Boundary Layer Characteristics over a High Altitude Station , Mauna Loa Observatory

The unique boundary layer at Mauna Loa Observatory (3396 meters) is examined with a combination of radiosondes launched from the observatory and a novel aerosol profiling technique called CLidar or camera lidar. This boundary layer is influenced by a combination of radiation winds, due to the heating and cooling of the surrounding lava, and off-island winds. Typically an upslope surface wind forms after sunrise as the ground heats up. The reverse occurs after sunset as the ground cools and a temperature inversion, tens of meters thick forms. Aerosol increases for the first 90 to 160 meters and then decreases to free tropospheric levels. The 90 to 160 m aerosol peak indicates the vicinity of the upslope/downslope interface in the air flow. An upper transition is seen in the aerosol gradient at about 600 meters above the observatory (4000 m Above Sea Level). This transition is also seen in radiosonde potential temperature data. The sondes indicate that the air above the nighttime downslope surface region usually has an upslope component. Some of this counter-flowing air can be entrained in the downslope air, possibly influencing the sampling of aerosols and trace gases at the observatory.


INTRODUCTION
Mauna Loa Observatory (MLO) is operated by the National Oceanic and Atmospheric Administration (NOAA) for atmospheric monitoring of gases, aerosols, solar radiation, and other climate forcing quantities.MLO is located on the northern side of the active Mauna Loa volcano at an altitude of 3400 meters.The summit of Mauna Loa is at 4170 meters and the slope at MLO is about 7%.To the North is the steeper Mauna Kea, a dormant volcano with a summit of 4205 meters.The two mountains form a "saddle" between them and the height of the saddle point is about 2000 meters, Fig. 1.
Although MLO is usually above the planetary boundary layer (PBL), especially at night, there is a unique flow of air flowing up and down the slope.A couple of hours after sunrise the dark black and brown lava is solar-heated enough to create a flow of air up the slope of the mountain.This flow carries local, more humid, island-influenced air to MLO.After sundown the lava rapidly cools and the reverse happens; the surface flow is down the slope of the mountain.Many atmospheric measurements are taken at MLO. Depending on the measurement, some data are only taken during this downslope period to avoid local, island influences.
This flow pattern, known as the radiation wind, was identified soon after the establishment of MLO in 1956 by surface wind measurements (Mendonca and Iwoka, 1969).Tethered balloon flights (Mendonca, 1969) on three days first probed the winds above MLO and found a counterflowing region of air above the nighttime downslope flow.This limited experiment found a 55 m thick downslope region and an upper region with upslope flow extending to at least 600 m.An 11-day study using several surface sites situated at different altitudes (Garrett, 1980) described other aspects of the flow, and suggested a model of the circulation showing these counter-flowing regions.With the construction of the 40 meter tower at MLO, further near-surface studies were possible (Herbert et al., 1994) which showed that the nighttime temperature inversion, formed as the lava cooled, did not extend much beyond the height of the tower.
The boundary conditions at MLO are produced by an interaction of the free-tropospheric (FT) flow interacting with the island, termed barrier wind, and the radiation winds.If the FT (off-island) winds are light, the radiation wind will dominate at MLO.If off-island winds are strong, the radiation wind effect may be undetectable.Ryan (1997) used the tower measurements with in-situ ozone, and radiosonde and ozonesonde flights from Hilo, Hawaii to describe the influence of the prevailing FT wind on this upslope/downslope phenomenon.Enough data were available to divide the wind into sectors and produce correlations between the MLO winds and the FT winds.One implication of the correlations was that there are times, even at night, when the source of the air at MLO (3400 m) is coming from lower altitudes.
In this study a novel aerosol profiling technique is used to directly observe the boundary layer above MLO from the surface to the free troposphere during the upslope/downslope transition after sunset by using aerosols as tracers.The technique, called CLidar or camera lidar (Barnes et al., 2003(Barnes et al., , 2007;;Sharma et al., 2011;Barnes and Sharma., 2012), was developed at MLO and has been regularly used since 2005.The altitude resolution is sub-meter near the surface and resolves the details of this boundary layer very well.The CLidar sensitivity for measuring aerosols easily distinguishes the very clean free-tropospheric air from the relatively-clean upslope air.In this study additional instruments are used in conjunction with CLidar data to more fully characterize the prevailing atmospheric conditions.These include balloonborne radiosondes launched from Hilo Hawaii and from MLO measuring wind direction, temperature, pressure and relative humidity, traditional lidar measuring water vapor, and ground-based and/or tower-based meteorological instruments at MLO measuring wind direction, relative humidity, temperature, and total aerosol light scattering.In the following Data section these data sets are described in detail.Next, the various measurements are combined to describe the behavior of the boundary layer over the observatory in the Discussion.Finally, the most important results are stated in the Conclusion.

Meteorological
MLO has a 40 meter tower instrumented for temperature measurements at 2, 10 and 38 meters above the ground.
The wind is measured at 10 and 38 meters and humidity is measured at 2 meters.The database used includes hourly averages from 1977 through 2014.Details can be found at the MLO web site (GMD, 2015).The NOAA National Weather Service flies radiosondes from Hilo, Hawaii (55 km away at sea level) twice a day at 2:00 and 14:00 Hawaii Standard Time (HST).These radiosonde times are generally when the day/night, upslope/downslope flows at MLO are well established.Planetary Boundary Layer (PBL) heights have been determined by finding peaks in the derivative of the potential temperature (Siedel et al., 2010).The PBL heights shown in Fig. 1 where determined from radiosondes from 2005 to 2014 selected from days the lidar and CLidar were operated.
The MLO daily temperature cycle, Fig. 2, is very similar in summer and winter although slightly warmer in the summer and with a longer daytime.The temperature at 38 meters is far more stable than the 2 meter temperature.This depicts the very strong heating influence of the bright sun and dark lava.The nighttime temperature inversion is easily seen even though the difference in sensors is only 36 meters.There is an asymmetry with the sunset cooling taking about twice as long as the sunrise warming.A limited number of radiosonde flights directly from MLO showed the temperature profile generally converged with the FT radiosonde profiles, within a 200 m or less of the ground.

Surface Aerosol
Total scatter (m -1 ), absorption (m -1 ), and condensation nuclei (# cm -3 ) are measured from a 10 meter intake (GMD, 2015).Hourly averages of total scatter at 500 nm are shown in Fig. 4 for the year 2013 which was fairly typical.The higher total scatter in April is a result of the well known springtime Asian dust transport (Bodhaine et al., 1980) which normally occurs in March, April, and May.The sampling system uses both 10 micron and 1 micron impactors which alternate every 6 minutes.The data shown are for the 10 micron impactor.The nighttime total scatter is very low at MLO.The yearly average is about 4 Mm -1 for most of the night, but on many occasions the total scatter is below 1 Mm -1 .These conditions have been quite useful in developing the CLidar technique since nearly the all the light is scattered by air molecules.

CLidar and Lidar
The CLidar system at MLO is a bistatic lidar system which uses a Nd:YAG GCR-6 laser (330 mJ per pulse, 30 Hz, 532 nm) as its light source, and a CCD camera equipped with wide angle optics and 10 nm laser line interference filter as its detector.The laser light was perpendicularly polarized to the camera.The polarization is an important parameter in the CLidar analysis (Barnes, 2003(Barnes, , 2007)), The system is limited to near dark and dark operating conditions.The CLidar images the 532 nm laser beam from the NOAA/MLO aerosol/water vapor/temperature lidar (Barnes and Hofmann, 1997).The CLidar camera is located 139 meters upslope from the lidar which is inside the main station building.The original system operated from 2005 to 2010 which was then replaced by a newer camera and fisheye lens increasing the signal by a factor of 50.The aerosol profile for 2013/7/3 was acquired with a smaller 0.29 Watt continuous laser.This laser light was circularly polarized for the observation.The older system (Barnes et al., 2003) integrated the signal for 5.5 minutes, and the newer system for 3 minutes.The entire laser beam is imaged from the side simultaneously, from ground to zenith.The scattering altitude is determined from the geometry of the laser and detector locations.The scattering angle ranges from near 90 degrees at ground level to 180 degrees at zenith.The side scattered intensity at each pixel (altitude) is then used to derive aerosol extinction.The signal is normalized to match molecular scattering in a high altitude fitting range, as is typically done with backscatter lidar, to remove the molecular scattering component.Then an aerosol scattering phase function (showing the fraction of scattering into each angle) must be used to iteratively correct for transmission and produce a profile of the ratio of aerosol to molecular scatter.With the aerosol phase function this ratio can be converted to total aerosol scatter and, assuming a single scattering albedo, to total extinction.
The aerosol phase function could be measured directly at ground level with a polar nephelometer if one were available.The aerosol phase functions used for the evening CLidar runs were derived from daytime sunphotometer measurements made with Aerosol Robotic Network (AERONET) instruments at MLO (Holben et al., 1998).Since the photometers measure column aerosol properties, the phase functions inverted from them can be considered as representative of an average phase function over altitudes.If AERONET data are not available for a particular day, modeled phase functions for various aerosol classes used by the NASA space lidar, CALIOP, can be used (Omar et al., 2009).A benefit of the CLidar system is its ability to measure aerosols all the way to ground level with high altitude resolution.The CLidar instrument fills an important niche at MLO where direct sampling instruments are often limited to the height of the instrumented 40 m tower and remote aerosol measurements such as the NOAA/MLO lidar are limited to 6000 m and higher.
An example of a CLidar aerosol profile from the newer system is shown in Fig. 5.The fundamental single-angle scatter measured has been converted into extinction by using the CALIOP Polluted Continental aerosol phase function.This function closely approximates the average AERONETinverted aerosol phase functions that have been acquired at MLO.There was an unusual amount of structure in the profile on this night which persisted for about 20 minutes.Layers are resolved which are only 10s of meters thick.A clear top to the layer is seen at 900 meters and there is a large decrease between 100 and 200 meters.

DISCUSSION
A key concept in understanding the unique MLO boundary layer is the interaction of the free tropospheric (FT) offisland wind and the radiation wind.A simple correlation of the standard 700 hPa wind speed from the radiosondes (3156 m average), with the speed at the same time of day at MLO, shows a slope of about 0.75, meaning the mountain barrier generally reduces the FT wind.An estimate of the radiation wind speed can be seen in the correlation when the FT winds are limited to very weak speeds of 1 or 2 m s -1 .In this case the non zero radiation wind is dominating at MLO.The average speed in this case is about 2.85 ± 1.0 m s -1 in the daytime, with the nighttime wind being slightly lower, 2.66 m s -1 .These are similar to results from Ryan (1997).
In Fig. 6 an average of CLidar aerosol profiles is shown for eight evenings with intermediate FT winds averaging 4.74 m s -1 .The MLO wind at twilight averaged 4.47 m s -1 coming downslope, so the nighttime radiation wind was already established.On each of the evenings between 30 and 40 individual profiles were acquired.The natural variation of the eight profiles dominates the standard deviation; the typical error in the aerosol scatter measurement was about 5%.The single-angle scatter profiles were converted to extinction in most cases using AERONET-inverted aerosol phases.The CALIOP Polluted Continental function was used when AERONET was not available.When AERONET data were used the single scattering albedo assumed was 0.972, and when CALIOP modeled phase functions were employed the SSA from the employed phase function was used.
A doubling of the aerosol is seen in the first 100 meters followed by a consistent decrease until FT typical values below 1 Mm -1 are reached.This reversal of the aerosol gradient indicates a transition in the flow which was seen in an early study using a tethered balloon (Mendonca, 1969).The balloon was flown over three separate autumn days when the FT winds were light (2.54 m s -1 average) and indicated a transition from upslope to downslope winds at 55 m.The altitude of the balloon was limited to about 600 m above the ground.The lidar water vapor is an average over several years of lidar observations.It also shows a low altitude peak which is higher than 100 m aerosol peak.But the altitude resolution (300 m) is quite coarse and does not constrain the altitude of the peak well.
The CLidar average extinction profile shows a change in gradient at about 600 m altitude.Eight out of eleven of the radiosondes launched from MLO (on different days) also show a peak in the gradient of the potential temperature calculated from their data at approximately 600 m.This Layer (PBL) height using the peak in the derivative of the potential temperature.The Hilo 14:00 sonde showed a PBL height of 3500 m, which is above the altitude of MLO.The next day Hilo 2:00 sonde yielded a PBL height of 2700 m.MLO sonde calculations show a peak in the potential temperature gradient at 4000 ± 20 m above sea level (ASL) for the first three MLO sonde data times.The 2:27 Nx/Dy flight did not show a peak.These three match well with the sharp drop off in the 19:22 and 21:20 MLO sonde RH data at 4000 m.The data demonstrate that the RH can transition rapidly within a very narrow altitude range of 50 to 100 m.Thus to characterize the conditions above MLO instrumentation with good altitude resolution is desirable.The twice daily Hilo radiosonde data also do not fully capture the time varying nature of the RH and boundary layer.Flying additional sondes from MLO offer more data but there are cost and staff limitations for the number of possible flights.An instrument which offers more continuous data on atmospheric conditions at and above the station would be useful.The CLidar instrument, which has both excellent near ground altitude resolution and the ability to monitor conditions over continuous segments of time was used to provide additional insights in subsequent studies.show the winds at MLO were (atypically) downslope before sunset, and also downslope (as typical) in the hours that followed sunset.Yet it can be seen from Fig. 9 that very high (for MLO) values of aerosol extinction, above 30 Mm -1 , persisted for hours past sunset.
Hence the air at MLO was not clean despite the downslope direction.As described in Ryan (1997) the downslope radiation wind at the surface of the mountain has an upslope flow above it and there is expected to be some region of turnover where these mix.In this case it is likely that the upslope portion of the wind entrained the humid local air from just below the observatory and as the flow overturned this air was carried to the observatory in the downslope flow, yielding the high aerosol values observed with the CLidar.Interestingly, the CLidar data also show increasing aerosol from the MLO ground level to 160 m above the observatory.Possibly the local air mixing with the cleaner downslope air is gradually diluted by free tropospheric air at higher altitudes.An example of conditions on an evening where the winds at MLO were in the expected upslope direction before sunset and in the typical downslope direction after sunset is given in the data of 7/3/2013 shown in Figs. 10 and 11.Note the extinction values in Fig. 11 are much lower than in Fig. 9.These were more stable conditions and in Fig. 10 the wind measurements from the sondes launched from MLO agree with the Hilo sonde wind values above 4400 m.The post-sunset (21:09) MLO sonde shows upslope winds at altitudes above 300 m (3700 ASL) above MLO.
The CLidar data this evening were taken while testing a smaller, 0.29 W continuous laser which the reduced signal and resulted in noisier profiles.They show increasing aerosol  from the surface to an altitude of about 90 m.The aerosol persists at diminished intensity to an altitude of 1000 m above the station.There is a very thin layer of high altitude cloud present at 3800 m above MLO (7200 m ASL).Even when wind patterns follow the traditional upslope to downslope transition in the evening and no unusual weather events are present, measurable low level aerosol can be present at MLO, especially in the lowest few hundred meters of atmosphere.

CONCLUSIONS
Mauna Loa Observatory (MLO) is situated above the average daytime and nighttime marine PBL, and the nighttime air is typically quite clean.The boundary layer above MLO is unique and influenced by a combination of off-island (free tropospheric) winds and radiation winds (from heating and cooling).Typically in evenings the radiation wind transitions from daytime upslope to nighttime downslope creating a temperature inversion that usually is only a few 10s of meters thick.Two transition altitudes can usually be seen in the aerosol profiles measured by the CLidar.The aerosol increases from the ground to about 90-160 m, and then decreases to free tropospheric values.The 90-160 m aerosol peak indicates the vicinity of the upslope/downslope interface in the air flow.The rate of decrease in aerosol extinction changes near 600 m above MLO (4000 m ASL).This transition is also seen in potential temperature and humidity measurements of sondes launched from MLO, indicating the transition is from a convectively less stable region below to a more stable region above.
The high altitude resolution and near-ground capabilities of the CLidar technique are particularly useful in this application since the boundary layer is so close to the ground.The aerosol transitions help illuminate transitions in flow that are otherwise difficult to measure.Downslope winds are not a guarantee of entirely clean free tropospheric air at MLO, but the very low aerosol levels measured at the surface and aloft constrain local contamination.It would be very informative to combine the CLidar with a type of wind profiler, which will hopefully be done in a future study.

Fig. 1 .
Fig. 1.Side view of the Big Island of Hawaii along a line including the summits of Mauna Loa and Mauna Kea.Typical daytime "radiation" wind directions are shown on the left while typical nighttime "radiation" wind directions are shown on right.Average boundary layer heights over the ocean are indicated by the dashed lines.

Fig. 2 .Fig. 3 .
Fig. 2. Average temperature measurements for a winter and a summer month at MLO.The tower has temperature sensors at 2, 10, and 38 meters.The periods of twilight are also shown.Following sunset it is dark enough to start operating the CLidar roughly halfway through the twilight period.

Fig. 8 Fig. 7 .Fig. 8 .
Fig. 7. Hilo radiosonde, MLO tower meteorology, and MLO launched radiosonde data for GPS sondes launched from MLO, 2005/06/15.Times are HST.a) Wind direction is shown.Zero degrees is North looking towards Mauna Kea.-90 to 0 to 90 winds have an upslope component (Convention: angle is direction wind is coming from).b) Relative Humidity data are shown.