Remote sensing of river temperatures in eastern Canada

Collaborators

Ms Anik Daigle, Cégep Garneau and project principal investigator

Gespe’gewa’gi Institute of Natural Understanding (GINU)

Direction principale de l’expertise sur la faune aquatique (MELCCFP)

Funding

Natural Sciences and Engineering Research Council of Canada (NSERC) Applied Research and Development (ARD) Program

Objectives

Water temperature in rivers and lakes influences water quality and aquatic habitats.
In particular, it affects dissolved oxygen levels, pollutant toxicity, the proliferation of invasive species, and the metabolism of aquatic organisms.

However, monitoring networks in Canada remain limited in both time and space. As a result, our understanding of thermal regimes remains fragmented. This complicates the assessment of water quality, aquatic habitats, ice-free and ice-covered seasons, and their impacts on spring flooding, species phenology, and fisheries. Without these data, it is difficult to anticipate the effects of global warming.

Our solution: to enhance the thermal profile of the river network in eastern Canada using satellite remote sensing.

Methodology

1. Automated Identification of “Water” Pixels

We automatically identify “water” pixels in Landsat images. We then use this data to:

• Characterize the thermal regime at each pixel.
• Evaluate its temporal evolution.

Through a user-friendly interface, users can view and download temperature time series and descriptors tailored to their needs.

2. Identification of “sentinel” pixels

We cross-reference multiple data sources to assign a probability of “water” membership to pixels.
This allows us to target only pixels that are 100% within a watercourse to avoid false measurements.

3. Longitudinal interpolation

Sentinel pixels provide temperature measurements at multiple points along lakes and rivers.
Since these measurements are discrete, we perform interpolation along the longitudinal axis using:

• Conventional methods (linear or cubic).
• Regressions based on local characteristics (elevation, stream order, width, slope, etc.).

4. Field validation

We validate the data using historical measurements from RivTemp and DataStream Atlantic. This allows us to assess the accuracy and any potential biases of the sentinel pixels.

5. Spatio-temporal characterization of thermal regimes

We extract and measure information from thermal profiles:

• Probability that the pixel belongs to the “water” category.
• Available temperature series.
• Interannual average profile of the thermal regime.
• Simplified three-parameter model (annual maximum, date of occurrence, duration of the warm season).
• Temporal trends (in °C/year) of annual and monthly average temperatures.
• 60 descriptive statistics to assess thermal habitat quality, including the thermal growth index for juvenile Atlantic salmon.

Setting up the distribution interface

We use Google Earth Engine (GEE) to deploy an online interface.
With Apps Engine, users can:

• Explore data in real time.
• Dynamically analyze geospatial information.
• Access interactive maps, charts, and dashboards to visualize results intuitively.

Equipments

• Google Earth Engine
• Python and Javascript programming
• Thermographs

Domains

Resource management

Environmental monitoring