using copernicus atmosphere products to improve the ... · bastien rouquié, olivier hagolle,...

13
Using Copernicus Atmosphere products to improve the estimation of the Aerosol Optical Thickness in MAJA Bastien Rouqui´ e Olivier Hagolle Camille Desjardins Fran¸cois-MarieBr´ eon Olivier Boucher Samuel R´ emy Centre d’ ´ Etudes Spatiales de la Biosphere June 13, 2018 1 / 13

Upload: others

Post on 09-Aug-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Using Copernicus Atmosphere products to improve theestimation of the Aerosol Optical Thickness in MAJA

Bastien Rouquie Olivier Hagolle Camille DesjardinsFrancois-Marie Breon Olivier Boucher Samuel Remy

Centre d’Etudes Spatiales de la Biosphere

June 13, 2018

1 / 13

Page 2: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Aerosol Optical Thickness (AOT) estimation by MAJA

Multi-temporal and multi-spectral criteria

surface reflectances relatively stable with time ⇒ minimize thedifference between the surface reflectance of two consecutive dates.

ρblue = 0.45× ρred ⇒ verify this multi-spectral relationship.

Aerosol type defined as constant with time and location.

arid sites vegetated sites 2 / 13

Page 3: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Copernicus Atmosphere Monitoring Service (CAMS)

Part of the European Union program for environment monitoring

Characteristics

Operational status (access throughECMWF)

2 forecasts per day

C-IFS (Composition-IntegratedForecasting System) model(Morcrette2009)

Assimilating MODIS data(Benedetti2009)

Range of products

air quality

atmospheric composition

solar radiation

climate forcing

ozone layer

surface fluxes

3 / 13

Page 4: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Copernicus Atmosphere Monitoring Service (CAMS)

CAMS forecast AOT on 8 April 2018

AOT forecasts at 550 nm forfive aerosol types:

Dust

SeaSalt

Sulfate

OrganicMatter

BlackCarbon

Total AOT not preciseenough (complex forecastprocesses, RMSE=0.176)⇒ estimate the aerosol type

4 / 13

Page 5: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Using CAMS to estimate the aerosol type

Creation of Look Up Tables (LUT):

for each aerosol typecorrespondence between surface and TOA (Top Of Atmosphere)reflectanceparameterized by AOTSeaSalt, Sulfate and OrganicMatter depend on Relative Humidity(RH) ⇒ one LUT per RH percentage: [30,50,70,80,85,90,95]

LUT used by MAJA

Linear interpolation weighted by CAMS AOT

Aerosol Sulfate SeaSalt OrganicMatter BlackCarbon DustProportion 84% 7% 6% 2% 1%

LUT = 0.84*LUT Sulfate + 0.07*LUT SeaSalt +0.06*LUT OrganicMatter + 0.02*LUT BlackCarbon + 0.01*LUT Dust

5 / 13

Page 6: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Dataset

Sentinel-2A time series: 1 April to 15 October 2016

Tile AERONET Type

32SNE Ben Salem (Tunisia) arid

29SPD Badajoz (Spain) arid

31PDR Banizoumbou (Niger) arid

29RNQ Ouarzazate (Morocco) arid

36RXV Sede Boker (Israel) arid

35JPM Pretoria (South Africa) arid

34LGJ Mongu (Zambia) arid

31TFJ Carpentras (France) vegetated

30TXM Zaragoza (Spain) vegetated

32TPR Sirmione (Italy) vegetated

31TCG Montsec (Spain) vegetated

31TCJ Toulouse (France) vegetated

35TPN Kishinev (Moldova) vegetated

21LWK Alta Floresta (Brazil) vegetated

51NXH Marbel (Philippines) vegetated

6 / 13

Page 7: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

AOT validation with AERONET - arid sites

Stable cases

temporal stability of AERONET observations

no more than 10% of clouds within a 10-km neighborhood

(a) CAMS (b) constant aerosol (c) 5 aerosol types

⇒ RMSE reduced by 28% when using 5 CAMS aerosol types

7 / 13

Page 8: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

AOT validation with AERONET - vegetated sites

(a) CAMS (b) constant aerosol (c) 5 aerosol types

RMSE slightly increased, but same order of magnitude

Passing from a constant aerosol to a variable one ⇒ additional noise

Constant aerosol gives good performances: standard continentalmodel designed to optimize performances over vegetated sites.

8 / 13

Page 9: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Smoothness of surface reflectances

Surface reflectances relatively stable with time ⇒ evaluate the noise

Noise criterion defined by Vermote2009√∑n−2i=1 (ρi+1 − ρi+2 − ρi

di+2 − di(di+1−di )−ρi )2

n−2

ρi , ρi+1 and ρi+2: surface reflectances of dates di , di+1 and di+2

Assumes a linear variation of surface reflectance within a few daysMaximum 20 days between i and i + 2

B2 (496 nm) B3 (560 nm) B4 (664 nm) B8 (832 nm) B11 (1613 nm) B12 (2198 nm)

constant aerosol 0.005 0.007 0.011 0.015 0.018 0.0155 CAMS aerosol types 0.005 0.007 0.011 0.015 0.018 0.015

Table: Noise criterion of MAJA surface reflectance time series, for arid sites.

B2 (496 nm) B3 (560 nm) B4 (664 nm) B8 (832 nm) B11 (1613 nm) B12 (2198 nm)

constant aerosol 0.006 0.007 0.010 0.022 0.019 0.0155 CAMS aerosol types 0.006 0.007 0.010 0.022 0.019 0.015

Table: Noise criterion of MAJA surface reflectance time series, for vegetated sites.

9 / 13

Page 10: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Improvements over a specific site

31PDR - Banizoumbou - Niger

Aerosol Optical Thickness

RMSE of MAJA AOT vs AERONET

constant aerosol 0.4645 CAMS aerosol types 0.210

Noise criterion

B2 (496 nm) B3 (560 nm) B4 (664 nm) B8 (832 nm) B11 (1613 nm) B12 (2198 nm)

constant aerosol 0.004 0.007 0.016 0.017 0.025 0.0235 CAMS aerosol types 0.006 0.008 0.012 0.012 0.019 0.019

10 / 13

Page 11: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Additional use of CAMS AOT

Use CAMS AOT as default value: replace gap-filling method byCAMS AOT values, for pixels with clouds, snow, water.Introduce CAMS AOT in cost function: small weighting coefficient

Dates with at least 75% of gap-filled pixels around AERONET site.

Reference With CAMS AOT 11 / 13

Page 12: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Conclusions

New features

constrain the aerosol type based on CAMS AOT

Rouquie et al. (2017)Using Copernicus Atmosphere Monitoring Service Products to Constrain theAerosol Type in the Atmospheric Correction Processor MAJA.Remote Sensing 9(12), 1230.

use CAMS AOT as default value

introduce CAMS AOT in inversion cost function

Summary of results

Better AOT estimation

Smoothness of surface reflectances unchanged

No need to select the aerosol type in advance

12 / 13

Page 13: Using Copernicus Atmosphere products to improve the ... · Bastien Rouquié, Olivier Hagolle, Camille Desjardins, François-Marie Bréon, Olivier Boucher, Samuel Rémy Created Date:

Conclusions

Perspectives

New features implemented in MAJA V3 operational version

Processor available from https://logiciels.cnes.fr

Integration within THEIA processing: October 2018

Continuous improvements of CAMS products:

forecasting systemaerosol optical properties

13 / 13