Monitoring of Lake Water Quality Using Hyperspectral CHRIS

ESA CHRIS-PROBA Workshop, Frascati, 28.�30.04.2004 GeoFoschungsZentrum Potsdam

Monitoring of Lake Water Quality Using HyperspectralCHRIS-PROBA Data

Sandra Mannheim, Birgit Heim, Karl Segl, Hermann Kaufmann

GeoForschungsZentrum Potsdam (GFZ)Department of Geodesy and Remote Sensing

Section 1.4 Remote Sensing TelegraphenbergD-14473 Potsdam

Tel.: ++49-331-288-1198, Fax: ++49-331-288-1192, Email: [email protected]

IaGB


Monitoring of Lake Water Quality Using HyperspectralCHRIS-PROBA Data

23. April 2003 10. August 2003 26. September 2003

Overview:

� Introduction� Radiometric correction (destriping and atmosphere)� Extraction of water parameters� Experimental results� Conclusions


Motivation and Objectives

Need for

Water quality monitoring of inland water bodies according to Europeanand German (EU Water Framework 2000, LAWA) water directives

Goals

Investigation of seasonal and spatial variations of water quality parameters

Quantification of the trophic parameters� chlorophyll-a� secchi depth� dissolved organic carbon (DOC) � suspended matter (SPM)

Determination of phytoplankton composition (green algae, diatoms, cyanobacteria, cryptophyta)


565Mazurian2003-09-11

660Mazurian2003-09-10

40Rheinsberg2003-09-26



60Mueritz2003-08-11

62Mueritz2003-08-09

610Mueritz2002-10-10

Ground truth points

Cloud cover [%]Test siteDate

565Mazurian2003-09-11

660Mazurian2003-09-10




60Mueritz2003-08-11

62Mueritz2003-08-09

610Mueritz2002-10-10

Ground truth points

Cloud cover [%]Test siteDate

CHRIS-PROBA Acquisition


München

Rostock

Germany

Berlin

0 2 4 km

Test Sites

Rheinsberg lake district

� located about 70 km northwest of Berlin

� covers about 300 km²� includes more than 30 lakes with

various depths, sizes and trophicstates

� formed by the last ice age


Ground Truth

Field spectroscopy

Field spectrometer FieldSpec Pro (ASD)Spectral range: UV to SWIR (350 � 2500 nm)

� Above water measurements of water albedo R(λ), down-welling Ldown(λ) and up-welling radiance Lup(λ), as well as sky radiance LSky

� Observation angles of 45° in zenith and 135° in azimuth direction (Ocean Optics Protocols)

Sea truth

� secchi depth� chlorophyll-a� dissolved organic carbon � suspended matter

� phytoplankton composition� phytoplankton morphology (spherical, bacillary)

+ +


Radiometric Correction of CHRIS-PROBA Data

� strong artefacts (stripes) in along track direction

2003-04-23, Band 2

7 %

6 %


Iterative Radiometric Correction

Correction model (column by column):

Assumption:

Least squares solution:

� averaging only similar pixels (75%)

Estimation of


Radiometric Correction

Adoption 1: grey value extrema Adoption 2: sensor errors

Solution:� selection of similar pixels� normalization of grey values

to standard thresholds

Solution:� no similar pixels� standard processing


Radiometric Correction


Atmospheric Correction

Ze1

Sw1

V1

Wu1

Z1 Zo1

Empirical line method

1. Correction of master scene using in-situ reference spectra

2. Correction of remaining scenes using reference spectra (dark dense vegetation, artificial surfaces) extracted from the master scene

Reflectance spectra

500 600 700 800 900 1000 nm

0.02

0.04

0.06

0.08Ze1V1Sw1Z1Zo1Wu1


Extraction of Water Parameters

400 600 800 1000 1200Wavelength [nm]

Rheinsberger SeeSchwarzer SeeZootzenseeBraminsee

CHL: ChlorophyllCAR:CarotenoidesPC: Phycocyanin

0.01

0.02

0.03

0.04

0.05

Ref

lect

ance

CH

LC

AR

CH

LP

C

ASD reflectance spectra 10th August 2003Lake


Automated Extraction of Water Parameters

Spectral library

Ze1V1Sw1Z1Zo1Wu1

Ground truth data

2Zo1

5Z1

21Ze1Chl-aPoint

Automated feature correlation

Controlling parameters

� spectral features� wavelength range� step size

Best correlation result / feature


Automated Extraction of Water Parameters

Mean value Maximum/Minimum

Polynomial adaption1th order

Polynomial adaption2th order

Absorption depth +/-

-+

RatioStandard deviation

Area

+-

λ λ λ λ

λ λ λ λ

Spectral Features


Extraction of Chlorophyll-a

chan

nel8

chan

nel1

2

Seasonal Approach

AreaChl-aLake

AreaChl-aLake

-0,26-0,17

-0,45-0,28

17 µg/lSchwarzer See34 µg/lZethner See

September

21 µg/lZethner See14 µg/lSchwarzer See

August

Wavelength [nm]

Ref

lect

ance


Extraction of Chlorophyll-a

R2 = 0,95

R2 = 0,97

5

10

15

20

25

30

35

40

-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1Feature values

µChl-a [ g/l]

Seasonal Approach

Chl-a summerChl-a spring / autumn


Results

0 - 3>3 - 10>10 - 18>18 - 33>33 - 60>60 - 105>105

Chl-a content [µg/l]

Limnological classificationaccording toGerman lake directive (LAWA)

� Chlorophyll map extracted from CHRIS PROBA data

2004-08-10


Results

0 - 3>3 - 10>10 - 18>18 - 33>33 - 60>60 - 105>105

Chl-a content [µg/l]

Chlorophyll map Topographic base

� Observation of spatial variations within lakes


ResultsSeptemberApril August

limnological summer average

0 - 3>3 - 10>10 - 18>18 - 33>33 - 60>60 - 105>105

Chl-a content [µg/l] � Monitoring of the seasonal chlorophyll dynamic of the lakes

� Evaluation of thelimnological summer average


Conclusion and Outlook

CHRIS-PROBA data have high potential for monitoring lake water quality

Successful calibration is possible

Seasonal chlorophyll quantification with high accuracy

Further works

Adaptation and validation of the developed seasonal chlorophyll algorithms

Quantification of the additional water quality parameters

Determination of the trophic state and evaluation of the results in cooperation with limnologists

Documents

Monitoring of Lake Water Quality Using Hyperspectral CHRIS