Presentation is loading. Please wait.

Presentation is loading. Please wait.

Automatic near-real time mapping of MERIS data in support of CASIX, DISMAR and as input to a phytoplankton classifier Jamie Shutler, Peter Miller, Steve.

Similar presentations


Presentation on theme: "Automatic near-real time mapping of MERIS data in support of CASIX, DISMAR and as input to a phytoplankton classifier Jamie Shutler, Peter Miller, Steve."— Presentation transcript:

1 Automatic near-real time mapping of MERIS data in support of CASIX, DISMAR and as input to a phytoplankton classifier Jamie Shutler, Peter Miller, Steve Groom and Jim Aiken jams@pml.ac.uk jams@pml.ac.uk http://www.pml.ac.uk Plymouth Marine Laboratory, U.K. HAB Low High

2 Overview What are CASIX and DISMAR ? Why are we interested in algal species ? Approach taken: –MERIS 7 day rolling archive. –Brockman Consult’s MERCI. –Near-real time mapping. Multivariate spectral classifier. Initial results. Ongoing developments. Conclusions.

3 Centre of excellence for observation of Air-Sea Interactions and fluXes - CASIX U.K. National Environment Research Council centre of excellence. Virtual centre consisting of the U.K. MET office and leading UK oceanographic institutions (3) and university departments (6). Remit: Exploit new-generation Earth observation data to advance the science of air-sea interactions towards reducing the errors in the prediction of climate change. Cruises require the timely supply of near-real time data.

4 Data Integration System for Marine Pollution and Water Quality - DISMAR Information Society Technologies (IST) funded project under FP6. Objective: to develop an advanced (intelligent) information system for monitoring and forecasting the marine environment – pollution To support: public administrations and emergency services. Users require the timely supply of near-real time data for decision making.

5 Why are we interested in algal species ? Water quality monitoring. e.g. harmful algal bloom detection. Providing data for coastal observatories. Guidance and support of research vessels. Species of current interest: –Karenia mikomotoi –Cyanobacteria

6 Approach MERIS 7 day rolling archive. Global level 2 data automatically downloaded. Data enrolled into Brockman Consult MERCI. Data converted into hdf formats and full resolution geolocation generated. Data mapped to standard areas and made available on the web. The use of a GRID engine enables multiple full orbit passes to be processed simultaneously. Water leaving radiance data used as input to a multivariate spectral classifier.

7 MERis Catalogue and Inventory MERCI Ability to search database dependent on spatial and temporal search criteria. Quick look images of scenes. Ability to view coverage of a pass. Easy identification of time-series data. Brockman Consult, Germany.

8 Near-real time downloading and mapping

9 Multivariate discrimination Ocean colour scenes SYSTAT Multivariate analysis Classifier Manual training No Bloom HAB Harmless algae Training samples Lwn(λ) a(λ) bb(λ) Ocean properties

10 Karenia ground truth data Kelly-Gerreyn, B.A., M.A. Qurban, D.J. Hydes, P. Miller, and L. Fernand, Coupled “FerryBox” Ship of Opportunity and satellite data observations of plankton succession across the European Shelf Sea and Atlantic Ocean, in International Council for the Exploration of the Sea (ICES) Annual Science Conference, Vigo, Spain, 2004. Sample stations on Research Vessel Corystes 26 June – 09 July 2003 Identified as Karenia mikimotoi with Chl-a > 50 mg m -3

11 Karenia HAB Initial MERIS results 11:04 UTC 27 June 2003 Karenia HAB HAB likelihood High Low True colour composite

12 Initial MERIS results 10:39 UTC 01 July 2003 HAB likelihood High Low True colour composite Karenia HAB Karenia HAB

13 Initial MERIS results 11:24 UTC 20 July 2003 HAB likelihood High Low True colour composite Problems: Intense Coccolithophore blooms

14 Combining spectral and spatial approaches J. D. Shutler, M. G. Grant and P. I. Miller, Towards spatial localisation of harmful algal blooms; Statistics- based Spatial anomaly detection, SPIE Remote Sensing Europe 2005 (Image and Signal processing for remote sensing XI), Belgium, September 2005. Automatic mapping of Coccolithophore phytoplankton bloom extent to remove false positives. South west approaches, U.K. 13:48 UTC 15 June 2004 P.I. Miller, J.D. Shutler, G.F. Moore, and S.B. Groom, SeaWiFS discrimination of harmful algal bloom evolution,, International Journal of Remote Sensing, in press.

15 Using these data in DISMAR Interactive online GIS system allowing coincident viewing of multiple data sets on the same projection. http://dispro.ucc.ie

16 Future developments Investigation of spatial techniques to reduce false positives. Further training. Development of the Cyanobacteria classifier is ongoing. MERIS full resolution data would enable more accurate mapping of algal concentrations. –Important for coastal monitoring applications providing timely information allowing HAB detection and preliminary species identification.

17 Conclusions Developed a near-real data supply system for researchers within the CASIX. –Support of research cruises –Coastal observatories (monitoring and research) Mapped data are made available within half an hour of reception. Data are currently providing input to a spectral classifier trained to distinguish algal types. Data form a test input into a experimental IST (FP6) water quality monitoring system - DISMAR

18

19 Initial results: SeaWiFS HAB discrimination SeaWiFS enhanced ocean colour scenes showing harmful algal blooms HAB England 20 Jul. 2000 HAB HAB likelihood: LowHigh Baltic Sea 05 Jun. 2002 HAB Miller, P.I., J.D. Shutler, G.F. Moore, and S.B. Groom, SeaWiFS discrimination of harmful algal bloom evolution, International Journal of Remote Sensing, in press.


Download ppt "Automatic near-real time mapping of MERIS data in support of CASIX, DISMAR and as input to a phytoplankton classifier Jamie Shutler, Peter Miller, Steve."

Similar presentations


Ads by Google