**5.5 Correlative space-borne and ground-based lidar measurements**

Atmospheric profiling by a network of ground-based lidar stations is an optimal approach for validation of results obtained by space-borne lidars, providing supporting data to fully exploit the information from satellite lidar missions. Such a mission is the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO). The Cloud-Aerosol LIdar with Orthogonal Polarization (CALIOP), mounted on the CALIPSO satellite, is a Nd:YAGlaser-based lidar specially designed for aerosol and cloud monitoring. Several years correlative ground-based lidar measurements, performed by the EARLINET stations as synchonized with CALIPSO overpasses, contribute to the specialized database, illustrating the potential of the lidar network to provide a sustainable ground-based support for spaceborne lidar missions (Pappalardo et al., 2009, 2010).

The Sofia CuBr-lidar group is involved in correlative measurements for CALIPSO since June 2006 (Grigorov et al., 2007). Results of mesurements performed on 28 April 2009 by the ground-based lidar and by the CALIOP lidar are presented in Figs.9a and 9b, respectively.

Vertical red lines on the plots indicate the time of satellite passage over Sofia. On the first plot two aerosol layers can be distinguished. The lower one, located at 1-1.5 km altitude, is due to air convection in the PBL. As seen on the corresponding forecast map of Sahara dust load (Fig.9c), the region covered by the dust flow is in the immediate vicinity of Sofia.

LIDAR Atmospheric Sensing by Metal Vapor and Nd:YAG Lasers 365

data. The satellite flies over Bulgaria from North-East to South-West. On the CALIOP lidar image, thick aerosol layers occur at 6-9 km altitudes close to the moment of satellite passage over Sofia (see Fig.9b). Those layers, moving from West, are observed by the ground-based

Lidar sensing in the visible range is typically performed at the second harmonic of the Nd:YAG laser radiation – 532 nm, combined in some lidar systems with the Raman line of the nitrogen molecule at 607 nm. The interval between 532 nm and 1064 nm is not widely used in lidar sounding, mainly because of the dominating usage of Nd:YAG lasers. The US lidar network MPLNET makes use on laser emissions within approximately the same spectral bandwidth (523 nm or 527 nm). As clear from preceding considerations, the metal-vapor lasers cover partially the spectral interval mentioned above. In this Section we consider multiwavelength lidar probing by MV and Nd:YAG lasers. Some useful theoretical simplifications, providing more correct determination of atmospheric backscattering coefficient profiles, and

**6.1 Lidar equations in the case of multi-wavelength sensing by metal vapor lasers** 

1 and 2

2=578.2 nm). The case of lidar

2(*h*) are the geometrical overlapping

*m*1(*h*) and

2 2

20 20

. (13b)

, *Th h h h Th h* 

, *Th h h h Th h* 

1 1

10 10

(13a)

 

 

*<sup>m</sup>*2(*h*) are the

*h dh*

Let us consider for simplicity the case of lidar sensing at two close wavelengths

sensing at more than two wavelengths can be analyzed using similar approach.

1 11

2 22

 

  Assuming a vertical sounding and replacing further the distance *r* with the height *h*, the normalized range-corrected-signal profiles *S*1(*h*) and *S*2(*h*) for both wavelengths are given by

*Sh C h h hTh* 1 11 1 1 1

*Sh C h h hTh* 2 22 2 2 2

1(*h*) and

*a*2(*h*) are the aerosol backscattering profiles;

are the atmospheric transmissions. Let us choose some joint reference height *h*0 for the two channels and denote by *S*1(*h0*) and *S*2(*h0*) the corresponding RCSs, as well as by

Normalizing the lidar profiles *S*1(*h*) and *S*2(*h*) by their values at *h=h*0, one can obtain (for *h > h*0)

;

;

1=510.6 nm and

 

 

*a m* , (12a)

*h dh* and 2 2 <sup>0</sup> ( ) exp[ 2 ( ') '] *<sup>h</sup> T h*

*<sup>m</sup>*2(*h*0)] - the total scattering coefficients.

1 0

2 0

*a m* , (12b)

lidar in about 1 hour (at ~ 01:50 h UTC, Fig.9a).

related experimental results are described and discussed.

**6. Multi-wavelength lidar sensing** 

(e.g. the ones emitted by a CuBr laser:

where *C*1 and *C*2 are the lidar constants;

*m*1(*h*0)] and

molecular backscattering profiles; 1 1 <sup>0</sup> ( ) exp[ 2 ( ') '] *<sup>h</sup> T h*

2(*h*0)=[*a*2(*h*0)+

1 0 1 0 10 1 , , ( ) *a m o*

2 0 2 0 20 2 , , ( ) *a m o*

*Sh h h L hh h h Sh h* 

*Sh h h L hh h h Sh h* 

functions;

1(*h*0)=[*a*1(*h*0)+

*a*1(*h*) and

the dimensionless lidar profiles:

Fig. 9. Sofia-lidar-station and CALIPSO correlative measurements: a) ground-based CuBr lidar data; b) CALIPSO satellite lidar data; c) BSC-DREAM forecast map of Saharan dust load; d) HYSPLIT backward air mass trajectories. The vertical red lines on plots (a) and (b) show the satellite overpass time above Sofia.

Two of the calculated backward air mass trajectories (colored in red and blue on Fig.9d) originate from North Europe. Altough passing over regions loaded with Saharan dust, they are not related to aerosol layers perceptible by the lidar. The green trajectory has an origin above the Atlantic Ocean, indicating for probable transport of humid air which, mixed with the fine Saharan dust particles, can form the aerosol fields appearing at 7.5-8 km height. Similar aerosol layers are present at the same altitude on the plot of CALIPSO satellite lidar data. The satellite flies over Bulgaria from North-East to South-West. On the CALIOP lidar image, thick aerosol layers occur at 6-9 km altitudes close to the moment of satellite passage over Sofia (see Fig.9b). Those layers, moving from West, are observed by the ground-based lidar in about 1 hour (at ~ 01:50 h UTC, Fig.9a).

#### **6. Multi-wavelength lidar sensing**

364 Advanced Photonic Sciences

a) b)

c) d)

show the satellite overpass time above Sofia.

Fig. 9. Sofia-lidar-station and CALIPSO correlative measurements: a) ground-based CuBr lidar data; b) CALIPSO satellite lidar data; c) BSC-DREAM forecast map of Saharan dust load; d) HYSPLIT backward air mass trajectories. The vertical red lines on plots (a) and (b)

Two of the calculated backward air mass trajectories (colored in red and blue on Fig.9d) originate from North Europe. Altough passing over regions loaded with Saharan dust, they are not related to aerosol layers perceptible by the lidar. The green trajectory has an origin above the Atlantic Ocean, indicating for probable transport of humid air which, mixed with the fine Saharan dust particles, can form the aerosol fields appearing at 7.5-8 km height. Similar aerosol layers are present at the same altitude on the plot of CALIPSO satellite lidar Lidar sensing in the visible range is typically performed at the second harmonic of the Nd:YAG laser radiation – 532 nm, combined in some lidar systems with the Raman line of the nitrogen molecule at 607 nm. The interval between 532 nm and 1064 nm is not widely used in lidar sounding, mainly because of the dominating usage of Nd:YAG lasers. The US lidar network MPLNET makes use on laser emissions within approximately the same spectral bandwidth (523 nm or 527 nm). As clear from preceding considerations, the metal-vapor lasers cover partially the spectral interval mentioned above. In this Section we consider multiwavelength lidar probing by MV and Nd:YAG lasers. Some useful theoretical simplifications, providing more correct determination of atmospheric backscattering coefficient profiles, and related experimental results are described and discussed.

#### **6.1 Lidar equations in the case of multi-wavelength sensing by metal vapor lasers**

Let us consider for simplicity the case of lidar sensing at two close wavelengths 1 and 2 (e.g. the ones emitted by a CuBr laser: 1=510.6 nm and 2=578.2 nm). The case of lidar sensing at more than two wavelengths can be analyzed using similar approach.

Assuming a vertical sounding and replacing further the distance *r* with the height *h*, the normalized range-corrected-signal profiles *S*1(*h*) and *S*2(*h*) for both wavelengths are given by

$$S\_1(h) = \mathbb{C}\_1 \gamma\_1(h) \bigsqcup \beta\_{a1}(h) + \beta\_{m1}(h) \bigsqcup T\_1(h) \tag{12a}$$

$$\mathcal{S}\_2(h) = \mathcal{C}\_2 \mathcal{V}\_2(h) \Big[ \mathcal{B}\_{a2}(h) + \mathcal{B}\_{m2}(h) \Big] \mathcal{T}\_2(h) \quad , \tag{12b}$$

where *C*1 and *C*2 are the lidar constants; 1(*h*) and 2(*h*) are the geometrical overlapping functions; *a*1(*h*) and *a*2(*h*) are the aerosol backscattering profiles; *m*1(*h*) and *<sup>m</sup>*2(*h*) are the molecular backscattering profiles; 1 1 <sup>0</sup> ( ) exp[ 2 ( ') '] *<sup>h</sup> T h h dh* and 2 2 <sup>0</sup> ( ) exp[ 2 ( ') '] *<sup>h</sup> T h h dh* are the atmospheric transmissions. Let us choose some joint reference height *h*0 for the two channels and denote by *S*1(*h0*) and *S*2(*h0*) the corresponding RCSs, as well as by 1(*h*0)=[*a*1(*h*0)+*m*1(*h*0)] and 2(*h*0)=[*a*2(*h*0)+*<sup>m</sup>*2(*h*0)] - the total scattering coefficients. Normalizing the lidar profiles *S*1(*h*) and *S*2(*h*) by their values at *h=h*0, one can obtain (for *h > h*0) the dimensionless lidar profiles:

$$L\_1(h, h\_0) = \frac{S\_1(h)}{S\_1(h\_0)} = \frac{\mathcal{J}\_{a1}(h) + \mathcal{J}\_{m1}(h)}{\mathcal{J}\_1(h\_0)} \Gamma\_1(h, h\_0); \quad \Gamma\_1(h, h\_0) = \frac{T\_1(h)}{T\_1(h\_0)} \frac{\gamma\_1(h)}{\gamma\_1(h\_0)}\tag{13a}$$

$$\Gamma\_2\left(h, h\_0\right) = \frac{S\_2\left(h\right)}{S\_2\left(h\_0\right)} = \frac{\beta\_{a2}\left(h\right) + \beta\_{m2}\left(h\right)}{\beta\_2\left(h\_0\right)}\Gamma\_2\left(h, h\_0\right); \quad \Gamma\_2\left(h, h\_0\right) = \frac{T\_2\left(h\right)}{T\_2\left(h\_0\right)}\frac{\gamma\_2\left(h\right)}{\gamma\_2\left(h\_0\right)}.\tag{13b}$$

LIDAR Atmospheric Sensing by Metal Vapor and Nd:YAG Lasers 367

particular, the BAE values can be influenced by the relative humidity of the atmosphere (Del Guasta & Marini, 2000; Del Guasta, 2002). As calculated from lidar data, the BAE represents a range-resolved function, in contrast to the classical Ångström exponent, thus providing information about the range variations of the aerosol size distribution. In lidars emitting very different wavelengths (say, the first and the second harmonics), the large wavelength

mutual dispositions of the aerosol size distribution modes with respect to the laser wavelengths. This is due to the inequalities *B*a1(*h*) *B*a2(*h*), indicating for some difference in wavelength dependent scattering mechanisms. It is evident that the conditions (16, 19), when *B*a1(*h*) *B*a2(*h*) can be satisfied within some defined wavelength domain of bandwidth

, depending on the aerosol size distribution and the aerosol composition. In these cases one can accept some similarity in the aerosol scattering mechanisms. In a clear atmosphere

ranges. In the case of multimode size distribution, it can be quite narrower, depending also

It is worth to discuss the opportunities for solving equations (14a,b) for the two wavelengths

solving single wavelengths equations (16). The number of arguments could also be reduced

well as applying some well-known models for the standard molecular atmosphere. Using then the model for aerosol backscattering wavelength dependence present above, the number of arguments could also be additionally reduced. The application of this approach

 The above analysis shows that the simultaneous multi-wavelength MVL lidar sensing in the mid-visible range, based on the application of the backscattering Ångström exponent profile, is attractive for characterizing vertical aerosol size distribution variations in the submicron and near-micron ranges. The combination with Nd:YAG lidar sensing (1064 nm, 532 nm) is a good approach for a reliable characterization of the most typical atmospheric

We present below some target experimental results obtained with the CuBr lidar system

purpose, let us define the profile of the ratio *R*(*h*,*h*0) of the two normalized lidar profiles

*<sup>h</sup> hh h Rhh*

 

 

1 0 1 1

*h hh*

20 1 1

1=510.6 nm and

*Rhh L hh L hh* , ,, 0 2 01 0 , 0 *h h* . (20)

 

*a m a m*

2). An important requirement here is the normalized lidar profiles *L*1(*h*,*h0*) and *L*2(*h*,*h0*) to be well distinguished with respect to the noise. As seen, the number of arguments

1(*h*0) and

*m*1(*h*) and

1, 

importance for extracting more and reliable information from the lidar sensing.

can practically cover the entire visible and a part of infrared

2). The validity of equations (16,19) is of great

(*h*), depending on shapes and

2(*h*0) could be determined by

*<sup>m</sup>*2(*h*) [see Eq.(15)], as

2=578.2 nm. To this

. (21)

differences can cause large variations of the calculated factor

exceeds the number of equations. The parameters

using the links of both molecular scattering coefficients

**6.2 Estimation of aerosol-to-molecular scattering proportions** 

0

After some transformations, using the simplified expressions (15) we obtain:

( ) , ( )

(Kolarov et al., 1995) emitting two basic wavelengths:

(1, 

(single mode distribution)

on the choice of both wavelengths (

is out of the scope of this analysis.

aerosol loadings.

*L*1(*h*,*h0*) and *L*2(*h*,*h0*):

As seen, the normalized lidar profiles do not depend on lidar constants *C*1 and *C*2 and thus, on some lidar parameters as the emitted powers, receiver sensitivities, etc. The dependence on the overlapping functions can be minimized, if the reference height *h*0 is chosen so that 1(*h*)~2(*h*)~1, for *h>h*0. For non-absorbing atmosphere one could also accept the atmospheric transmissions for 1 and 2 to be close, i.e. *T*1(*h*)/*T*1(*h0*) ~ *T*2(*h*)/*T*2(*h0*) ~ 1, and thus, 1(*h*,*h0*) ~ 2(*h*,*h0*) ~1. As a result, the normalized profiles *L*1(*h*,*h0*) and *L*2(*h*,*h0*) will depend on the atmospheric parameters by the profiles of the aerosol and molecular backscattering coefficients. Thus, expressions (13a,b) can be written in the forms:

$$
\beta\_{\rm u1}(\text{h}) + \beta\_{\rm m1}(\text{h}) = \beta\_1(\text{h}\_0)L\_1(\text{h}\_\prime \text{h}\_0),
\tag{14a}
$$

$$
\beta\_{\rm u2}(\text{h}) + \beta\_{\rm w2}(\text{h}) = \beta\_2(\text{h}\_0)L\_2(\text{h}\_\prime \text{h}\_0). \tag{14b}
$$

The molecular scattering coefficients are expressed by <sup>4</sup> *m m* 1 1 *h Bh* and <sup>4</sup> *m m* 2 2 *h Bh* , where *Bm*(*h*) does not depend on (Measures, 1984). Their ratio is given by:

$$
\partial \left( \mathcal{k}\_1, \mathcal{k}\_2 \right) = \mathcal{J}\_{m2} \left( h \right) \Big/ \mathcal{J}\_{m1} \left( h \right) = \left( \mathcal{k}\_1 / \mathcal{k}\_2 \right)^4. \tag{15}
$$

In the case of CuBr lidar, (1,2) =0.6.

By analogy with the molecular scattering, the wavelength dependence of the aerosol backscattering coefficients could be presented in the form:

$$\mathcal{B}\_{a1}(h,\mathcal{Z}\_1) = \mathcal{B}\_{a1}(h)\mathcal{Z}\_1^{-\eta(h)} \; ; \; \mathcal{B}\_{a2}(h,\mathcal{Z}\_1) = \mathcal{B}\_{a2}(h)\mathcal{Z}\_2^{-\eta(h)} \; ; \; \; h \ge h\_0 \; \; \tag{16}$$

where *Ba*1,2(*h*) and (*h*) do not depend on the wavelength in broad spectral domains. The factor typically varies within the range 0.57 for different types of aerosol (Toriumi et al., 1994). Based on the initial supposition stating closeness of the two wavelengths, one can assume

$$B\_{a1}(\mathbf{h}) \approx B\_{a2}(\mathbf{h}) = B\_{a}(\mathbf{h}) \tag{17}$$

and the ratio (*h*,1,2) of backscattering coefficients *a*2(*h*) and *<sup>a</sup>*1(*h*) is expressed by a dependence similar to (15):

$$
\mu\left(\hbar,\mathbb{X}\_1,\mathbb{X}\_2\right) = \beta\_{a2}\left(\hbar\right)\Big|\mathcal{J}\_{a1}\left(\hbar\right) = \left(\mathbb{X}\_1/\mathbb{X}\_2\right)^{\eta\left(h\right)}.\tag{18}
$$

The vertical profile of (*h*) used for characterizing aerosol types is given by

$$\ln \eta(h) = \ln \left[ \mu(h, \lambda\_1, \lambda\_2) \right] \Big/ \ln(\lambda\_1/\lambda\_2) = \ln[\beta\_{a2}(h)] / \beta\_{a1}(h) \Big/ \ln \left(\lambda\_1/\lambda\_2\right). \tag{19}$$

The parameter (*h*) as defined in (19) is also called aerosol backscattering-related Ångström exponent (BAE) (Del Guasta, 2002; Kamei et al., 2006). It is involved by analogy with the Ångström exponent (Ångström, 1929, 1964; Shuster et al., 2005) which is normally expressed in terms of aerosol optical depth or extinction. Generally, (*h*) is a complex function and characteristics of the aerosol particle size distribution and mode volume fractions. In

As seen, the normalized lidar profiles do not depend on lidar constants *C*1 and *C*2 and thus, on some lidar parameters as the emitted powers, receiver sensitivities, etc. The dependence on the overlapping functions can be minimized, if the reference height *h*0 is chosen so that

~ 2(*h*,*h0*) ~1. As a result, the normalized profiles *L*1(*h*,*h0*) and *L*2(*h*,*h0*) will depend on the atmospheric parameters by the profiles of the aerosol and molecular backscattering

The molecular scattering coefficients are expressed by <sup>4</sup> *m m* 1 1

12 2 1 12

By analogy with the molecular scattering, the wavelength dependence of the aerosol

; 21 2 2 , *<sup>h</sup>*

et al., 1994). Based on the initial supposition stating closeness of the two wavelengths, one

 *Ba*1(*h*) *Ba*2(*h*) = *Ba*(*h*) (17)

 12 2 1 12 , , *<sup>h</sup>*

 

12 1 2 1 12 <sup>2</sup>

exponent (BAE) (Del Guasta, 2002; Kamei et al., 2006). It is involved by analogy with the Ångström exponent (Ångström, 1929, 1964; Shuster et al., 2005) which is normally expressed

characteristics of the aerosol particle size distribution and mode volume fractions. In

(*h*) used for characterizing aerosol types is given by

 

 

coefficients. Thus, expressions (13a,b) can be written in the forms:

 *a*1(*h*)+*m*1(*h*) = 

 *a*2(*h*)+*m*2(*h*) = 

, where *Bm*(*h*) does not depend on

 

backscattering coefficients could be presented in the form:

 11 11 , *<sup>h</sup> a a h Bh*

typically varies within the range 0.57

 

*h h* ln , , ln( ) ln[

in terms of aerosol optical depth or extinction. Generally,

  (*h*) do not depend on the wavelength

2) of backscattering coefficients

*a a h hh*

 

 

 

(1,2) =0.6.

2(*h*)~1, for *h>h*0. For non-absorbing atmosphere one could also accept the atmospheric

2 to be close, i.e. *T*1(*h*)/*T*1(*h0*) ~ *T*2(*h*)/*T*2(*h0*) ~ 1, and thus, 1(*h*,*h0*)

<sup>4</sup>

*a a h Bh*

 (,) *m m h h* . (15)

1(*h*0)*L*1(*h*,*h0*), (14a)

2(*h*0)*L*2(*h*,*h0*). (14b)

*h Bh*

, 0 *h h* , (16)

for different types of aerosol (Toriumi

 

(*h*) is a complex function and

in broad spectral domains. The

*<sup>a</sup>*1(*h*) is expressed by a

(Measures, 1984). Their ratio is

and

 

 

> 

*a a h h* ]/ln . (19)

(*h*) as defined in (19) is also called aerosol backscattering-related Ångström

*a*2(*h*) and

. (18)

1(*h*)~

transmissions for

<sup>4</sup> *m m* 2 2

In the case of CuBr lidar,

where *Ba*1,2(*h*) and

factor 

can assume

and the ratio

The parameter

(*h*,1,

dependence similar to (15):

The vertical profile of

*h Bh*

given by:

1 and  particular, the BAE values can be influenced by the relative humidity of the atmosphere (Del Guasta & Marini, 2000; Del Guasta, 2002). As calculated from lidar data, the BAE represents a range-resolved function, in contrast to the classical Ångström exponent, thus providing information about the range variations of the aerosol size distribution. In lidars emitting very different wavelengths (say, the first and the second harmonics), the large wavelength differences can cause large variations of the calculated factor (*h*), depending on shapes and mutual dispositions of the aerosol size distribution modes with respect to the laser wavelengths. This is due to the inequalities *B*a1(*h*) *B*a2(*h*), indicating for some difference in wavelength dependent scattering mechanisms. It is evident that the conditions (16, 19), when *B*a1(*h*) *B*a2(*h*) can be satisfied within some defined wavelength domain of bandwidth , depending on the aerosol size distribution and the aerosol composition. In these cases one can accept some similarity in the aerosol scattering mechanisms. In a clear atmosphere (single mode distribution) can practically cover the entire visible and a part of infrared ranges. In the case of multimode size distribution, it can be quite narrower, depending also on the choice of both wavelengths (1, 2). The validity of equations (16,19) is of great importance for extracting more and reliable information from the lidar sensing.

It is worth to discuss the opportunities for solving equations (14a,b) for the two wavelengths (1, 2). An important requirement here is the normalized lidar profiles *L*1(*h*,*h0*) and *L*2(*h*,*h0*) to be well distinguished with respect to the noise. As seen, the number of arguments exceeds the number of equations. The parameters 1(*h*0) and 2(*h*0) could be determined by solving single wavelengths equations (16). The number of arguments could also be reduced using the links of both molecular scattering coefficients *m*1(*h*) and *<sup>m</sup>*2(*h*) [see Eq.(15)], as well as applying some well-known models for the standard molecular atmosphere. Using then the model for aerosol backscattering wavelength dependence present above, the number of arguments could also be additionally reduced. The application of this approach is out of the scope of this analysis.

 The above analysis shows that the simultaneous multi-wavelength MVL lidar sensing in the mid-visible range, based on the application of the backscattering Ångström exponent profile, is attractive for characterizing vertical aerosol size distribution variations in the submicron and near-micron ranges. The combination with Nd:YAG lidar sensing (1064 nm, 532 nm) is a good approach for a reliable characterization of the most typical atmospheric aerosol loadings.

#### **6.2 Estimation of aerosol-to-molecular scattering proportions**

We present below some target experimental results obtained with the CuBr lidar system (Kolarov et al., 1995) emitting two basic wavelengths: 1=510.6 nm and 2=578.2 nm. To this purpose, let us define the profile of the ratio *R*(*h*,*h*0) of the two normalized lidar profiles *L*1(*h*,*h0*) and *L*2(*h*,*h0*):

$$R\left(h\_{\prime}h\_{0}\right) = L\_{2}\left(h\_{\prime}h\_{0}\right) \left\langle L\_{1}\left[h\_{\prime}h\_{0}\right] \right\rangle, \quad h \ge h\_{0} \,. \tag{20}$$

After some transformations, using the simplified expressions (15) we obtain:

$$R\left(h, h\_0\right) = \left(\frac{\beta\_1(h\_0)}{\beta\_2(h\_0)}\right) \frac{\mu\left(h\right)\beta\_{a1}\left(h\right) + \theta\beta\_{m1}\left(h\right)}{\beta\_{a1}\left(h\right) + \beta\_{m1}\left(h\right)}.\tag{21}$$

LIDAR Atmospheric Sensing by Metal Vapor and Nd:YAG Lasers 369

wavelengths are closely disposed (but at well distinguished lidar profiles) so the uncertainties introduced by the parameters of two lidar channels can be minimized (see Eqs.13a,b). In this sense, the application of MVL in lidar atmospheric probing can be

substantial for improving the measurement accuracy.

a) b)

zone, profiles are quite smooth, without differentiated bulges.

scattering) atmosphere on 17.07.2008.

Fig. 10. a) Normalized lidar profiles 1 and 2 (on both wavelengths, respectively) and the calculated ratio *R*(*h*,*h*max,*h*0) 3, measured in a clear atmosphere on 10.07.2008; b) Normalized lidar profiles 1 and 2 and the calculated ratio 3, measured in mixed (aerosol and molecular

**6.3 Two-wavelength lidar probing of aerosol mode fractions over complex terrain** 

Results of lidar observations on atmospheric aerosols over a complex terrain (Fernando, 2010), representing adjoining city-, plain-, and mountain zones in Sofia region, are described below. A residential city zone is located along with the first 2 kilometers of the laser beam path. The distance range 2 - 5 km covers city outskirts and suburbs (plain zone) and the one from 5.5 km to about 9 km covers the mountain foot, slope, and ridge (mountain zone).

Lidar measurements (Fig.11) are carried out simultaneously at wavelengths 1064 nm and 532 nm (Peshev et al., 2011). Range profiles of evaluated aerosol backscattering coefficients (molecular component subtracted) on 28.01.2008, averaged over the period of measurements (17:10 h – 18:15 h GMT), are presented in Fig.12a. Over the plain area, the backscattering coefficient at 532 nm is permanently higher than the one at 1064 nm, starting with a ratio of about 1.5-2 and gradually decreasing to equalization close to the interface between the plain and mountain. Humps observed at the initial parts of the two profiles (city zone) are identified to be due to increased anthropogenic aerosol emissions. For the rest of the plain

Over the mountain zone, the situation is inverted – the backscattering at 1064 nm exceeds the one at 532 nm, both having values considerably higher than those for the plain zone, denoting

The variation range of *R*(*h*,*h*0) for 0 *h h* can be easily estimated. At the reference height *h*<sup>0</sup> one will obtain *R*(*h*,*h*0)=1. Further, at a proper choice of the reference height *h*0 the contribution of the molecular scattering could be neglected for some heights 0 *h h* or *a*1,2(*h*) >>*m*1,2(*h*) and then, (*h*)~*R*(*h*,*h*0)2(*h*0)/1(*h*0) as well as (*h*0)~a2(*h*0)/a1(*h*0). In the opposite case, one could neglect the contribution of the aerosol scattering with respect to the molecular scattering (Sec.2) above some height *h*max (*h*max>*h*0), where a1,2(*h*max) << m1,2(*h*max). For this case an approximate estimate for the ratio *R*(*h*max,*h*0) can be also obtained. Using Eq.(21) one will obtain the estimate *R*(*h*>*h*max,*h*0)~1(*h*0)/2(*h*0)~, if assuming 1(*h*0)/2(*h*0)~ 1 for lower heights [see Eqs.(16-19)]. Thus, the range of the height variations of the profile *R*(*h*,*h*0) will vary approximately within the borders:

$$
\theta \le \mathsf{R}(\mathsf{h}\_{\mathsf{h}} \mathsf{h}\_{\mathsf{0}}) \le 1 \quad \text{.} \tag{22}
$$

The use of the ratio *R*(*h*,*h*0) provides a clear and easy determination of the height *h*max where one can accept an absence of an aerosol loading. In most of lidar systems it is typically accepted to attach the molecular scattering profile to the range-corrected profiles *S*1(*h*) and *S*2(*h*) at heights above 12-15 km. But, in the presence of intensive daily background this attachment could be even incorrect as the lidar profiles *S*1(*h*) and *S*2(*h*) can contain only noise at higher altitudes. The knowledge of the altitude *h*max can provide the correct attachment of the molecular profiles in such cases. Moreover, this approach can provide well defined estimates of the aerosol scattering contribution in the point of attachment, if it is lower than *h*max and thus, an attachment to the molecular scattering in the presence of an aerosol contribution. The measured ratio *R*(*h*,*h*0) can also be used for estimating the profile of aerosol-to-molecular backscattering coefficients.

Below, we demonstrate some experimental results concerning determination and analysis of calculated profiles of the ratio *R*(*h*,*h*0) for different aerosol loadings. The plots in Fig.10a present the two normalized lidar profiles: *L*1(*h*,*h*0) (curve 1) and *L*2(*h*,*h*0) (curve 2) at wavelengths 1 and 2, respectively. The reference height *h*0 is equal to 2 km. Note that both normalized profiles are well distinguished. As expected, the profile for 1 is situated in the region of higher values than the profile for 2. This is due to the higher molecular scattering for shorter wavelengths. The ratio *R*(*h*,*h*max,*h*0) is larger than the lower estimate in relation (22) for height up to ~ 4 km. It becomes ~ 0.6 just at this altitude and above, therefore, *h*max is of the order of 4 km. The attachment of the molecular scattering profile can be implemented at heights above 4 km. As evident, the height *h*max for attachment of the Rayleigh profiles can be estimated here from lidar data only. This is essential for the accuracy of retrieved aerosol parameters, providing an opportunity to avoid sometimes the application of complicated set of calibration procedures (Bösenberg & Hoff, 2007).

The results in Fig.10b correspond to the case, when the calculated ratio *R*(*h*,*h*max,*h*0) is higher and approximately equal to ~ 0.8 in a large region of altitudes up to 10 km. This case can be characterized as a case of significant aerosol loading up to altitudes of 10 km. It must be noted that the both normalized profiles are approximately parallel and well distinguished as well. Some additional small aerosol contribution above 6km are well seen also in the profile of the ratio *R*(*h*,*h*max,*h*0). This fact could be explained by some changes of the aerosol size distribution at these altitudes, due to the presence of additional thin cloud loading having different aerosol scattering structure. For lower heights, the ratio *R*(*h*,*h*max,*h*0) tends to unity as expected. As seen, the ratio *R*(*h*,*h*max,*h*0) becomes very informative when the two

The variation range of *R*(*h*,*h*0) for 0 *h h* can be easily estimated. At the reference height *h*<sup>0</sup> one will obtain *R*(*h*,*h*0)=1. Further, at a proper choice of the reference height *h*0 the contribution of the molecular scattering could be neglected for some heights 0 *h h* or

opposite case, one could neglect the contribution of the aerosol scattering with respect to the

For this case an approximate estimate for the ratio *R*(*h*max,*h*0) can be also obtained. Using

1 for lower heights [see Eqs.(16-19)]. Thus, the range of the height variations of the profile

The use of the ratio *R*(*h*,*h*0) provides a clear and easy determination of the height *h*max where one can accept an absence of an aerosol loading. In most of lidar systems it is typically accepted to attach the molecular scattering profile to the range-corrected profiles *S*1(*h*) and *S*2(*h*) at heights above 12-15 km. But, in the presence of intensive daily background this attachment could be even incorrect as the lidar profiles *S*1(*h*) and *S*2(*h*) can contain only noise at higher altitudes. The knowledge of the altitude *h*max can provide the correct attachment of the molecular profiles in such cases. Moreover, this approach can provide well defined estimates of the aerosol scattering contribution in the point of attachment, if it is lower than *h*max and thus, an attachment to the molecular scattering in the presence of an aerosol contribution. The measured ratio *R*(*h*,*h*0) can also be used for estimating the profile of

Below, we demonstrate some experimental results concerning determination and analysis of calculated profiles of the ratio *R*(*h*,*h*0) for different aerosol loadings. The plots in Fig.10a present the two normalized lidar profiles: *L*1(*h*,*h*0) (curve 1) and *L*2(*h*,*h*0) (curve 2) at

application of complicated set of calibration procedures (Bösenberg & Hoff, 2007).

for shorter wavelengths. The ratio *R*(*h*,*h*max,*h*0) is larger than the lower estimate in relation (22) for height up to ~ 4 km. It becomes ~ 0.6 just at this altitude and above, therefore, *h*max is of the order of 4 km. The attachment of the molecular scattering profile can be implemented at heights above 4 km. As evident, the height *h*max for attachment of the Rayleigh profiles can be estimated here from lidar data only. This is essential for the accuracy of retrieved aerosol parameters, providing an opportunity to avoid sometimes the

The results in Fig.10b correspond to the case, when the calculated ratio *R*(*h*,*h*max,*h*0) is higher and approximately equal to ~ 0.8 in a large region of altitudes up to 10 km. This case can be characterized as a case of significant aerosol loading up to altitudes of 10 km. It must be noted that the both normalized profiles are approximately parallel and well distinguished as well. Some additional small aerosol contribution above 6km are well seen also in the profile of the ratio *R*(*h*,*h*max,*h*0). This fact could be explained by some changes of the aerosol size distribution at these altitudes, due to the presence of additional thin cloud loading having different aerosol scattering structure. For lower heights, the ratio *R*(*h*,*h*max,*h*0) tends to unity as expected. As seen, the ratio *R*(*h*,*h*max,*h*0) becomes very informative when the two

normalized profiles are well distinguished. As expected, the profile for

2, respectively. The reference height *h*0 is equal to 2 km. Note that both

1(*h*0)/2(*h*0)~

1(*h*0) as well as

(*h*0)~a2(*h*0)/

*R*(*h*,*h*0)1 . (22)

, if assuming

2. This is due to the higher molecular scattering

1 is situated in the

a1,2(*h*max) <<

a1(*h*0). In the

m1,2(*h*max).

1(*h*0)/2(*h*0)~

wavelengths

1 and 

*a*1,2(*h*) >>

*m*1,2(*h*) and then,

Eq.(21) one will obtain the estimate *R*(*h*>*h*max,*h*0)~

aerosol-to-molecular backscattering coefficients.

region of higher values than the profile for

*R*(*h*,*h*0) will vary approximately within the borders:

(*h*)~*R*(*h*,*h*0)

molecular scattering (Sec.2) above some height *h*max (*h*max>*h*0), where

2(*h*0)/

wavelengths are closely disposed (but at well distinguished lidar profiles) so the uncertainties introduced by the parameters of two lidar channels can be minimized (see Eqs.13a,b). In this sense, the application of MVL in lidar atmospheric probing can be substantial for improving the measurement accuracy.

Fig. 10. a) Normalized lidar profiles 1 and 2 (on both wavelengths, respectively) and the calculated ratio *R*(*h*,*h*max,*h*0) 3, measured in a clear atmosphere on 10.07.2008; b) Normalized lidar profiles 1 and 2 and the calculated ratio 3, measured in mixed (aerosol and molecular scattering) atmosphere on 17.07.2008.

## **6.3 Two-wavelength lidar probing of aerosol mode fractions over complex terrain**

Results of lidar observations on atmospheric aerosols over a complex terrain (Fernando, 2010), representing adjoining city-, plain-, and mountain zones in Sofia region, are described below. A residential city zone is located along with the first 2 kilometers of the laser beam path. The distance range 2 - 5 km covers city outskirts and suburbs (plain zone) and the one from 5.5 km to about 9 km covers the mountain foot, slope, and ridge (mountain zone).

Lidar measurements (Fig.11) are carried out simultaneously at wavelengths 1064 nm and 532 nm (Peshev et al., 2011). Range profiles of evaluated aerosol backscattering coefficients (molecular component subtracted) on 28.01.2008, averaged over the period of measurements (17:10 h – 18:15 h GMT), are presented in Fig.12a. Over the plain area, the backscattering coefficient at 532 nm is permanently higher than the one at 1064 nm, starting with a ratio of about 1.5-2 and gradually decreasing to equalization close to the interface between the plain and mountain. Humps observed at the initial parts of the two profiles (city zone) are identified to be due to increased anthropogenic aerosol emissions. For the rest of the plain zone, profiles are quite smooth, without differentiated bulges.

Over the mountain zone, the situation is inverted – the backscattering at 1064 nm exceeds the one at 532 nm, both having values considerably higher than those for the plain zone, denoting

LIDAR Atmospheric Sensing by Metal Vapor and Nd:YAG Lasers 371

The BAE takes gradually decreasing negative values for distances corresponding to the mountain zone (Fig.12b). Negative values of the extinction-related or backscatter-related Ångström exponent (Kamei, 2006; Lu, 2011; Guerrero-Rascado, 2009) occur in cases of large aerosol particles in the over-micron size range (e.g. large water droplets, ice particles, volcanic or fire ashes, partcle aggregates, etc.), as well as in case of large mode volume fractions of coarse aerosols with respect to those of fine aerosols. In these cases, the backscattering at longer wavelengths dominates over the one at shorter wavelengths,

Various lidar technologies have been developed since the first lidar demonstration more than 40 years ago. In this chapter we presented some of the basic applications of lidars in the atmospheric research. The high informativity of lidar probing is due to the strong interaction of optical waves with atmospheric particles – aerosols and molecules, combined with range-resolved acquisition of lidar signals. The fast spread of lidars all over the world in the last two decades led to their organizing in regional lidar networks as EARLINET (Europe), MPLNET (USA), AD-Net (Japan), etc., integrated now in the Global lidar network (GALION). As lidars are practically the only instruments for high resolution vertical atmospheric profiling, their further improvement is one of the most important tasks for the lidar community. One of the ways to enhance the quality of lidar output information and its significance for the global and regional atmospheric monitoring is the creation of complex multispectral lidar systems, capable to provide more detailed and reliable data for the retrieval of aerosol optical, microphysical, and radiative properties, etc. This approach requires the development of novel more effective inverse algorithms, as well. It is also expected the future lidar networks to operate in a close cooperation with other existing networks as the sun-photometer network, groundlevel in-situ aerosol monitoring networks, satellite measurements (lidar and multispectral radiometers), radars, etc. The synergy resulting from such cooperation was demonstrated in a large number of experiments. The expected increase of the lidar station density all over the world is a good indication for their great significance in analysing local and global atmospheric processes and trends. In this connection we note the European project ACTRIS (Aerosols, Clouds, and Trace gases Research InfraStructure Network (www.actris.net)). Started in 2011 under 7-th Framework Program, it integrates the EARLINET, EUSSAR, CLOUDNET, and a new trace gas network into single-coordinated ground-based networks, with monitoring impact on the climate changes, air quality, and

The results described here were funded partly by the Bulgarian National Science Fund under projects Ph-63, Ph-447, Ph-811 and the European Commission under the project EARLINET-ASOS, grant RICA-025991 EC FP6. The authors thank the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT backward trajectories. We also thank the Barcelona Supercomputing Center for Dust Regional Atmospheric Model (DREAM) and for

especially for substantially different wavelengths, as in Fig.12b.

**7. Conclusions** 

long-term transport of pollutants.

the provision of Saharan dust forecast maps.

**8. Acknowledgments** 

presence of dense aerosol layers, most probably - water aerosol (fog or orographic clouds near the surface). The available meteorological data for the day support this conclusion.

Fig. 11. Schematic view of the lidar experiment over complex terrain; WS – wavelength separator; A1 and A2 – aerosol channels at 1064 nm and 532 nm, respectively; R – Raman channel at 607 nm.

Fig. 12. a) Time-averaged profiles of aerosol backscattering coefficient at the two wavelengths of measurement; b) Backscattering-related Ångström exponent profile.

The backscatter-related Ångström exponent defined in Eq.(19) is used as a qualitative indicator of aerosol particle properties. In Fig.12b, the range profile of BAE is presented, corresponding to aerosol backscatter profiles shown on Fig.12a. As one can see, the values of vary within the range 0.5-1.25 for distances of up to 5 km (plain zone). These values indicate for prevalence of the fine-mode aerosol particle fraction, most probably of anthropogenic origin, as typical for urban areas under dry atmospheric conditions.

Around the plain-to-mountain interface zone located about 5.5 km away from the lidar station, BAE values approach zero indicating for prevailing contribution of the coarse-mode aerosol fraction.

The BAE takes gradually decreasing negative values for distances corresponding to the mountain zone (Fig.12b). Negative values of the extinction-related or backscatter-related Ångström exponent (Kamei, 2006; Lu, 2011; Guerrero-Rascado, 2009) occur in cases of large aerosol particles in the over-micron size range (e.g. large water droplets, ice particles, volcanic or fire ashes, partcle aggregates, etc.), as well as in case of large mode volume fractions of coarse aerosols with respect to those of fine aerosols. In these cases, the backscattering at longer wavelengths dominates over the one at shorter wavelengths, especially for substantially different wavelengths, as in Fig.12b.
