The planetary boundary layer (PBL) height is a key parameter in air quality
control and pollutant dispersion. The PBL height cannot, however, be directly
measured, and its estimation relies on the analysis of the vertical profiles
of the temperature, turbulence or the atmospheric composition. An
operational PBL height detection method including several remote sensing
instruments (wind profiler, Raman lidar, microwave radiometer) and several
algorithms (Parcel and bulk Richardson number methods, surface-based
temperature inversion, aerosol or humidity gradient analysis) was developed
and tested with 1 year of measurements, which allows the methods to be
validated against radio sounding measurements. The microwave radiometer
provides convective boundary layer heights in good agreement with the radio
sounding (RS) (median bias < 25 m,
The height of the planetary (or atmospheric) boundary layer (PBL) is a key parameter for air quality analysis, pollutants dispersion and quantification of pollutant emissions and sources. The PBL controls the interactions of the atmosphere with the oceans and land and determines the air volume available for the dispersion of all atmospheric constituents, including anthropogenic pollution and water vapor, emitted at the Earth's surface. Hence the PBL contributes to the assessment of the pollutant concentration near the surface and the PBL height is thus a key parameter of all air pollution models. Despite its critical importance, the PBL cannot be directly measured but has to be estimated by upper-air instruments.
The COST (European Cooperation in Science and Technology) action 710 (Harmonisation of the pre-processing of meteorological data for atmospheric dispersion models) defined the daytime PBL height as “the height of the layer adjacent to the ground over which pollutants or any constituents emitted within this layer or entrained into it become vertically dispersed by convection or mechanical turbulence within a time scale of about an hour” (COST action 710 – Final report, 1998). The PBL height can consequently be estimated by the measurement of mechanical turbulence, of the temperature enabling convection or of the concentration of PBL constituents. These detection methods are based on various atmospheric parameters, various measuring instruments and different analysis algorithms, leading to several PBL height estimations that are not always consistent with each other. The first measurements of the PBL were performed using surface and tower observations of vertical wind profiles and deeply investigating wind turbulence. The intense development of remote sensing instruments nowadays offers a wide field of vertical profiles up to several kilometers, which allows PBL height detection from the surface with high temporal resolution.
Diurnal cycle of the PBL height over land for a clear convective day (adapted from Stull, 1988).
The PBL experiences a marked diurnal cycle that depends on both the synoptic and local weather conditions. In the case of fair weather days, the PBL height has a well-defined structure and diurnal cycle (Fig. 1), leading to the development of a convective boundary layer (CBL), also called a mixing layer, during the day and of a stable boundary layer (SBL), which is capped by a residual layer (RL) during the night (Stull, 1988). In the case of cloudy or rainy conditions and in the case of advective weather conditions, free convection is no longer driven primarily by solar heating, but by ground thermal inertia, cold air advection, forced mechanical convection and/or cloud top radiative cooling. In those cloudy cases, the CBL development remains weaker than in the case of clear sky conditions, with slower growth and lower maximum height. The boundary layer is said to be neutral if the buoyancy is near zero; these neutral cases are found for overcast conditions with strong winds but little temperature differences between the air and the ground. Neutral conditions are frequently met in the RL but rarely near the surface. The PBL development under clear sky conditions (i.e., more than 50 % of solar radiation during the CBL development) that leads to strong convection driven principally by solar heating will be called CBL. For all “no clear-sky” cases with partial or total cloud coverage but without precipitation, the PBL will be called cloudy-CBL.
While the definition and the measurement of the CBL, the neutral boundary
layer (NBL) and the cloudy-CBL are well established, the nocturnal SBL
presents a more complicated internal structure. It is comprised of a stable
layer caused by radiative cooling from the ground, which gradually merges
into a neutral layer called the RL (Stull, 1988; Salmond and McKendry, 2005;
Mahrt et al., 1998). The stable layer can be characterized by a
surface-based temperature inversion (SBI), and its top can be estimated by
the height at which the gradient of the potential temperature (
Contrary to radio sounding (RS), launched usually only twice a day, continuous remote sensing measurements allow the determination of the diurnal cycle of the different layers constituting the PBL. The use of remote sensing instrumentation to detect the PBL height was recently summarized by Emeis (2009). Recent studies compared several detection methods or retrieval techniques (Bianco and Wilczac, 2002; Seidel et al., 2010; Beyrich and Leps, 2012; Haeffelin et al., 2012; Summa et al., 2013), remote sensing with RS measurements (Baars et al., 2008, Liu and Liang 2010, Granados-Muñoz et al., 2012; Milroy et al., 2012; Sawyer and Li, 2013; Cimini et al., 2013) and/or several remote sensing instruments (Wang et al., 2012; Zahng et al., 2012). In most of these studies, good correlations are found in the case of strong or weak convective weather conditions with differences of 100–300 m between the various instruments and/or methods. Non-convective weather conditions corresponding in most of the cases to cloudy and rainy conditions lead to much greater discrepancies in the PBL height estimations. In these cases, the difference becomes even greater if the methods/instruments are designed to detect various types of PBL such as CBL, NBL or RL. If temperature profiles are measured, bulk Richardson number (bR) or Parcel (PM) methods are usually considered as the most relevant methods for daytime PBL height detection. Some studies also compared measurements with models predictions (Baars et al., 2008; Seidel et al., 2012; Ketterer et al., 2014), the results depending on both the model and the measurement type.
Climatologies of PBL height have been performed on time series from 1 to 25
years long in Europe and the United States (Baars et al., 2008; Schmid and
Niyogy, 2012; Beyrich and Leps, 2012; Granados-Muñoz et al., 2012;
Sawyer and Li, 2013) and over continents (Seidel et al., 2010, 2012). For continental stations, a clear CBL seasonal cycle is usually
found with a maximum height reaching 1000 to 2000 m above ground level
(a.g.l.) in summer and and a minimum height reaching 500 to 1200 m a.g.l in
winter. The seasonal cycle of the nocturnal SBL was only addressed on the
basis of temperature (
In this study, an operational system for PBL height detection has been
developed based on the analysis of vertical atmospheric profiles of
List of abbreviations.
This paper gives a description of the instruments and the methods used to
derive the PBL heights, some examples of PBL height estimations, the
inter-comparison and validation of the experimental methods, a comparison
with COSMO-2 model and a 2-year climatology. Recommendations about the
most comprehensive set of instruments for an operational detection of the
PBL diurnal cycle are given in the conclusion. Abbreviations for the sites,
instruments and methods as well as for the different PBL layers are
summarized in Table 1. Throughout the paper, the various PBL detections are named
by the measuring instrument and the applied method, RS/PM being for example
given for Parcel method (PM) applied to RS measurements. If not specified,
elevations or heights are given as above ground level (a.g.l.) and time as
LT (equal to UTC
For this study, a 2-year (2012–2013) data set from the two upper-air remote
sensing sites Payerne (491 m a.s.l., 46.799
The WPs are Degreane PCL1300 (Degreane Horizon, 2006) with five antennas
operating at 1290 MHz (
The MWR is a passive remote sensing instrument that measures electromagnetic
radiation emitted from the atmosphere in the microwave band. From the
measured radiation spectrum, the atmospheric
The PAY aerological station is equipped with a fully automated and operational Raman lidar designed for continuous measurements of tropospheric water vapor, aerosols and temperature in dry conditions (Dinoev et al., 2013). The transmitter is a Nd:YAG-laser emitting UV pulses (300 mJ per pulse, 30 Hz repetition rate) at a wavelength of 355 nm. The receiver consists of four telescopes of 0.3 m diameter each, which are fiber coupled to the polychromator, which spectrally separates the backscattered light. Separate photomultipliers simultaneously detect vibrational Raman scatter from nitrogen (387 nm) and water vapor (407 nm) signals, two portions of the pure rotational Raman spectrum and the elastic backscatter. The aerosol scattering ratio is then derived from the sum of the rotational Raman signals and the elastic signal (Dinoev et al., 2010). The maximum range varies from 4000 m during the day up to 8000 m during the night for the water vapor measurements and from 7000 m (day) to 12 000 m (night) for aerosol backscatter ratio measurements. The first range level is located at 110 m. The vertical resolution is dynamically adapted to the measurement conditions, varying from 30 m near the surface to a maximum of 300 m in the upper troposphere. However, the signal-to-noise ratio is very high in the boundary layer and the vertical resolution remains constant (30 m). The effective time resolution of profiles is 30 min. No measurements are possible during precipitation and in the presence of low clouds, i.e., the lidar powers down if the clouds are below 900 m or there is precipitation and powers up as soon as the cloud base rises above 2000 m and there is no precipitation.
A ceilometer (CBME80 from Eliasson) measuring at
In addition to the remote sensing instruments, the Payerne station performs
routine RS providing pressure (
The COSMO-2 model (
The SwissMetNet meteorological surface network provides surface
The cloud cover is detected by an automatic partial cloud amount detection
algorithm (APCADA) that estimates in realtime the sky cloud cover from
surface-based measurements of long-wave downward radiation,
Measurements from both the MWR and lidar are necessary to calculate the
virtual potential temperature (
PBL detection methods based on
The Parcel method (Holzworth, 1964; Fisher et al., 1998) defines the PBL
height as the elevation to which an air parcel with ambient surface
The bulk Richardson number (
For both PM and bR methods, the surface
Detection of the SBL from RS
The nocturnal SBL can only be detected by the
Upper panel: automatic detection of PBL height from all remote
sensing instruments, RS and COSMO-2 model for a convective day in summer
2012 at PAY; the background signal corresponds to the
lidar/ASR. Lower panel: sunshine duration, vertical heat flux and temporal
gradient of surface
The radar echo measured by the WP is generated by inhomogeneities in the
refractive index, which are characterized by the structure constant
The aerosol scattering ratio (ASR) is the ratio between the total and the molecular backscatter coefficients. Since the PBL top is characterized by a sharp decrease in concentration of all pollutants, the absolute minima in the vertical gradient of the lidar/ASR and of the RS/RH profiles can be associated with the CBL height during day and to the RL during night. A continuity algorithm similar to the WP/SNR method (see Sect. 2.2.2) was applied, with the modified condition that the local minimum has to be lower than 10 % of the absolute minimum. According to the WP/SNR method, the uncertainty is considered equal to the FWHM of the selected peak in the lidar/ASR gradient profile and is on the order of 100–250 m.
Both the PM and bR methods can be applied not only to
The operational procedure calculates the PBL heights each hour. Examples of
the resulting plots are presented in Figs. 4 and 5. All PBL heights from the
various instruments and methods are plotted on the lidar/ASR (Fig. 4) or on
the WP/SNR (Fig. 5) in the upper panel, whereas the vertical heat flux, the
sunshine duration and the temporal gradient of the surface
Example of cloudy-CBL detection under cloudy conditions in winter (14 February 2013) plotted on WP/SNR signal as background. For symbol description see Fig. 4.
Linear regression of PBL height detected
by
The first example of a clear CBL height diurnal cycle (Fig. 4) was measured
at PAY during a clear-sky convective day on the 23 July 2012, where all
principal PBL features of Fig. 1 were measured:
The layered structure of the nocturnal PBL between midnight and the sunrise:
(1) the SBI is detected by both RS (dark blue triangles) and MWR (reversed dark blue triangles)
at about 100 m, (2) the SBL detected by MWR/bR (white squares) and the top of the stable layer
detected by the MWR/SBLpT (magenta triangles) peak both at the same altitude of 500 m until 03:00
and decreases to about 200 m thereafter, (3) the SBL detected by the COSMO-2/bR (orange diamonds) stays
constant at 250 m until sunrise, (4) the RL is detected by both the WP/SNR (light blue circles) and the
lidar/ASR (green circles) at 1500 m, the WP catching another turbulent layer at 700–800 m between 03:00
and 09:00 corresponding to a jet of northeasterly wind (15 m s The CBL development from sunrise to mid-afternoon: (1) one hour after sunrise, the
CBL height increase is very well caught by all the methods based on The nocturnal SBL development: after 18:00, the bR method continues to follow the CBL
decrease whereas the development of the nocturnal SBL can be detected by the MWR/SBI and MWR/SBLpT methods.
The second example of a winter day (Fig. 5) presents a stable cloud cover at
800–1200 m. The uniformity of the clouds is evident in the ceilometer
measurements. In this cloudy-CBL case, only the PBL height detection methods
based on
Boxplots of PBL height differences
Linear regression of PBL height computed
with various methods and instruments as a function of
The inter-comparison and validation process was performed at PAY on a set of
119 clear-sky convective days, representing one-third of the total measured days
in 2012. This means that the CBL pattern was clearly recognizable and that
at least half of the solar radiation was measured from sunrise to 13:00.
Hence, the presence of some clouds is not excluded. RS/PM at 12:00 was
chosen as the reference method for the validation due to the availability
and reliability of RS Due to the use of the same RS data with a very good vertical resolution, the
RS/bR and RS/RH gradient methods are the closest to RS/PM, with regression slopes near
1, coefficients of determination ( The MWR results are somewhat more scattered, but with very small median bias
(< 25 m) and interquartile ranges (100 m). The MWR/PM has the smallest
interquartile ranges and whiskers size due to the same applied detection method that,
contrary to bR, do not use the WP wind velocity. The WP/SNR method has the lowest correlation coefficients (0.49), the largest
median bias ( The comparison with lidar/ASR can only be done on a reduced data set (61 cases)
due to its lower data availability. Taking into account the very different detection
methods based, respectively, on
Since the CBL may not always be at its maxima at 12:00, an inter-comparison
on the same set of 119 convective days was performed with MWR/PM as
reference for the 12:00–15:00 time interval corresponding to CBL height
maxima for all seasons (Fig. 7). Similarly to the 12:00 case, the difference
between PM and bR is rather small, with interquartile ranges of 5 and 71 m and whiskers far below 200 m. The lidar/ASR also shows a very good agreement
with a median bias of 20 m and an interquartile range of about
Same as Fig. 6 but between 12:00 and 15:00 UT and with MWR/PM taken as the reference.
Example of CBL overestimation by COSMO-2/bR, the background signal corresponds to the WP/SNR. For a description of the symbols, see Fig. 4.
Each of the considered methods and instruments has their own uncertainties in
PBL height detection. The uncertainty minimum is usually obtained for
fully developed CBL reported in Figs. 6–7. Several type of uncertainties can
however be estimated. First, a statistical uncertainty (see for example the
climatology analyses Figs. 9–12) estimates the fluctuations of measurements
for cases that are considered as similar; these fluctuations reflect the
measurement uncertainties and illustrate the variation of the
atmosphere for “similar conditions”, but are unable to detect systematic
bias. A measuring uncertainty can in addition be derived for each instrument
providing an estimation of systematic bias and fluctuations; such analyses
have been up to now only partially made for some instruments, but not all,
impeding our ability to propagate these errors on the various PBL height
detection methods. Finally, the comparison to a reference (Figs. 6–7) allows one to statistically estimate the reliability of the other methods. The
uncertainties bounded to the methods and the instruments (see Sect. 2.2)
provide however a similar picture as the inter-comparison, with the greatest
precision for methods based on
Upper panel: CBL height 2-year climatology at PAY (left) and SHA (right). The symbols are the monthly median of the daily medians of the CBL height taken between 12:00 and 15:00; the error bars are the 25th and 75th percentiles. Lower panel: the number of CBL days used for calculating the monthly medians are given in grey for MWR/PM, WP/SNR and COSMO-2/bR, in black for MWR/bR and in green for lidar/ASR.
Finally, in addition to considering the differences in statistical and intrinsic uncertainties found between the various instruments and methods, one has to consider that the measured parameter (PBL height) is in reality not a fixed point but rather a transition layer between two atmospheric states. Both Stull (1988) and Garratt (1992) estimated the thickness of the entrainment zone as large as half the mixed layer depth. This transition layer reaches therefore between some tens to some hundreds of meters. Moreover, the remote sensing instruments measure an air volume with a thickness corresponding to the instrument level (see Sect 2.2) and not a precise point. The obtained differences between the experimental methods and their uncertainties remain on the same order of magnitude of this transition layer thickness.
Upper panel: cloudy-CBL height climatology at PAY (left) and SHA (right). Lower panel: number of cloudy-CBL days used to calculate the monthly medians. Symbols and colors as in Fig. 9.
The comparison of the COSMO-2/bR PBL heights to the references for the same
119 cases described in Sect. 3.3 (Table 3 and Figs. 6 and 7) shows that
the PBL heights calculated by COSMO-2/bR have a positive bias compared to
the measured PBL heights. The median biases are 275 m and 299 m when
compared to the RS/PM (12:00) and to the MWR/PM (12:00 to 15:00),
respectively. The interquartile ranges reach 200 to 350 m, and the maximal
whiskers are higher than 1000 m. A detailed analysis of the individual
determinations (see Fig. 8 for example) reveals that COSMO-2/bR often
overestimates the PBL height during the whole day and tends to show a too-rapid PBL growth in the morning. This behavior is not limited to clear-sky
convective days and is observed throughout the year. This significant
positive bias compared to all experimental methods and the asymmetry of the
distribution, which is seen on the histograms (Figs. 6 and 7), may have
several explanations:
Contrary to all the experimental methods, COSMO-2/bR determines the PBL
height from the The use of the bR method induces a positive bias compared to the PM
method, but the difference does not exceed some tens of meters as
demonstrated by the RS and MWR results. The bR method is very sensitive to the surface The occurrence of clouds, which may be missing in the model, can
temporarily reduce the surface heating and thus, the convection of air
masses. Therefore the occurrence of clouds can lead to a lower measured PBL
height.
Further studies are necessary to assess the details of how these various parameters cause the PBL height overestimated in COSMO-2/bR, but our results demonstrate the model consistently overestimates the PBL height.
Upper panel: SBL and RL heights for clear-sky conditions at PAY (left) and SHA (right). Lower panel: number of cloudy-CBL days used to calculate the monthly medians for each method, the colors correspond to the upper panel ones. Symbols and colors as in Fig. 9.
The 2-year climatology of CBL heights calculated from all instruments and
COSMO-2 is presented in Fig. 9 for PAY (256 days) and SHA (289 days). It has
to be noted that the same subset of days was taken for the MWR, the WP and
COSMO-2, whereas the lower availability of lidar/ASR data and to a lesser
extent of MWR/bR leads to a smaller data set that is still useful for
comparison with the CBL heights estimated from the other instruments. The
CBL heights have an annual cycle with a minima at 300–700 m in winter and a
maxima at 1200–1500 m during the May–August period. It has to be noted that
the CBL extremes occur at the solstices and not at the
The systematic overestimation of the COSMO-2/bR model observed at both
stations presents a clear annual cycle with a winter minimum and a summer
maximum that can reach 500–700 m. At PAY and to a lesser extent at SHA, the
WP/SNR and lidar/ASR detect a higher CBL (300–500 m) than the MWR/PM and
MWR/bR in winter. This difference is probably related to meteorological
conditions with high-altitude
Upper panel: SBL and RL heights for cloudy conditions at PAY (left) and SHA (right). Lower panel: number of cloudy-CBL days used to calculate the monthly medians, the colors correspond to the upper panel ones. Symbols and colors as in Fig. 9.
The CBL maxima measured over the Swiss plateau are similar to the PBL
heights maxima measured over Europe by RS (Seidel et al., 2012; Beyrich and
Leps, 2012), but lower than the lidar-measured PBL height over Leipzig
(Baars et al., 2008) and the PBL height detected by several methods (RS, MWR
and lidar) over Granada (Granados-Muños et al., 2012). The higher PBL
height over both regions can be explained for Leipzig by its lower altitude
(135 m a.s.l.), its northerly latitude leading to longer summer days and
similar annual
Cloudy-CBL cases have been selected as non-CBL days without rain between
06:00 and 15:00 and correspond to various meteorological conditions (e.g.,
high-altitude clouds, fog, advection, mixed conditions). As expected by its
more heterogeneous atmospheric structure, the cloudy-CBL climatology (271
days at PAY and 223 at SHA) presents more scattered results with larger
quartiles (Fig. 10) than the CBL one. The cloudy-CBL annual cycle based on
the
The SBL climatology was divided into clear-sky (Fig. 11) and cloudy nights
(Fig. 12) in order to differentiate cases with high and low radiative
cooling. Clear-sky (186 at PAY and 163 at SHA) and cloudy nights (126 at PAY
and 151 at SHA) were selected with the criteria of no precipitation between
00:00 and 05:00 in addition to 0–2 and 7–8 octa of the sky covered by clouds
estimated by APCADA, respectively. While some features of the SBL annual
behavior can be deduced, the low number of cases for some months,
particularly for cloudy conditions, does not allow us to draw strict
conclusions on the effective seasonal cycle of the different layers forming
the SBL. The following points can however be inferred:
During clear-sky nights, the complete SBL structure can be clearly observed at
PAY with SBI heights being between 100 and 500 m during the whole year, SBLpT being
lower than 500 m in winter and rising up to 800 m during the other seasons. The RL measured
by the lidar/ASR has a seasonal cycle completely similar to the CBL one (see Fig. 9), so that
the pollutant emitted during the preceding days remain at the altitude of the CBL maxima during
the night. During clear-sky nights, the WP/SNR method, which is more frequently subjected to false
attribution than the other methods (see Sect. 2.2.2), leads to much more scattered results
and large quartiles. The WP/SNR results are however comparable to the RL heights measured by the
lidar/ASR. During cloudy nights, the ground The COSMO-2/bR frequently computes SBL height lower than 50 m that can hardly represent
a real physical PBL height. These false estimations are due to a stable The MWR/bR method gives results usually similar to SBI in the case of clear-sky but clearly
higher in the case of cloudy nights. This difference is probably due to the direct dependence of SBI
height on the ground radiative cooling, whereas the bR method is more affected by wind turbulence
and katabatic jets that are not discriminated by the cloud amount.
Few SBL climatologies have been yet published probably due to the greater
complexity of PBL heights detection during night than during day. Cimini et
al. (2013) found MWR/SBL height lower than 500 m near Paris during the
March–August period that are comparable to our climatology over the Swiss
plateau. Martucci et al. (2007) found nighttime RL heights detected by
lidar/ASR between 500 and 1500 m in Neuchâtel (Switzerland), similar to
our results. Additionally, Beyrich and Leps (2012) and Seidel et al. (2010)
studied the 10-year climatology of PBL height detected by RS measurements
(twice a day). The SBL seasonal cycles over Europe were found to depend on
the method applied to the RS profiles: the PM method leads to almost
constant SBL during the whole year, whereas SBI has a seasonal minima in
summer and a maxima in winter. Unfortunately, our 2-year data set restricted
by the cloud coverage is not large enough to compare our SBL seasonal cycles
with these results. Finally, similarly to our results, the gradient method
applied to the RH or specific humidity profiles is maximal during summer and
minimal during winter. As expected, they also found that SBI yields the
smallest heights, followed by the PM method, while the humidity and the ASR
profiles similarly lead to much greater heights, corresponding to RL top.
Advantages and limits of detection methods and instruments to estimate the PBL height.
The difficulty of the PBL height detection comes first from the complexity of the troposphere itself, which can be composed of several layers with different thermal structures, wind regimes and concentrations of atmospheric constituents. Secondly, each detection method has good performances only for defined PBL structures and under specific meteorological conditions. Only the combination of several methods and instruments allows one to follow the complete diurnal cycle of the complex PBL layered structure.
For this study a system for automatic realtime detection of the PBL height based on several methods applied to various remote sensing observations was implemented and operated for 2 years (2012–2013) for two upper-air stations on the Swiss plateau to quantify the advantages and disadvantages of several techniques. The numerical weather prediction model COSMO-2/bR PBL height was also compared to the experimentally determined PBL heights. Relative RS/PM at 12:00 or the MWR/PM between 12:00 and 15:00 as a reference, the remote sensing and model results were then validated on a subset of 119 convective days. A 2-year climatology for daytime and nighttime PBL heights was calculated for convective days and clear-sky nights, as well as for cloudy convective days and nights without precipitation. The system for automatic detection of the PBL height is now implemented in an operational environment and the data are visualized and provided to end users in realtime.
The advantages and limitations of each detection/measurement method as an operational mode are summarized in Table 4. The greatest advantage of PBL detection by the various profiles measured by RS is its very good measurement precision and vertical resolution. Its temporal resolution (two measurements per day), however, does not provide the PBL diurnal cycle.
The MWR provides The PBL increases after sunrise, reaching its maximal elevation at the
beginning of the afternoon and then decreasing as soon as the vertical heat
flux vanishes after sunset. The SBI development and maximal height from sunset to sunrise that
corresponds to the layer in which the pollutants emitted during the night
are trapped. MWR/SBLpT measures the top of the nocturnal stable layer. MWR is
therefore able to detect the daytime and nighttime layers in which ground-emitted atmospheric constituents are trapped, but not the RL corresponding
to the air volume trapping the atmospheric constituents emitted some
hours/days before.
The Raman lidar has a higher vertical resolution than MWR but its data
availability is restricted by fog, low cloud coverage and precipitation. The
profiles of the aerosol or the humidity concentrations allow one to measure the
dynamics of atmospheric constituents and are consequently a direct
determination of the pollutant dispersion in the PBL. The comparison with
RS/PM and MWR/PM proves that the lidar/ASR is able to detect the CBL maxima
during the afternoon with a good precision and also sometimes part of the
CBL formation. During night, this method provides the RL height and can
therefore be considered as complementary to the MWR methods.
The comparison of WP/SNR with RS/PM and MWR/PM shows that, in most cases, the CBL maximum is well detected by WP, but with a lower precision and a greater amount of outliers. De facto WP/SNR maxima can be generated by turbulence at the PBL top, but also at cloud top or at wind shears. An operational PBL height measurement by WP is therefore much more difficult to implement without a human visual control to attribute the SNR maxima to the real atmospheric phenomena. In the case of cloudy condition, the WP/SNR tends to measure the cloud top instead of the PBL height, which could be exploited for other applications. For this study, the WP and the Raman lidar have been used in their operational configuration. However, it would technically be possible for both systems to go to higher temporal and vertical resolutions optimized for PBL height detection, which could slightly improve their performance. Moreover, more complex procedures for the CBL height detection and for the selection of clear CBL cases can greatly increase the accuracy of the results, but decrease the availability of the CBL height detection (see for example Bianco et al. 2008).
The forecast model COSMO-2 uses the bR method applied to the
We conclude that the MWR/PM is the most robust among the experimental methods under consideration and best suited for automatic realtime detection of the PBL height. It provides good results under a wide range of meteorological conditions. Moreover, the MWR/SBI and SBLpT allow the characterization of the nocturnal SBL. It is however necessary to have access to a ceilometer or lidar to monitor the RL height.
Taking advantage of all available upper-air measurements, the principal
features of the PBL are well depicted by the 2-year climatology. The
annual cycle of the CBL height with its maxima at 1500 m during the
May–August period is detected by all instruments and seems to follow the
solar radiation cycle rather than the
The authors greatly acknowledge Robert Sica for reviewing the manuscript. Edited by: J.-Y. C. Chiu