A MULTI-DEVICE TOOL FOR HARMFUL ALGAE BLOOM MONITORING IN INLAND
WATERS: STUDY CASE LAKE ÉRIE
Anas El Alem1 (anas.el_alem@ete.inrs.ca), Karem Chokmani1, Charles Gignac1, Claudie Ratté-Fortin1,
Rabiaa Ben Aicha1, Sarah Goubet1, and Aarthi Venkatesan1
1Centre Eau Terre Environnement, INRS, 490 de la Couronne, Québec, QC G1K 9A9, Canada
Context
The occurrence and ampleness of harmful algae bloom (HAB) over Lake Érie, threatening human and animal health, is increasing from one year to another. Traditional network sampling programs, based on some punctual stations, are costly, labor intensive and with low temporal and spatial coverage, especially for water bodies of large area such as Lake Érie.
Objective
Development of the ÉCHO (Érie CHlorOphyll) application, which is compatible with android phones, tablets and computers, based on MODIS imagery downscaled to 250 m spatial resolution and on Landsat (TM, ETM+ and OLI) data.
Models calibration and accuracy
MODIS
Calibration: Using an Ensemble
based system (EBS)
(R2 adjustment = 0.98) Validation: Cross-validation 100 102 104 106 100 102 104 106 Observed (mg Chl-a m-3) E st im at ed ( m g C hl -a m -3 ) Data Line (1:1) EBS R2 = 0.98 NASHr = 0.96 RMSEr = 15% BIASr = -2%
Landsat
Calibration: Using a Combined
bands ratios (CBR) (R2 adjustment = 0.90) Validation: Cross-validation 10 0 10 1 10 2 Observed (mg Chl-a m -3 ) 10 0 10 1 10 2 Estimated (mg Chl-a m - 3 ) Data Line 1:1 R2 = 0.90 NASHr = 0.89 RMSEr = 11% BAISr = -1.5% CBR B) C) A) Lake Erie Pymatuning Reservoir
Methodology
The ÉCHO application allows to:
1 2
4 3
Retrieving the latest MODIS images of Lake Erie
Using the algorithm for image processing to retrieve Chl-a maps
Updating statistics and models based on new data Transferring updated information to the database
5 Issuing early warning notifications to clients
(Stakeholders, riverine, and interested people)
Assist stakeholders making decisions:
• Easily
• Affordably
• Timely
Put satellite technology directly into the hands of the society
Monitor water quality, in NRT, for the entire
Lake Define points of
interest (POI) and alert thresholds
Upload pictures, field data, and personal
remarks
Share information in social media and/or with other users
Forecast Chl-a
concentrations and issue alerts
Usage plans
ADVANCED ($$)
• All archived, lastest and
current Chl-a maps
• Unlimited user-defined POI • Access to all time series
data and statistics available
• Access to short-term Chl-a
forecasts
• Social media sharing
BASIC (FREE)
• Latest Chl-a maps • Social media sharing
REGULAR ($)
• Lastest and current Chl-a
maps
• Three user-defined POI • Access to current time
series data and statistics
• Social media sharing Dailly Chl-a
Monthly averages
Annual averages
MODIS images downloading
weekly Chl-a Monthly averages Annual averages CBR EBS Processing Chl-a concentrations 2000 To da te Spatial databases Clie nt s LS images downloading 1984 to da te
What next?
Development of more robust forecasting models to estimate Chl-a concentrations;
Development of a model to estimate Chl-a based on Sentinel-2 (A and B) data at 20m spatial resolution;
Development of models to estimates suspended matters based on the MODIS, Landsat and Sentinel data.
Increase knowledge for efficient actions 82°0'0"W 82°0'0"W 83°0'0"W 83°0'0"W 42°0'0"N 42°0'0"N ´ 0 10 20 30 40 km CBR Chl-a High : 2581.32 Low : 0 G - R Chl-a High : > 10000 Low : 0
