Canadian Forest Service links

Estimation of Deforestation and CO₂ Emissions from Hydro Reservoirs in Quebec

Authors: David Tammadge, Don Leckie, Frank Eichel and Dennis Paradine

Partners: Environment Canada, Canadian Space Agency

Description: As part of Canada’s commitments to the United Nations Framework Convention on Climate Change Canada must report on the area of deforestation and related greenhouse gas emissions. A team within the Digital Remote Sensing group at Pacific Forestry Centre, Canadian Forest Service led the effort to make 1970-2004 deforestation estimates for Canada. Hydro reservoirs are individual deforestation events that, for the given year in which they occur, can dwarf all other causes of deforestation. It is therefore important to get a good estimate of the actual area flooded.

In the 1980s and 1990s Quebec in particular developed large hydroelectric reservoirs. Records from hydroelectric companies and regulatory agencies sometimes give the area of forest cleared before flooding but not the total area of forest flooded. Land cover classifications prior to flooding were not available and are time consuming to produce; especially as using Landsat MSS data would be necessary with the older reservoirs. A quick and inexpensive method to estimate the area of forest flooded was needed. An efficient approach was to use circa year 2000 EOSD land cover classifications to calculate the percentage of treed area surrounding the hydro reservoir and then to extrapolate this data into the area of the reservoir, obtaining the approximate area that was treed before the reservoir was flooded.

The procedure used involved obtaining the EOSD data from the scenes including and immediately surrounding each hydro reservoir. The area of each EOSD class was calculated by summing the area of pixels of each class. Then the total area less the no data, cloud and shadow classes (i.e., total mappable area) was calculated. From this the treed area (EOSD classes: coniferous dense, coniferous open, coniferous sparse, broadleaf dense, broadleaf open, broadleaf sparse, mixed wood dense, mixed wood open, mixed wood sparse) was divided by the total interpretable area to give the percentage of forested area for the scene. This percentage was then multiplied by the area of the reservoir to give the approximate area treed before the flooding of the hydro reservoir and therefore the area deforested by the flooding of the reservoir. By using EOSD classified data the area of deforestation was estimated and subsequently carbon dioxide emissions from hydro reservoirs were calculated.

Figure 1: This is an EOSD classified image of the La Forge 1 hydro reservoir. The reservoir is highlighted in brown.