GloboLakes: Global Observatory of Lake Responses to Environmental Change

Steve Groom, PML

sbg@pml.ac.uk

GloboLakes is a 5 year (2012 -2017) €3 million UK project to investigate the state of lakes globally using EO data & investigate their response to environmental change drivers

Introduction and consortium

Andrew Tyler, Peter Hunter, Evangelos Spyrakos: University of Stirling, UK

Steve Groom, Victor Vicente-Martinez, Gavin Tilstone, Giorgio Dall’Olmo: Plymouth Marine Laboratory, UK

Christopher Merchant, Stuart MacCallum: University of Edinburgh/University of Reading, UK

Mark Cutler, John Rowan, Terry Dawson, Eirini Politi: University of Dundee, UK

Stephen Maberly, Laurence Carvalho, Stephen Thackery, Alex Elliott: Centre for Ecology & Hydrology, UK

Claire Miller, Marion Scott, Ruth Haggarty: University of Glasgow, UK

Overview

• Rationale and Approach

• In situ sampling

• How will GloboLakes be of benefit for Sentinel 2 programme (and vice versa)

• S2 MSI simulations

• Future plans

• LIMNADES

• >300 million lakes globally – Provide essential ecosystem goods & services – Fundamental to global food and water security – Important in global biogeochemical cycling – Yet only <0.0001% sampled regularly

• Aim to observe from EO data ~1000 lakes spanning all climatic zones

• Timeliness – why now? – Increasing robustness of EO algorithms & ensemble

approaches – Impending launches of Sentinel 2 MSI and 3 OLCI offering

superior capabilities (2015) – By 2017 will be 20 years of ocean colour observations

Rationale

SeaWiFS: 1997-2010

MERIS: 2002-2012

MSI - Sentinel 2: 2015

OLCI -Sentinel 3: 2015

Lake Balaton, Hungary

Ice Cover

• Set up a satellite-based observatory with archive and near real-time (NRT) data processing for 1000 lakes globally

• Process archive/new data to produce time series of: i) Lake Surface Temperature; ii) Chl-a; iii) phycocyanin concentration; iv) Total Suspended Matter v) Coloured Dissolved Organic Matter • Investigate spatial & temporal trends & attribute

causes of change for 1000 lakes worldwide

• Investigate lake sensitivity to environmental change • Distribute data freely to external project partners

and end-users via web portal and ftp • Opportunistically investigate Sentinel 2 MSI

Our approach

Project structure

WP1: EO algorithms for lakes (Stirling)

WP2: Operational NRT processing

(PML)

WP3: Climate and catchment data

(Dundee)

WP4: Data integration and uncertainty assessment (Stirling/Glasgow)

WP5: Spatial and temporal patterns

(Glasgow)

WP6: Attributing causes (CEH)

WP7: Forecasting

models (CEH)

WP8: End-user applications for lake management (Stirling – All)

2012-15

2013-17

2013-15

2014-16

2012-17

EO model development and processing

Quality assurance

Data analysis and interpretation

End-user engagement

ESA ARC-Lakes

(Reading)

ESA DIVERSITY II (Brockman Consult)

Intercomparison and operationalisation of algorithms using data from a hierarchical populations of lakes

Level 1. Well understood, with excellent sampling ~15 lakes

Level 3. Unknown characteristics, validation data very limited or absent ~1000 lakes

Level 2. In-situ biogeochemical data available for validation ~50 lakes

Hierarchy of Study Lakes

WP1: GloboLakes in situ optics and biogeochemical data for algorithm development

WP1/2: Collaborator provided in situ biogeochemical data and high temporal resolution buoy data for validation

WP2: Output: Level 2/3 data provided to users

Field parameters Bio-optical parameters Laboratory parameters Secchi disk depth Remote sensing reflectance (Satlantic

HyperSAS; Trios RAMSES) Chl-a (spectrophotometric ISO method)

Water depth Subsurface irradiance reflectance (Satlantic HyperOCRs)

Size-fractionated Chl-a

Water temperature Spectral absorption coefficients (Wetlabs AC-S; PSICAM)

HPLC pigments (CSIRO method)

Wind speed

Spectral attenuations coefficients (Wetlabs AC-S)

Phycocyanin (adapted from Horváth et al., 2013)

Digital photos (water & sky conditions)

Spectral backscattering coefficients (Wetlabs BB3)

CDOM (a200-800 & SCDOM)

Aerosol optical depth (using a Microtops)

Temperature, depth, salinity profiles (Sea-Bird Electronics)

CDOM synchronous fluorescence scans (Universidade de Vigo)

TSM, PIM, POM (REVAMP protocol)

Particulate absorption (NASA Ocean Optics method)

DOC (Shimadzu TOC-VSCN)

POC (Perkin Elmer CHN analyzer)

Phytoplankton samples (preserved in Lugol’s)

Flow cytometry (preserved)

Sampling parameters

• Aim to measure as many parameters as feasible optical closure

• UK lakes: • Loch Leven (4 times),

Loch Lomond (7), Windermere (2), Bassenthwaite (1) & Derwent Water (1)

• Range of sizes: 5-71km2; oligotrophic to eutrophic; range of depths

• Hungary • Lake Balaton (4) & Kis Balaton (1) Hungary

Sampling in 2013

UK Lakes

• 82 stations sampled; 230 water samples; 520 optics casts

Results from 2013

Biogeochemical Summary: TSM 0.4 -10 mg L-1; chl-a 6 - 100 mg m-3; PC: 0 – 51 mg m-3; DOC 1.4 – 7 mg L-1

atot aph aNAP

[m-1

] N=35 N=17

N=7 N=5 N=5

N=11

Absorption Summary:

• Analysis is at an early stage

• Unfortunately there were no appropriate satellite observations

• GloboLakes team are a Sentinel 3 Validation Team member

• Radiance data will be used to simulate S2 MSI capabilities for retrieval of in-water properties, complementing optical modelling

• Are there sufficient bands to retrieve: SPM; chl-a; PC and over what conditions and sensitivity?

• What is the sensitivity to retrievals i.e. spectral characteristics; radiometric resolution; S/N

• Airborne hyperspectral data will be used to simulate S2 MSI to

test atmospheric correction methods

How will these data benefit S2 MSI

S2 MSI simulations

Turbid lake

Mesotrophic water 560nm

Extended in Europe through EC FP7 INFORM Mid May Loch Lomond, UK Early June Lake Geneva, Switzerland Mid June Loch Ness, UK July Lake Balaton, Hungary (EC FP7 INFORM/EUFAR) Mid August Cumbrian Lakes, UK Mid September Mantua Lakes, Italy (INFORM/EUFAR) Opportunistic Loch Leven Opportunistic Loch Lomond Summer NERC Airborne Remote Sensing flights (Leven, Lomond, Lough Neagh/ Windermere) Specim Fenix 400 – 2500nm Specim Owl superspectral Thermal IR

Proposed sampling for 2014

• Implement existing Diversity II processing chain (courtesy of Brockmann Consult) and re-produce Diversity II results

• Test existing and additional bio-optical algorithms using modelling, in situ data and simulated satellite reflectances (with uncertainty propagation studies)

• Diversity II models • aLMI (Brando) • NIR-red 2-/3-/4- band models • New MPH algorithm (Matthews) • New NN (with U. Vigo) • PC models (Simis, Hunter, Li etc.) • ….?

• Test water optical type classification scheme (a priori) and/or ensemble model (blending?) (a posteriori)

Proposed algorithm work plan 2014

Sentinel 3 validation… Opportunistic sampling targeting S2 overpasses (and S3 if available) with 24 h mobilization in weather windows) • Loch Leven • Loch Lomond Mid July Lough Neagh (NI) Mid August Lake Vanern (Sweden; INCIS-3IVE) Installation of radiometers on UKLEON buoy (Loch Leven or Loch Lomond; TBC) Real-time Chla, turbidity and CDOM from (at least) 11 UKLEON buoys (8 on lakes > 0.5 km2)

Proposed sampling for 2015

• >110 lakes • 20 countries • 5 continents

Lake bIo-optical MeasuremeNts And matchup Data for rEmote Sensing (LIMNADES) Community-owned database of AOPs, IOPs, constituent concentrations and ancillary data

LIMNADES

• GloboLakes is aiming to

• Develop improved approaches to retrieve bio-optical parameters over lakes globally

• Undertake archive and near-real time processing to produce a global time series in >1000 lakes worldwide

• Opportunistically use of Sentinel 2 MSI for observing key smaller lakes particularly in UK

• Other work (PML)

• EC FP7 INFORM: lake primary production and phytop. size classes • EC FP7 EartH2Observe: focus on high CDOM + cyanobacterial lakes in

Estonia • PML internal: S2 and airborne RS atmospheric correction

Thank you!

www.globolakes.ac.uk

Algorithm validation in Balaton

Four AC models • MERIS MEGS (Standard

ESA) • NASA SeaDAS • CoastColour (Brockman

Consult lead ESA project) • SCAPE-M (Guanter et al.,

2010)

Atmospheric correction validation

LSWT – Phase-2

Annual LSWT Range

Mean August LSWT

LSWT Phase-2

LSWT – Phase-3

Annual LSWT Range

Mean August LSWT

LSWT – Phase-2 LSWT Phase-3

Satlantic HyperSAS radiometers Surface and sky radiances; solar irradiance Remote sensing reflectance (Rrs(0+)) Tilt and heading sensor Wetlabs AC-S in situ spectrophotometer Spectral attenuation (c) Spectral absorption (a) Spectral scattering (b = c – a) 80 channels: 400-730 nm

Wetlabs ECO BB3 Spectral backscattering 3 channels: 470, 532, 650 nm

TriOS RAMSES radiometers (new for 2014) Surface and sky radiances; solar irradianc Remote sensing reflectance (Rrs(0+)) Tilt and heading sensor

TriOS OSCAR hyperspectral PSICAM (new for 2014) Spectral absorption (a) 256 channels: 360-750 nm

Sampling Strategy & Instrumentation

UK Lakes

Loch Lomond

Loch Leven

Windemere

Bassenthwaite Lake

Derwent Water

Size: 5 – 71 km2

Range of depths Oligotrophic to eutrophic

Sampling in 2013: UK lakes

Frame for above-water radiometer package

Davit for deployment of in-water instruments

Cabin workstation for onboard filtration

Deployment

Conceptual framework for the development of the LIMNADES database

LIMNADES: DEVELOPMENTS

Linkages and Developments

• Brockmann Associates

• Lake Algal Bloom Pilot Project

• KTAMOP: Ecological Status and Function of Lakes (Hungarian Academy of Sciences) – HICO (Hypespectral Imager of the Coastal Ocean)

• Colour of Water: Climate induced changes in water colour in Nordic Lakes. Estonia/Sweden

• Validation team for ESA Sentinel 3

• GLaSS: Global Lakes Sentinel Services: Water Insight, Wageningen • INFORM: Improved monitoring and forecasting of ecological status of European Inland

waters by combining Future earth ObseRvation data and Models (FP7 project pending): VITO, Belgium

• DANCERS: DANube macroregion: Capacity building and Excellence in River Systems (basin, delta and sea): GeoEcoMar, Romania

• CYANOCOST: Training and networking for EO

• NETLAKE: COST Action for in-situ sensors

Future Directions

Funding: COST ACTION: EU Funding for training – as a framework for standardisation in Sentinel 3 (&2) validation for Inland Waters Opportunity: INTERCOMPARISON: of radiometers /calibration for reflectance and deployment/sampling methodologies. GLaSS/INFORM/GloboLakes Meetings: END USER WORKSHOP: JRC Italy. 1st Week in May Conferences: • JOINT AQUATIC SCIENCES MEETING 2014 (http://sgmeet.com/jasm2014/)

Portland Oregan (18-23 May 2014) (7th Feb Submission Deadline) – Working towards interoperability standards for networks and observatories – Lakes and streams as engines in global processes: Insights from technical

advances in limnology • 15th World Lake Conference (Perugia, Italy, Sep 14) - Lakes remote sensing

special session & GEO