Spatial Variability and Vertical Distribution of Aerosols Over an - - PowerPoint PPT Presentation

Spatial Variability and Vertical Distribution of Aerosols Over an Eastern Mediterranean Region Using LIDAR and MODIS/MAIAC-Measurements. Haifa as a Case Study Alexandra Chudnovsky1, Albert Ansmann2, Irina Rogozovski1, Holger Baars2, Alexei Lyapustin3, and Yujie Wang3

1Porter School of Environment and Earth Sciences, Air-O Lab, Tel-Aviv University 2Leibniz Institute for Tropospheric Research, Leipzig, Germany 3GEST / UMBC, NASA Goddard Space Flight Center, Baltimore, MD, USA

10th International Conference on Urban Climate / 14th Symposium on the Urban Environment August 6-10, NY

Strength of Remote Sensing for air quality studies

 Existing PM2.5 ground-monitoring networks encompass a

relatively small number of stations per urban area, which is not sufficient to characterize spatial variability of particles within a large metropolitan area

 Expanding spatial coverage and “filling gaps” in areas where

there are no ground sensors (much of the third globe)

 A cost effective way of obtaining global information about the

earth-atmosphere system

 Event and emission identification and transport, atmospheric

composition determination, monitoring of long range transport events (forest fires, volcanic plumes, and stagnant haze masses

• ver highly industrialized areas) influencing local air quality

Objectives

 To describe the aerosol pollution state in the vertical

column in the PBL

 Studying the long-term regional variability of different

natural and anthropogenic aerosol components

 Buildup of an aerosol lidar center at TAU in the

framework of a long-term collaboration with TROPOS and its integration into the world-leading community

June 10, 2003 A A 0.8 AOD

MAIAC 1 km (left column) and MOD04 10 km (right column) , representing low pollution (A) and moderate pollution (B) days as measured by PM2.5. Note the loss of AOD variability on conventional images (right column). Consider left column only next. Image A and B show dynamic variation for a given region, regardless of pollution.

1 C 10 km 1 km June 17, 2003 10 km June 17, 2003 1 km June25, 2003 B A Boston Boston Boston June 25, 2003 10 km B A Boston

Chudnovsky, A., Kostinski, Lyapustin, A., and Koutrakis, P. (2013). Spatial scales of pollution from variable resolution satellite imaging. Environmental Pollution, Vol. 172, pp. 131-138.

Previous work

Collaboration with TROPOS and NASA Building 3D profile: understanding better the

• rigin of aerosols

Detailed correlation study: MAIAC (AOD) vs surface observation (PM), and AERONET (AOD) Statistical modelling of the

• correlation. What parameters and

conditions impact the most satellite retrieval error?

Merging different data sets

Basic Idea: Desert regions are darker at shorter wavelengths so aerosol show up better.

Satellite retrievals

Technion, LiDAR site

New Version of Existing Satellite Data

Towards a more Generalizable Linkage

Spatial Resolution

Lyapustin, A., Martonchik, J., Wang, Y., Laszlo, I., and Korkin, S (2011a). Multiangle implementation of atmospheric correction (MAIAC): 1. Radiative transfer basis and look-up tables. J. Geophys. Res., 116(D03210).

2000- current

Methodology

Flow chart that illustrates the overall methodology of this study. AERONET (AOD 440 nm) Ground Stations (PM2.5, PM10) MAIAC (AOD 470 nm) Polly LiDAR Ground Monitoring Establishment of the spatial-temporal aerosol distribution over Gush Dan area Data merging with different sources Establishment of a 3D spatial-temporal aerosol distribution over Haifa Data processing and screening Satellite data Comparison to the vertical profile of the atmosphere Good MAIAC/AERONET correlation Poor MAIAC/AERONET correlation Data collection Models HYSPLIT Synoptic patterns (Based on Osetisnsky, Alpert)

• A. Missing data
• B. Cloudy days
• C. Included data: polarization

MODIS MAIAC AOD MODIS RGB

0.00 - 0.05 0.05 - 0.10 0.10 - 0.15 0.15 - 0.20 0.20 - 0.25 0.25 - 0.30 0.30 - 0.35 0.35 - 0.40 0.00 - 0.05 0.05 - 0.10 0.10 - 0.15 0.15 - 0.20 0.20 - 0.25 0.25 - 0.30 0.30 - 0.35 0.35 - 0.40 0.00 - 0.05 0.05 - 0.10 0.10 - 0.15 0.15 - 0.20 0.20 - 0.25 0.25 - 0.30 0.30 - 0.35 0.35 - 0.40

Panel A (03.08.2012) represents missing data were the white circle highlights the area with no retrieved data; Panel B (30.08.2012) demonstrates a cloudy day and where the algorithm cloud mask was applied on MAIAC retrieval data and blocked a significant area of the image. Note high AOD values in the vicinity of the cloud mask (highlighted by white arrows); Panel C (08.08.2013) shows residue calibration errors in the original MODIS L1B data, related to the MODIS mirror-side difference at 10km (polarization effects). Note that data presented in Panel C was included in the analyses. Marked in a black box is the research area.

Data availability and data selection for the analyses

Sever et al. 2017, Atmospheric Environment

Long term analyses of data: Correlation between MODIS MAIAC and AERONET (arid) Sede Boker station (2001-2014)

12 Month Terra Aqua Jan 0.60 0.49 Feb 0.68 0.71 Mar 0.82 0.82 Apr 0.71 0.86 May 0.76 0.64 Jun 0.48 0.62 Jul 0.65 0.68 Aug 0.70 0.63 Sep 0.95 0.95 Oct 0.70 0.66 Nov 0.59 0.47 Dec 0.58 0.56 Total 0.81 0.82

0.0 0.2 0.4 0.6 0.8 1.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Percent Month Terra Aqua

0.64 0.77 0.62 0.91 0.62 0.79 0.64 0.91

0.0 0.2 0.4 0.6 0.8 1.0 Winter (DJF) Spring (MAM) Summer (JJA) Fall (SON)

Percent Month Terra Aqua

Panel A Panel B Correlation (R) between MODIS MAIAC (470nm) and AERONET (extrapolated 470nm). AERONET AOD values represent a

• ne-hour time interval, from half an hour

before the satellite pass-time to half an hour

• afterwards. MAIAC’s data represent an

average AOD for an area of 3x3 pixels, whereby the Sede Boker AERONET station in located in the central pixel. The monthly (panel A) and thereafter the seasonally (panel B) correlations are for the years 2000-2015 for Terra and 2002-2015 for AQUA. All presented correlations are based on daily data comparisons between MAIAC and

• AERONET. The values of the monthly

correlations are presented in the adjacent table.

Sever et al. Submitted

Monthly trend of AOD for Nes Ziona: comparison between AERONET and MAIAC AOD retrievals

Nes Ziona- April 2013 Nes Ziona- July 2013 Nes Ziona- August 2013 Nes Ziona- November 2013

Study area. Left: a map of Israel, with the Haifa region highlighted by the dashed line (major cities and the AERONET station marked); Right: an enlargement of the study area (AERONET station and PM2.5/PM10 stations marked).

Study area

Average AOD spatial variability for selected months: August and October 2016

Terra and Aqua

y = 1.121x + 0.021 R² = 0.74 0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 AERONET AOD (470 nm) MODIS AOD (470 nm)

AQUA

Underestimation Overestimation

y = 0.89x + 0.048 R² = 0.72 0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 AERONET AOD (470 nm) MODIS AOD (470 nm)

TERRA

Underestimation Overestimation

AERONET AOD MODIS MAIAC AOD

PM2.5 concentrations (µg/m3)

5 -12.5 12.5 -36.5 0.45 -1.0 1.01 - 1.35 1.35 - 1.7 440-675 Angstrom

Anthropogenic pollution

Good correspondence Bad Days: August 13, October 1 25-26 October, 29 September 3 October Ground pollution + lofted dust

AERONET AOD MODIS MAIAC AOD Wind direction 25-26 October

100.0 0.0 50.0 4000 500 1000 1500 2000 2500 3000 3500 Time (UTC) 08:00:22 10/25/2016 07:00:00 10/25/2016 07:10:00 10/25/2016 07:20:00 10/25/2016 07:30:00 10/25/2016 07:40:00 10/25/2016 07:50:00 10/25/2016 Range-corrected signal, 532 nm 100.0 0.0 50.0 4000 500 1000 1500 2000 2500 3000 3500 Time (UTC) 12:00:00 10/25/2016 11:00:00 10/25/2016 11:10:00 10/25/2016 11:20:00 10/25/2016 11:30:00 10/25/2016 11:40:00 10/25/2016 11:50:00 10/25/2016 Range-corrected signal, 532 nm

Polly lidar, Technion, Haifa, Israel 25 Oct, 07-08 UTC 25 Oct, 11-12 UTC

4 3 2 1 4 3 2 1

Altitude, km Altitude, km

Residual pollution layer Mixing layer Mixing layer Residual pollution layer

532 nm

Time Time

Polly lidar, Technion, Haifa, Israel

532 nm

100.0 0.0 50.0 4000 500 1000 1500 2000 2500 3000 3500 Time (UTC) 12:00:00 10/26/2016 10:00:00 10/26/2016 10:15:00 10/26/2016 10:30:00 10/26/2016 10:45:00 10/26/2016 11:00:00 10/26/2016 11:15:00 10/26/2016 11:30:00 10/26/2016 11:45:00 10/26/2016 Range-corrected signal, 532 nm 100.0 0.0 50.0 5000 500 1000 1500 2000 2500 3000 3500 4000 4500 Time (UTC) 14:29:12 10/26/2016 12:30:00 10/26/2016 12:45:00 10/26/2016 13:00:00 10/26/2016 13:15:00 10/26/2016 13:30:00 10/26/2016 13:45:00 10/26/2016 14:00:00 10/26/2016 14:15:00 10/26/2016 Range-corrected signal, 532 nm

26 Oct, 10-12 UTC

26 Oct, 12:30-14:30 UTC

Altitude, km Altitude, km Lofted layers of dust and pollution

Residual pollution layer Marine PBL Marine PBL Residual pollution layer

Lofted layers of dust and pollution

4 3 2 1

5 4 3 2 1

Time Time

October 26, 2016

AOD 0.44 Fine Fraction AOD 1

532 nm

Polly lidar, Technion, Haifa, Israel

100.0 0.0 50.0 4000 500 1000 1500 2000 2500 3000 3500 Time (UTC) 08:00:00 10/03/2016 07:01:30 10/03/2016 07:10:00 10/03/2016 07:20:00 10/03/2016 07:30:00 10/03/2016 07:40:00 10/03/2016 07:50:00 10/03/2016 Range-corrected signal, 532 nm 100.0 0.0 50.0 4000 500 1000 1500 2000 2500 3000 3500 Time (UTC) 12:00:00 10/03/2016 11:00:00 10/03/2016 11:10:00 10/03/2016 11:20:00 10/03/2016 11:30:00 10/03/2016 11:40:00 10/03/2016 11:50:00 10/03/2016 Range-corrected signal, 532 nm

3 Oct, 07-08 UTC 3 Oct, 11-12 UTC

Not well-mixed dust + pollution layer

4 3 2 1 4 3 2 1

Altitude, km Altitude, km Time Time

Conclusions

• Correspondence between MODIS MAIAC and

AERONET AOD. MAIAC gives an area measure (a 1x1 km pixel) and AERONET's AOD is based on narrow beam measurements

• Merge with active sensors is necessary to investigate

different synoptic conditions and pollution layering

• Noise of high resolution data
• The calculation of the 5 x 5 km box AOD averages is

preferable in presenting of AOD statistics if 1 km AOD data is under investigation.

• More data need to be processed and analyzed and

algorithm improvements are needed

Acknowledgments

• The Environment and Health Fund, (EHF) Israel
• Students: Irina Rogozovski (MA), Lee Sever (PhD),

Ran Pelta (MA)

• Collaborators: Albert Ansmann (TROPOS), Alexei

Lyapustin (NASA), Yujie Wang (NASA), Dietrich Althausen (TROPOS), Birgit Hess (TROPOS)

Thank you for attention

• Contact information:
• achudnov@post.tau.ac.il
• achudnov@hsph.harvard.edu,
• Alexandra Chudnovsky, Ph.D

Porter School of Environment and Earth Sciences, Air-O laboratory, Faculty of Exact Sciences, Tel-Aviv University, Israel