The results revealed that the use of GIS analysis and MCE may be employed for reforestation of a cut block. The project analyzed the enormous amounts of data affecting tree growing conditions, and determined, by using an environmental indirect approach, the optimal suitability locations for tree species. Further research and experimentation with different study sites will be required until knowledge and procedural transfer can be operational.
Upon evaluating the data collection strategy and analyzing the methodology, several operational and methodological problems were discovered and are acknowledged below:
Spatial Resolution:
Analysis should be done at the largest suitable scale for higher accuracy. The study site was chosen partly because the soil data had the smallest available scale of 1:100000 (i.e. ~50 m effective resolution) (Goodchild et al., 1993). This may be adequate for the cut block level but not precise for the tree stand level. Each pixel in a raster dataset represents an average of attributes over the real-world area that the cell covers. All layers used in this analysis were over-sampled to match the DEM’s 25-meter cell resolutions.
Temporal Resolution:
Most real-world natural features are variable. Soil structure, porosity, and organic-matter content of soil will change; especially, when disturbed by forest practices such as logging (Daniel et al., 1979). The course and depth of a river may also change. As will the topography over time.
Attribute Resolution:
Most real-world natural features are continuous. Representation of theses by entities in a database requires generalization, which reduces the attribute accuracy of these features. Soil is spatially continuous data but the dataset was discretized into areas containing predominantly one soi1 association. However, in many parts of the Taseko Lakes map different soi1 associations were too intimately mixed to be shown separately on the map. In other places it was impossible to map the exact boundaries between soi1 associations in the time allotted for the survey. Both these situations forced the grouping of soils from different associations within one area on the map.
Over simplification
More site factors can be incorporated in the decision-making. Bedrock geology determines many soi1 characteristics (Valentine et al, 1987). For example, where the bedrock source of soi1 parent material is coarse textured acidic granite with few bases the soils which form on that material will be coarse and relatively infertile. Where the source is basalt, which is base rich and finer grained, the soils will be finer and more fertile. A geology layer was obtained from BC Ministry of Energy, Mines and Petroleum Resources, Geological Survey Branch but it was determined not suitable because it was too large-scale (1:250 000) for the cut block level.
Unknowns:
Each factor in an MCE using weighted linear combination must be standardized to some common scale (i.e. 0-255). Explicit information about the relationships (e.g. deciding whether it is a normal or linear distribution) between each of these factors and ecological reality is not available or understood. Therefore, quantifying each relationship and determining the control points could only be approximated and hypothesized. This produces errors in MCEs produced using these factors since the estimated functions do not precisely describe real-world relationships.
This error is greatest for qualitative data because ranked values must be assigned to attribute values in these layers. These values are determined based on knowledge of the relationships of concern, but are analyst-defined. Difficulties were encountered in determining an ordinal scale for nominal entities; in other words, to quantify a qualitative measurement. For example, soils were described in language but a ranking had to be set based on the critical attributes of the tree species. As well, it was not recommended for spruce to be planted on steep slopes and pine thrives in a wide variety of topographic situations; but no definite slope degree was given. Perhaps further research could solve these problems.