Extensive Ecoforest Map of Northern Continuous Boreal Forest, Québec, Canada A. Robitaille¹, A. Leboeuf¹, J.-P. Létourneau¹, J.-P. Saucier¹ and É. Vaillancourt¹ 1. Ministère des Ressources naturelles et de la Faune du Québec, 880 chemin Sainte-Foy, 5 e étage, Québec, Canada, G1S 4X4; 1-418-627-8669; andre.robitaille@mrnf.gouv.qc.ca Introduction and context Since the late 1960s, the Ministère des Ressources naturelles et de la Faune du Québec (MRNFQ) has periodically carried out forest mapping activities at a scale of 1: 20,000 in Québec s forest areas south of 51º 30 north latitude, the territory under forest management. It integrates permanent environment and vegetation variables including potential vegetation. It is produced on a computer screen in 3D aerial photos and is updated annually with regard to forest work and disturbances such as forest fires (Robert and Robitaille 2009 and Robert and Robitaille 2009). In the early 2000s, new needs emerged for information about forest resources in a data-poor area located between 51º 30 and 53º north latitude and covering 240,000 km². To perform the work, the MRNFQ decided to explore new automated mapping approaches (Robitaille et al. 2008). Such approaches had to deliver high-quality products, be relatively inexpensive and be well adapted to remote or hard-to-reach areas. This paper presents the mapping method, software used and results obtained to date. The mapping program began in 2005 and will end in 2009. Aerial photos have been used to map physical environmental variables (surficial deposits, drainage) and satellite imagery to map vegetation variables. Integrating major disturbances with all of these variables completes the integrated ecoforest map, produced at a scale of 1:100,000. Mapped area The 240,000 km 2 area marks the transition between continuous spruce-moss forest to the south and southern extent of spruce-lichen open forest to the north (Figure 1). It covers a 1500 km east-west transect, which is characterized by multiple relief, altitude and climate zones, vegetation patterns, fire and insect disturbances, surficial deposits, and drainage. It is a sparsely populated, undeveloped area. There are only three majors roads across the territory. Figure 1. Integrated ecoforest map of the northern portion of the continuous boreal forest
In the western portion, it progresses from sea level to about 500 m. The terrain is gentle. In the west-central portion, the relief is gentle but altitude increases from an average of 500 to 700 m. There are also a few high hills with rugged topography up to 1000 m in altitude. The east-central portion is very rugged, with height differences that are frequently as great as 500 m. Several peaks reach 900 m. Finally, the relief in the eastern portion consists of rounded, relatively gentle hills with altitude from 500 m to sea level. The three following examples illustrate the effect of altitude, continental surface and water bodies on the territory s climate: Average annual temperatures range from 0 to 1 C on the west coast but are about -4 C in the centre of the territory. Average annual precipitation is greater in the centre, with 1000 mm and more on the higher peaks. It is less than 800 mm toward the west. Growing degree-days drop from 1100 degree-days in the west to nearly 500 in the centre. Classification structure As previously mentioned, the classification structure of the mapping is similar to the one developed farther south and integrates permanent environment and vegetation variables (Létourneau et al. 2008). Here are the major classes of this classification. Surficial deposits The composition of surficial deposits influences soil development, productivity of vegetation and drainage conditions and is a significant variable in land-use planning. The classes selected for the mapping refer to major genetic groups (glacial, fluvial, marine, etc.) and can be recognized by their morphology using aerial photos. Most of these classes are split into subcategories that are distinguished by compactness, granulometry and stoniness (Figure 2 and 3). Figure 2. Aeolian sand. Figure 3. Till from crystalline rocks. Drainage The 5 classes of drainage indicate levels of soil moisture: xeric, xeric-mesic, mesic, subhydric and hydric. The presence of seepage can be reported especially on the long slopes of hills. Vegetation The classes of vegetation are adapted to the MRNFQ s needs at this stage in the project, and the tools used allow them to be well recognized (Figure 4,5,6,7). Seven major characteristics can be used to describe the stands.
1. Cover type (deciduous, mixed or coniferous). 2. Understory vegetation (lichen, moss, shrubs). 3. Density classes (five classes from 10% to 100%). 4. Disturbances (fire, insects). 5. Development phase (mature, pre-mature or regeneration). 6. Vegetation without forest potential (wetland, barren, etc.). 7. No vegetation (water, rock, etc.). Figure 5. Coniferous (Picea pre- mariana) mature stand with moss understory Figure 4. Deciduous (Betula papyrifera) mature stand Figure 6. Fire from 2006 Figure 7. Coniferous (Picea mariana) mature stand with lichen understory Approach and tools The mapping approach includes three principals elements (Figure 8): (1) first, surficial deposits, (2) second, drainage boundaries and (3) third, vegetation. The interpretation of surficial deposits and drainage is initially carried out by analyzing aerial photos on a computer screen in 3D using DVP software (Groupe Alta inc. 2008) in conjunction with ESRI software. This makes it possible to simultaneously synchronize the topographic map and the photos so that the contours recorded on the screen by the geomorphologist automatically generate a GIS shapefile and an associated database. The polygons generated are then integrated into the Definiens software (Definiens inc. 2006). For each polygon, the software performs segmentation and classification of the vegetation, based on satellite imagery Landsat TM in this case. Image segmentation consists of automatically delineating polygons based on thematic maps and image homogeneity patterns. The user controls the size of polygons and
smoothness of contours. Cover type, density, barrens, spruce-lichen and spruce-moss forests are then distinguished by analysts, using spectral characteristics of images and ancillary data such as topographic maps, fire history maps, etc. Finally, contours and years of major disturbances, such as forest fires and insect epidemics, are integrated into the map. The integrated ecoforest mapping includes prior field checks by geomorphology and forestry specialists to guide and validate the process. 1 2 3 Figure 8. Northern forest inventory process Results and discussion The quality of the map was assessed in 2005 when the approach was developed for a training project. It was a qualitative assessment carried out using randomly selected GPS polygons. Results obtained for 240, 000 km 2 were very satisfactory and helped to identify the following advantages and points to improve. Advantages - Forest overview of a huge territory. Limits - Some attributes are not mapped, (e.g. height, species) or were difficult to discriminate. Conclusions and perspectives The approach developed by the MRNFQ provides an excellent picture of the vegetation and physical environment of these huge, poorly known areas. In fact, the substantial amount of
information about surficial deposits, drainage, vegetation and disturbances will be essential to improve our knowledge about this fragile ecosystem. This map will be linked to an ecological classification. So, for each polygon there will have information about the potential natural vegetation. This map associated with the analysis of ground sample plots acquired within the information acquisition framework, will allow in next years to elaborate the frontiers between spruce-moss forest and spruce-lichen open forest. Finally, in spite of the limitations observed, the approach achieves good precision rates and requires low investments, since satellite images cover large areas and archived aerial photos can be used to map surficial deposits and drainage. These conclusions pave the way to extrapolating this approach to other remote areas of the boreal field. Acknowledgments We acknowledge support of several collegues from MRNFQ for them supports, advices and insightful discussions: Geneviève Auclair, Jacques Brunelle, Christian Cantin, Lyne Carrier, Benoît De Serres, Marie-Pierre Drouin, Pierre Grondin, Sebastian Matejek and Sonia Watts. Literature Cited Definiens inc. 2006. Definiens professional version 5. GroupeAlta inc. 2008. DVP version 7. Létourneau, J.P., Matejek, S., Morneau, C., Robitaille, A., Roméo, T., Brunelle, J., and Leboeuf, A. 2008. Norme de cartographie écoforestière du Programme d inventaire écoforestier nordique. Ministère des Ressources naturelles et de la Faune du Québec. 51 pp. Robert, D., and Robitaille, A. 2009. Cartographie forestière. In Manuel de foresterie, Ordre des ingénieurs forestiers du Québec, Éditions Multi Mondes, pp.507-540. Robert, D. and Robitaille. 2009. Integrated Ecoforest Mapping of Southern Québec Forests. Extended forest inventory and monitoring over space and time, IUFRO the International Union of Forest Research Organizations, may 19-22 th. Québec City, Canada. Robitaille, A., Leboeuf, A., Létourneau, J.-P., Saucier, J.-P., and E., Vaillancourt, 2008. Integrated Ecoforest Mapping of the Northern Portion of the Continuous Boreal Forest, Québec, Canada. Workshop Proceedings in press: Circum Boreal Vegetation Mapping. 3-6 November, Helsinki, Finland