Remote sensing of aerosol from the GOES-R Advanced Baseline Imager (ABI)
description
Transcript of Remote sensing of aerosol from the GOES-R Advanced Baseline Imager (ABI)
Remote sensing of aerosol from the GOES-R Advanced Baseline Imager (ABI)Istvan Laszlo1 , Pubu Ciren2, Hongqing Liu2, Shobha Kondragunta1, Xuepeng Zhao3, J. Dan Tarpley1 and Mitch Goldberg1
1Center for Satellite Applications and Research, NOAA/NESDIS, Camp Springs, Maryland2QSS Group, Inc., Maryland
3ESSIC/CICS, University of Maryland, College Park, Maryland
Real time monitoring of aerosol optical depth from Geostationary Operational Environmental Satellite (GOES) data have been routinely conducted at the National Oceanic and Atmospheric Administration (NOAA). The current algorithm uses only a single channel to retrieve aerosol optical depth (AOD), provides no information on particle size, uses a fixed aerosol model and estimates the surface effect with large uncertainty that leads to inaccurate AOD for certain times and regions. The GOES-R Algorithm Working Group (AWG) Application Team (AT) for aerosol at the NOAA Center for Satellite Applications and Research (STAR) has currently started developing methods that address some of the shortcomings of the current GOES aerosol algorithm. This is done by taking advantage of the new capabilities offered by the Advanced Baseline Imager (ABI) onboard GOES-R. These capabilities are similar to the multispectral observations currently provided by the Moderate Resolution Imaging Spectroradiometer (MODIS) flown on the NASA Earth Observing System (EOS) satellites, and to those that will be available from the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Visible/Infrared Imager/Radiometer Suite (VIIRS). In addition, ABI on GOES-R will expand the frequency and coverage of aerosol remote sensing that can support a wide range of weather and environmental applications including the monitoring of air quality.
In the sections that follow shown are:• aerosol retrieval from the current GOES satellite (2)• proposed retrieval over land from ABI (3)• current work for estimating the surface contribution from ABI (4)• aerosol models adopted to provide un-retrieved properties (5)• generation of clear ABI proxy reflectances for aerosol retrieval (6), and• example of VIIRS retrieval for demonstrating ABI capability (7).
1. INTRODUCTION
2. GOES-M AEROSOL RETRIEVAL
ABSTRACT
In this poster the current and planned work at NOAA/NESDIS/STAR on aerosol remote sensing from GOES-R ABI are highlighted. Examples of estimating the visible surface reflectance from the near infrared ABI channel, adaptation of time-space-dependent most-probable aerosol models for prescribing aerosol microphysics, and simulation of clear ABI channel radiances using MODIS-derived atmosphere and surface products are presented.
Clear 0.65 mreflectance
Clear composite of0.65 m reflectance
from previous 21 days
Update clear composite reflectance
Estimate surface albedoassuming AOT=0.02
for every observation time and every 4-km pixel
LUTfor a single
aerosol model
Estimate AOT at 0.65 m
Shortcomings: • Only a single channel is available for retrieving
both aerosol and surface reflectance.• VIS channel is not calibrated onboard.• Single aerosol model used, independent of
time and space.• Variability in gaseous absorption is not
accounted for.
Current Aerosol Retrieval from GOES
Aerosol optical depth derived from GOES-12 visible radiances and provided in the GOES Aerosol and Smoke Product (GASP).
5. AEROSOL MODELS
Aerosol properties that are not retrieved by the algorithm are prescribed from aerosol models. These models are currently an adaptation of the MISR aerosol climatologies.
Left: Aerosol components in the MISR aerosol climatology for (a) January, (b) April (c) July and (d) September. The components are as follows:
1a. Carbonaceous+ Dusty Maritime 4a. Carbonaceous+ Dusty Continental 1b. Carbonaceous+ Dusty Maritime 4b. Carbonaceous+ Dusty Continental1c. Carbonaceous+ Dusty Maritime 4c. Carbonaceous+ Dusty Continental2a. Dusty maritime+ Coarse Dust 5a. Carbonaceous+ Black Carbon Continental2b. Dusty maritime+ Coarse Dust 5b. Carbonaceous+ Black Carbon Continental3a. Carbonaceous+ Black Carbon Maritime 5c. Carbonaceous+ Black Carbon Continental3b. Carbonaceous+ Black Carbon Maritime
Right: Simulated Ch1 (a) and Ch2 (b) TOA reflectances as a function of viewing zenith angle for the 13 MISR aerosol types.
7. VIIRS aerosol product as proxy for ABI
a. c.b. d.
Simulated VIIRS TOA reflectances in channels centered at 0.48um (a), 0.672 um (b) and 2.25um (c), respectively. Figure d) is the retrieved aerosol optical depth at 0.55um. The center wavelength of these three VIIRS channels are very similar to those of channels 1, 2 and 6 of ABI, and the ABI algorithm is based partly on the VIIRS method, so the ABI aerosol retrievals are expected to “look” similar to the VIIRS retrievals.
6. PROXY CLEAR ABI REFLECTANCES
Components:
• MODIS-derived fields of surface reflectance and AOD, O3 and H2O
• Aerosol models• RT model
Advantage:
• Realistic-looking reflectance field with known components → evaluation of algorithm is “easy”.
ABI Clear-reflectance simulator
4. VIS VS. NIR SURFACE REFLECTANCE
Retrieval of surface reflectance follows the MODIS/VIIRS approach: surface reflectance in the shorter-wavelengths “blue” and “red” ABI channels are estimated from TOA reflectance in the NIR, where it is assumed that aerosol has minimal effect on the radiance. For this, the relationship between the blue, red, and NIR channels must be established. Two approaches have been explored:
(a) (b) 1. Relationship derived from spectral reflectance data in the ASTER library:
a) Blue (ABI: Ch1, MODIS: band 3) vs. NIR (ABI: Ch 6, MODIS: band 7) surface reflectance.
b) The same as (a) but for the red channel (ABI: Ch 2, MODIS: band 1).
Solid line shows relation used in the MODIS algorithm.
2. Relationship derived from hyperspectral satellite data:
Simulated ABI TOA reflectances in channels 1, 2, 3 and 6 obtained from Hyperion data. After correction for atmospheric effect, the relationship between NIR and visible surface reflectances can be derived.
ch1 ch2 ch3 ch6 109:30:15 (RGB)
(b)
ch1 ch2 ch3 ch6 50:32:10 (RGB)
(a)
(a) Agricultural site
Location: Kansas/Oklahoma
Target Latitude: 37.476 N
Target Longitude: 97.4223 W
Acquisition Date: March 9, 2002
(b) Forest site
Location: New England
Target Latitude: 43.1861 N
Target Longitude: 72.4335 W
Acquisition Date: February 28, 2002
3. GOES-R AEROSOL RETRIEVAL
calculate pathradiance L0 ratio
estimate surface reflectancesat 0.47 m and 0.64 m
select pixels with Rmin ≤R2.26≤Rmax
and calculate average reflectances
estimate shape of surface BRDFfrom clear-composite reflectances
water pixel
ocean algorithm
coastal pixelproceed with AOT retrieval but degrade quality flag
26.264.047.0 ,, RRR 26.264.026.247.0 , RgRRfRss
Clear ABI reflectances Rat 0.47, 0.64, 2.26 min a box of 10x101-km pixels
derive AOTindependently
at 0.47 m and 0.64 m from space-time dependent
LUT
047.0
064.0
L
Lr
r > rtuse dust model and re-derive AOT
Y
N Y
N
calculate Angstrom exponent
Y
N
Determine if dust
dust slope10.9-12.2 m
radiancesfrom HES
H2O, O3
future modules
Proposed GOES-R Aerosol Algorithm over Land
Multiple channels of ABI in the VIS/NIR offer potential improvements for aerosol remote sensing. Separate algorithms are designed for land and ocean. They combine techniques used in GASP, MODIS, VIIRS and AVHRR aerosol retrievals.
c
b d
1a 1b 1c 2a 2b 3a 3b 4a 4b 4c 5a 5b 5c
a
0 20 40 60 80 100
0.08
0.12
0.16
0.20
0.24
0.28
SZA=20, RAZ=40, Srf=0.0, aod=0.64
TO
A r
efle
ctance
C
H1 (0
.47 u
m)
Viewing zenith angle
model 1a model 1b model 1c model 2a model 2b model 3a model 3b model 4a model 4b model 4c model 5a model 5b
0 20 40 60 80 100
0.04
0.06
0.08
0.10
0.12
0.14
0.16
TO
A r
efle
ctance
(0.6
4 u
m)
Viewing zenith angle
model 1a model 1b model 1c model 2a model 2b model 3a model 3b model 4a model 4b model 4c model 5a model 5b
For algorithm development and testing proxy clear-sky TOA reflectances in the appropriate ABI channels are needed. These are generated using current GOES, MODIS, SEVIRI and AVHRR data. Another possibility is to use satellite-derived atmospheric and surface parameters along with a radiative transfer model. The figure on the left shows such a system.
GOES-R ABI reflectances
6S RT code
MYD09ASurface reflectance
8 day-composite500m by 500m
MYD04_L2Aerosol optical depth
10 km by 10 km
MYD35_L2Cloud mask
1 km by 1km
MYD07_L2Total ozone5km by 5km
MYD07_L2Total water vapor 5 km by 5 km
Ch 1
Ch 2
Ch 6
Common grid