J. L. Guerrero-Rascado1,2,3, R. F. da Costa3, A. E. Bedoya4, R. Guardani5,

L. Alados-Arboledas1,2, A. E. Bastidas4 and E. Landulfo3

1Instituto Interuniversitario de Investigación del Sistema Tierra en Andalucía (IISTA-CEAMA), Granada, España

2Dpto. Física Aplicada, Universidad de Granada, Granada, España3Centro de Lasers e Aplicações, Instituto de Pesquisas Energéticas e Nucleares

(IPEN), São Paulo, Brasil 4Escuela de Física, Universidad Nacional de Colombia, Medellín, Colombia

5Dpto. Engenheria Química, Universidade de São Paulo, São Paulo, Brasil

rascado@ugr.es

Dust loading maps ( g/m2) calculated by DREAM model The emission of pollutants in megacities and industrial areas can

have strong impact: environment and human health

Serious health effects: premature death, heart attacks, acute bronchitis and aggravated asthma among children

Premature mortality can be associated to the presence of particulate matter [Molina and Molina, J. Air Waste Manag. Assoc. 2004]

VIII WLMLA, Cayo Coco 9th April

Dust loading maps ( g/m2) calculated by DREAM model Cubatão is an industrialized city, the

most polluted city in the world (United Nations) in the 80s

A large pollutant monitoring program started

In 1983, however, CETESB carried out a pollution control program to reduce air, water, and soil pollution

Sluggish dispersion of air pollutants extreme discomfort and long term health problems for the population (particularly blood-related and respiratory diseases, and birth defects)

VIII WLMLA, Cayo Coco 9th April

Dust loading maps ( g/m2) calculated by DREAM model

VIII WLMLA, Cayo Coco 9th April

One of the sources of pollutants in the city comes from biomass burning and oil refineries

Lidar: very useful due to very high spatial (~m) and temporal resolution (~s)

Its performance in the near range must be tested to get trustable information from flames

OBJECTIVES: 1-. characterization of the electronic subsystem 2-. development of methodology to characterize optical and

microphysical properties inside industrial flares

VIII WLMLA, Cayo Coco 9th April

CEPEMA (Centre for Training and Research in Environment) Cubatão (23º 53' S, 46º 26' W, 10 m asl), state of São Paulo (Brazil) surrounded by mountains to N, W, and E (600-1000 m), and by the

sea to S 40 km2 with: 23 large industry complexes, 1 steel plant, 1 oil

refinery, 7 fertilizer plants, 1 cement plant, and 11 chemical/petrochemical plants

in total 260 pollutant emission sources

~ 400 m

Nd:YAG laser (CFR 450, Quantel SA) Emission/detection: 355, 532, 1064 nm Repetition rate: 20 Hz Analog (AN) and Photon Counting (PC) detection Newtonian telescope Scanning system: 2 AP sun tracker (Kipp & Zonnen) selectable spatial resolutions: 3.75, 7.5, 15 and 30 m

VIII WLMLA, Cayo Coco 9th April

Different tests routinely applied in LALINET and EARLINET networks to detect potential anomalies in the electronic subsystem of the Cubatão scanning lidar

VIII WLMLA, Cayo Coco 9th April

Zero-bin calibration:

Trigger in a lidar is used to fire pulses and also to activate the data acquisition

Ideally laser and chronometer are synchronized

In practise… not

wrong range determination!!!

scatterers

Lidar (light detection and ranging) … but ranging based on timing

laser chronometer

R range

/upt R c

2 /total up downt t t R c

VIII WLMLA, Cayo Coco 9th April

How to measure zero-bin calibration?

Zero-bin test: a target is placed at the output of the laser window in order to produce strong backscattered radiation. Thus, the first intense peak observed by the detector system should correspond to the zero position of our measurements

lidar

telescope Laser beam output

zR

Our target is placed at zero-bin, however…

VIII WLMLA, Cayo Coco 9th April

Bin-shift calibration:

PC signals saturated in near-range zero-bin calibration does not allow for unambiguously determining the zero-bin position for these channelsComparison of the shape of AN and PC signals: both signals should simultaneously detect atmospheric targets at the same distance for ideal systems

VIII WLMLA, Cayo Coco 9th April

How to measure bin-shift calibration?

Bin-shift test: Bin-shift is determined by the best linear fit in a certain distance range between AN and PC signals (using AN signal as a reference), sliding the PC signal around it

10-min measurement, slidingwindow between −20 and +

20 bins

VIII WLMLA, Cayo Coco 9th April

Dark current:

Response exhibited by a receptor of radiation during periods when it is notactively illuminated.

lidar

telescope

Laser beam output

-measurement taken by covering totally the telescope or detectors performed with enough averaging time (~10 min)

-all parameters (voltages, pulse repetition frequency …) are configured as a usual measurement

VIII WLMLA, Cayo Coco 9th April

12x12 virtual matrix used to discretize the flare region in steps of 0.02º (zenith) and 0.03º (azimuth), integration time 10 s

VIII WLMLA, Cayo Coco 9th April

Lidar signal at a given distance:

Forming the ratio at the leading and trailing edge of the flame :

Some considerations:

Flare stack is at 400 m from the instrument O(r)=1

Atmosphere is horizontally homogeneous around the flame β(r1)=β(r2)

VIII WLMLA, Cayo Coco 9th April

Particle extinction coefficient, assuming homogeneity between r1 and r2 :

Finally, the extinction-related Angström exponent:

VIII WLMLA, Cayo Coco 9th April

Different spatial resolutions (i.e., at 3.75, 7.5, 15 and 30 m) in order to check the instrumental characterization at each setting

Zero-bin position is located at bins #10 and #11 (depending on wavelength)Zero-bin position is different for different spatial resolutions selectedLidar technique is a timing technique (which in turn is converted into a ranging technique through the light velocity) the temporal delay between the laser beam emission and start of data acquisition is converted to a different number of bins when different spatial resolutions are used

VIII WLMLA, Cayo Coco 9th April

Relationship among zero-bin positions at different spatial resolutions:

n = 0,1,2,3 determine the different spatial resolutions (resolution(n) = 3.75·2n)

VIII WLMLA, Cayo Coco 9th April

Orographic conditions led to low clouds formation (typically <1 km)

Bin-shift calibration requires a distance-range where both AN and PC properly detect atmospheric structures (aerosol layers or clouds) simultaneously operation at a zenith angle of 25° to guarantee profiling a target at 1.5-2.5 km from the system

VIII WLMLA, Cayo Coco 9th April

All bin-shifts were found to be negative, meaning all PC signals are displaced several bins towards the instrument (with respect to the AN signals)Bin-shift test not applicable to 1064 nm due to extremely low SNR

.

#0 bins for most of them synchronization between emission and start of data acquisition

VIII WLMLA, Cayo Coco 9th April

AN dark current was constant with positive values for the entire distance range sounded, except very close to the instrument where stray-light peaks are detected up to 250 m from the instrument

PC dark current showed an expected value of zero in most of the distance range

VIII WLMLA, Cayo Coco 9th April

Values at each virtual pixel were computed as the mean value of 5 bins along the lidar line-of-sight (range 367.5-397.5 m bins 49-53)

better identification of flame boundaries at larger wavelengths

VIII WLMLA, Cayo Coco 9th April

Optical properties:

α355,532 largest values at the center of the flame base α1064 is maximum in a ring around the center of the flame base

Larger AE in the bottom of the flame (hot region) than in the top (cold region) nucleation and subsequent coagulation increases as distance from the emission point

* Hot region (<3 pixels surrounding the emitter point)* Cold region (>4 pixels respect to the emitter point)

VIII WLMLA, Cayo Coco 9th April

[Gobbi et al, ACP, 2007] & [Perrone et al. ACP, 2014]: graphical method based on spectral variation of AE (i.e. its curvature) is related to particle size distribution (Rfine & η)

Since 1993 several studies point out the relevance of the spectral variation of the Angström exponent (i.e. its curvature) to obtain information of the corresponding particle size distribution

Hot region particles: mainly 0.15< Rfine <0.25 µm & 50%< η <100% Cold region particles: mainly 30%< η <70% less concentration of fine particles as a consequence of the coarse mode particles generation by coagulationRfine slightly decreasing in cold region higher efficiency of the fine particles with larger sizes to coagulate

The results indicates that to properly use the lidar signals in any lidar procedure (i.e. derivation of range corrected signals plots, Klett-Fernald method and Raman algorithm, among others) some corrections must be applied such as some bins of the AN and PC signals must be removed and dark current must be subtracted.

The case study allowed to illustrate the procedure to retrieve optical and microphysical properties (Gobbi modified graphical method) of particles in flames. This work demonstrates the capabilities of the multiwavelength elastic scanning lidar as a promising tool for industrial flare research.

VIII WLMLA, Cayo Coco 9th April

VIII WLMLA, Cayo Coco 9th April

… thank you very much

More details in:

J.L. Guerrero-Rascado, R. da Costa, A. E. Bedoya, R. Guardani, L. Alados-Arboledas, A. E. Bastidas, and E. Landulfo, Multispectral elastic scanning lidar for industrial flare research: characterizing the electronic subsystem and application, OPTICS EXPRESS, 31063, Vol. 22, No. 25, DOI:10.1364/OE.22.031063 (2014)