., and Coauthors. . 2000: The Department of Energy's Atmospheric Radiation Measurement (ARM) unmanned aerospace vehicle (UAV) program. Bull. Amer. Meteor. Soc., 81 , 29152937.

    • Search Google Scholar
    • Export Citation
  • Whiteman, C., 1990: Observations of thermally developed wind systems in mountainous terrain. Atmospheric Processes over Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 5–42.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., J. Egger, and V. Wirth, 2001: Diurnal winds in the Himalayan Kali Gandaki valley. Part II: Modeling. Mon. Wea. Rev., 129 , 10621078.

    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Map of the Kali Gandaki valley: airplane ascents, dots; permanent surface stations, crosses with circles; villages and sites mentioned in the text, crosses. Pibal ascents were made in Jomsom, Dhumpha, Chuksang, Tangye, and Lo Manthang. LK is Langpoghyun Kola. Height contours, solid (m); contour interval is 500 m. Horizontal distances as indicated at the axes (km). The map is based on topographic data with a resolution of 30″ × 30″. These data have been interpolated to a 1 km × 1 km grid. See also Fig. 1 of KG1. Dashed, Kali Gandaki and LK

  • Fig. 2.

    Isentropes (contour interval, 1 K; solid) and wind (vectors) in a section along the Kali Gandaki valley as obtained in the reference run REF of KG2 in the afternoon (t = 15 h). Shading: light, wind speeds 10–15 m s−1; medium, 15–20 m s−1; dark, >20 m s−1. See also Fig. 7a of KG2. The bold letters mark the locations of Tukuche, Marpha, Jomsom, and Kagbeni. Height is above MSL. The narrowest point of the valley is located near Marpha

  • Fig. 3.

    Monthly mean values of the hourly mean wind speed U (m s−1) as observed in Kagbeni in Feb–Mar 1990 at a height of 9 m

  • Fig. 4.

    (a) Prototype Kali. (b) The pilot is wearing the special binoculars needed to follow the plane at heights of more than 1000 m above the ground

  • Fig. 5.

    Time series of wind speed υ (m s−1) (bold) and direction (dots), temperature T (°C) and wet-bulb temperature Tw (°C) (dashed), specific humidity q (g kg−1), and surface pressure p (hPa) in Marpha for 19 and 20 Mar

  • Fig. 6.

    The same as Fig. 5 but for Kagbeni

  • Fig. 7.

    Correlation of the temperature and the wind velocity in Marpha (bold) and Kagbeni (dashed) for positive lags; i.e., wind perturbations are shifted by τ (h) with respect to temperature. Diurnal and semidiurnal cycles are removed

  • Fig. 8.

    Potential temperature θ (K) as a function of height during ascent (solid) and descent (broken) at 0620 LST 25 Mar in Jomsom. Also given is the specific humidity q (g kg−1) during ascent and descent (triangles). The vertical coordinate z is defined such that z = 0 at Jomsom airport at 2751 m above MSL

  • Fig. 9.

    Potential temperature θ (K) and specific humidity q (g kg−1) as a function of height during ascent and descent on 19 Mar in Jomsom and Eklobati. Also given is the wind speed (bold) and the wind direction (dots) as obtained at the base in Jomsom. As can be seen from Fig. 1 upvalley winds occur for 180° ≤ dir ≤ 240°: (a) 1100, (b) 1200, (c) 1300, and (d) 1400 LST. Wind velocities too noisy to be shown. Jomsom (Eklobati) is at 2751 m (2810 m) above MSL. Jomsom airport: z = 0 m

  • Fig. 9.

    (Continued)

  • Fig. 10.

    Track (dots) of a balloon released at 1026 LST 21 Mar near Dhumpha. Bold shows height lines (m); T1T2 shows the baseline. The numbers along the track give the height of the balloon above the starting position; y (x) axis pointing toward the north (east)

  • Fig. 11.

    Potential temperature θ (K) and specific humidity q (g kg−1) during ascent (bold) and descent (broken) in Jomsom and Tukuche on 24 Mar. Also given are wind speed (bold) and direction (crosses) as observed at the Jomsom baseline: (a) 1130 and (b) 1330 LST. Jomsom (Tukuche start) is at 2751 m (2670 m) above MSL. Jomsom airport: z = 0 m

  • Fig. 12.

    Potential temperature and specific humidity during ascent (bold) and descent (broken) at 1200 LST 28 Feb in Tangye (3600 m above MSL). Also given are wind speed and direction; z = 0 at the observing site

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 280 73 1
PDF Downloads 189 51 1