Long Point Environmental Folio, Chapter 12 J.G. Nelson and K.L. Wilcox, Editors, 1996

Analysis of Land Cover Change in the Long Point Area*

Patrick Lawrence, Karen Beazley and Chi Ling Yeung

*This chapter is largely derived from two sources: Yeung, C.L. 1993. "Analysis of Land Use/Land Cover Change of the Long Point Region Using Landsat MSS Images" Long Point Environmental Folio Series. Technical Note #1. Heritage Resources Centre, University of Waterloo, Waterloo, Ontario; and
Lawrence, P.L. and Beazley, K. 1994. "Analysis of Land Cover and Land Use Change in the Long Point Area from 1955 to 1990 Using Aerial Photography". Long Point Environmental Folio Series. Technical Note #2. Heritage Resources Centre, University of Waterloo, Waterloo, Ontario.

Introduction

In the Long Point Folio project, efforts were made to interpret land cover changes and their environmental and other effects in two ways. The first was to analyze images taken from satellites circling the earth at high altitudes. The second was by analyzing aerial photography taken from airplanes at lower altitudes. The results of these efforts are useful but not as complete and definitive as would be wished for planning and management purposes. The initial analysis of satellite images could only be completed for two images: 1974 and 1984. It would be very useful to have a comparable analysis for the 1990's since appreciable change seems to have occurred since 1984. The analysis of aerial photographs was completed for an area adjoining Long Point and is difficult to compare to the satellite image analysis of the larger Haldimand-Norfolk area. However, the analyses do indicate some important changes - for example in wetlands/marshes - and support the need for more detailed land use studies and monitoring of changes as a basis for planning, management and decision-making.
In reading this chapter, it is important to recognize that what was mapped from the satellite images and the aerial photographs was land cover change, not land use change. Land cover refers to the forests, marshes, ponds and open water, urban areas, cultivated fields and other patterns visible on the images and photographs. Land cover often tells us a lot about actual land use but is not strictly the same thing. Mapping forests, or wetlands, or urban areas, for example, does not give details on how they are used by humans.

Analysis of Land Cover Change from 1974-1984 using Landsat Images

Two Landsat Multispectral Scanner (MSS) images, dated June 21, 1974 and July 7, 1984, were compared to investigate regional land cover changes in the Long Point area. A 1991 Landsat Thematic Mapper (TM) image was also available for comparison but due to technical problems, analysis of this image was not completed for this project. Since the 1974 and 1984 images were taken during the same season and less than a month apart, they are assumed to be comparable in terms of the state of vegetation growth and land characteristics.
The satellite images cover a rectangular area consisting mainly of the Regional Municipality of Haldimand-Norfolk (Figure 1). The area is bounded in the east and west by Port Burwell and Dunnville and north and south by the Six Nations Indian Reserve and Lake Erie.
The MSS images have a spatial resolution of 79 m2 per pixel, which is the size of the smallest distinguishable feature. However, smaller features with sharply defined boundaries - for example roads through forested areas - are often visible on the images. The Landsat MSS records ground reflectance in four bands: band 4 (green; 0.5-0.6 mm), band 5 (red; 0.6-0.7 mm), band 6 (near-infrared; 0.7-0.8 mm) and band 7 (near-infrared; 0.8-1.1 mm). Each band offers different potential for distinguishing ground features.

Figure 12.1 Study Area: Long Point Biosphere Reserve and Region

Usually, band 6 and 7 detect water bodies the best while band 4 and 5 are good for cultural features. A false colour composite is produced for visible interpretation by assigning a red colour to band 7, green to band 5 and blue to band 4. This shows vegetation in different intensities of red, cultural features and bare surfaces in light colours, and water bodies in very dark colours. False colour composites for the two images were produced for visual comparison. Figure 2 shows forest cover of the Long Point area in 1974.

Figure 12.2 1974 Forest Cover (Deciduous and Coniferous Forests in green-blue).

Table 12.0 Six classifcations of land cover/land use

A simple classification scheme was developed with six land cover/land use classes. They are:

Agricultural Area Land under crops, pasture, grass and low shrub and herb, as well as area laid fallow (bare soil with some grasses). Abandoned fields and golf courses also were in this category.
Deciduous Woodland Land under deciduous trees, both planted and natural. Riparian forests are included as well as swamp.

Coniferous Woodland Land under coniferous trees, both planted or natural.

Marsh Inland and shore marshes with low lying partly submerged vegetation

Built-up Area Land under concrete or a mixture of concrete and vegetation.

Open Water Any water body.

Agricultural Area	Land under crops, pasture, grass and low shrub and herb, as well as area laid fallow (bare soil with some grasses). Abandoned fields and golf courses also were in this category.
Deciduous Woodland	Land under deciduous trees, both planted and natural. Riparian forests are included as well as swamp.
Coniferous Woodland	Land under coniferous trees, both planted or natural.
Marsh	Inland and shore marshes with low lying partly submerged vegetation
Built-up Area	Land under concrete or a mixture of concrete and vegetation.
Open Water	Any water body.

Beach, sand dune and bare ground often resemble Built-up Areas and can be mistakenly included in this class. Their high reflectance makes them very difficult to separate from Built-up Areas.

Interpretation of Images

Attempts were made to interpret the MSS images in terms of the foregoing land cover types, with the interpretations being checked by field reconnaissance in the study area. A major problem encountered in interpretation was confusion between Agricultural and Built-up Areas. Both of them involve a mixture of vegetated and non-vegetated surface. In the case of Agricultural Area, this is a mixing of fields in fallow, different densities of crops, and crops in different stages of maturity. In the case of Built-up Area, the mixing consists of buildings, construction sites, urban vegetation and lawn. Moreover, Agricultural Area usually has structures like barns, farm houses and tobacco kilns which show similar reflectance as urban buildings. Although the peak reflectance of these two land cover types are different, the variances in each case are so considerable that there is a large area of overlap where they are indistinguishable on the satellite images.
Another problem was the diversity and complexity of the Long Point spit or peninsula which possesses a variety of land cover types. Marsh and Open Water seem to be classified accurately; however, sand dunes and oak savanna are too small to be classified separately. Some interpretation also is necessary for other land cover types. The cottage community at the base of the peninsula is generally visible but owing to the large amount of lawn and trees among the cottages, much of this area looked like land classified as Agricultural. Similarly, the sparsely vegetated oak savanna and low shrub communities appeared to be similar to Agricultural. Non-vegetated sand dunes and beaches resembled Built-up Area.

Results

In general, no dramatic change in land cover pattern was discerned in the 10 years from June 21, 1974 to July 7, 1984. The area was classified and mapped as dominantly agricultural with abundant woodlots among the fields, riparian forest along streams, and plantations near the St. William Forestry Station and Turkey Point Provincial Park. Marshes were mapped at the Big Creek Estuary, Long Point and Turkey Point Peninsulas, the Grand River Estuary and scattered about inland lakes. Established urban centers such as Port Dover and Simcoe were clearly detected, as well as the Nanticoke Industrial Complex. However, the coarse spatial resolution of the MSS and the simple classification scheme did not reveal the complexity of the land cover on the Long Point peninsula.
Some subtle changes appear to have occurred during the 10 year study period (Table 1).

Table 12.1 Estimated Percentage Change in Land Cover for the Haldimand-Norfolk Region 1974-1984 (from Yeung, 1993)

Land Cover/Land Use Type 1974 1984

Agricultural Area 84.5 84.3

Deciduous Woodland 10.8 12.1

Marsh 2.2 1.9

Built-up Area 1.8 1.1

Coniferous Woodland 0.7 0.5

Unclassified 0.0 0.2

Land Cover/Land Use Type	1974	1984
Agricultural Area	84.5	84.3
Deciduous Woodland	10.8	12.1
Marsh	2.2	1.9
Built-up Area	1.8	1.1
Coniferous Woodland	0.7	0.5
Unclassified	0.0	0.2

At a regional scale, the percentage of Agricultural Area remains more or less the same. However, it increases significantly closer to the Long Point area rising from 47% to 57%, even when the confusion with Built-up Area is taken into account (Table 2).

Table 12.2 Estimated Percentage Change in Land Cover for the Long Point Area 1974-1984 (from Yeung, 1993).

Land Cover Type 1974 1984

Agricultural Area 47.2 56.7

Marsh 34.3 29.9

Deciduous Woodland 7.1 0.3

Coniferous Woodland 4.8 3.3

Built-up Area 6.6 0.9

Land Cover Type	1974	1984
Agricultural Area	47.2	56.7
Marsh	34.3	29.9
Deciduous Woodland	7.1	0.3
Coniferous Woodland	4.8	3.3
Built-up Area	6.6	0.9

This increase seems to be at the expense of the marshes in the north shore area of the Inner Bay. The marshes register a substantial decrease from more than 34% to less than 30% locally and from 2.2% to 1.9% in the region. There appears to be some increase in deciduous woodland, from 10.8% to 12.1% regionally and 7.1% to 9.3% locally. No further fragmentation of the woodland is evident. Reforestation programs and the gradual maturity of the forests seem to have increased the overall forest cover of the region. However, changes in species or in the nature or composition of forest cover could not be detected in this analysis. For instance, the classification was not sensitive to the replacement of indigenous forest by planted woodland. Similarly, Agriculture Areas which have been inactive or abandoned may still be classified as Agricultural since a grassy surface is still dominant in such areas. The analysis indicated a decrease in coniferous woodland from 0.7% to 0.5% regionally and 4.8% to 3.3% locally. This may be the result of a shift to deciduous species used in reforestation programs as most of the coniferous woodland is in forestry plantations.
As a measure of the level of urbanization of the region, Built-up Areas are very difficult to interpret because of confusion with Agricultural Areas. The estimated decrease in Built-up Area from 1.8% to 1.1% regionally during the 10 year period is unlikely in view of the gradual expansion of towns like Simcoe and Port Dover, and the growth of cottage development during the same period. A likely explanation is that newly developed residential areas of the 1970's had become better vegetated in the 1980's, resulting in more of those areas being misclassified as Agricultural land. The biggest change in Built-up Area occurs at the Nanticoke Industrial Complex with the addition in the 1980's, of the Texaco Refinery and the Stelco Lake Erie Plant to the Thermal Generating Station already visible in the 1974 image.

Analysis of Land Cover Change from 1955 to 1990 Using Aerial Photography

Land cover analysis was also completed on the basis of 1955 and 1990 aerial photography for a smaller area closer to Long Point. These two sets represented the most complete and accessible photography for the study area. In addition, aerial photography from 1978 was also used to make note of major changes in the period from 1978 to 1990. Field observations completed in 1992 also provided more information on the most recent changes.
Seven main classes of land cover were defined for the Long Point study area: Forests; Wetland/marshes; Parkland/Savanna; Agriculture; Built-up Area; Coastal Deposits; and, Open Water. Forest and Parkland/savanna are identified by their vegetation cover patterns. Forest includes both deciduous and coniferous trees. Parkland/Savanna is a term used to refer to vegetation consisting of a mixture of grassland and scattered trees, often oak. Wetland/marshes are areas that could be identified as having dense aquatic emergent vegetation. They tend to be located mainly along the shoreline. Agriculture areas are sites dominated by open fields or cultivated fields with planted crops. Built-Up areas include all types of human development, such as urban communities, rural residential buildings (cottages), services (roadways), commercial uses, industry, marinas, dredged sites, and recreational sites (golf courses).
Coastal deposits include shoreline bluffs, beaches and dunes and are found mainly on the Long Point peninsula. The open water class includes only enclosed ponds or inlets on the mainland and peninsula and not the Inner Bay or Lake Erie.

Results

In 1955 the dominant land cover type in the study area (Figure 3) was Agriculture, consisting of 53 % of the total land cover area and including cultivated fields with grains, market gardens, and tobacco as the significant crop types. Wetland/marshes occupied 22% and Forest 17% - with deciduous trees being the most common. The other main land cover type was Built-up area with 6% coverage; it included cottages, marinas, dredged areas, and urban developments. Open Water (2%) and all other cover types, including Coastal Deposits and Parkland/Savanna (less then 0.5%) made up the rest of the total.

Figure 12.3 The Long Point Study Area 1955 (from Lawrence and Beazley, 1994)

By 1990 (Figure 4), Agricultural land area had increased to 56%. Wetland/marshes in the Inner Bay area showed a decrease to 15% of the total land cover, a loss of 1/3 in area since 1955. A slight increase occurred in Forest with little change in the other classes. An increase in deciduous trees may represent tree planting initiatives.

Figure 12.4 Long Point Area 1990 Land Cover (from Lawrence and Beazley, 1994)

Built-up area appears to have remained unchanged as a percentage of the total land cover from 1955 to 1990 with small declines in some places and limited and site specific growth in some marina, residential, urban and cottage areas. It is also possible that forest canopy cover is masking development resulting in classification of some cottage and residential areas as forest.
The examination of aerial photographs revealed several sites where land cover changes occurred from 1955 to 1990. Many of the changes are highlighted on Figure 5.

Figure 12.5 Significant Land Use Change 1955-1990 (Adapted from Lawrence and Beazley, 1994)

Infilling of vacant lots, roadway construction, and expansion of cottage developments have occurred affecting Wetland/Marsh areas. Construction of an impoundment and development of an enclosed marsh west of the causeway at Big Creek have led to more open water and reduction in emergent vegetation. Cottages on Hastings Drive have been removed due to flooding and erosion damage. Urban growth along main roadways at Port Rowan and St. William's has increased the size of these communities. Dredging, infilling, and channelization have occurred at marina sites along the Inner Bay marshes. Wetland infilling and channel dredging have resulted in increased development at Turkey Point. Associated landuse activities from 1978 are highlighted in Figure 6.

Figure 12.6 Long Point Region: Land Use Activities between 1978 and 1992 (Adapted from Lawrence and Beazley, 1994)

The accuracy of the results of a study such as this can be affected by errors arising from several factors including the scale of photography and mapping, interpretation of the land cover types, and the use of hand measurements of the land cover types on the aerial photographs. It should be noted that the accuracy of interpretation of land cover types from aerial photography is limited by the ability to detect and identify features that are characteristic of each land cover type. Further detailed field studies and mapping would be required to develop more specific land cover types and improve the analysis. The development of more detailed analysis techniques, perhaps by digitizing the land cover maps and utilizing computer technology (Geographic Information Systems), might reduce the margin of error in calculations. Obtaining more precise and complete aerial photography coverage would also allow for improved interpretation.
It is difficult to determine direct cause-effect relationships for many of the land cover changes. Many are a result of a combination of several changes in land uses and/or natural processes. For example, wetland/marsh loss can be attributed to water level changes or variations in emergent vegetation or the removal of vegetation due to channelization and infilling. Further research on understanding the nature and rate of land cover changes is necessary especially in those parts of Long Point where rapid changes seem to be occurring which could adversely affect wetland, marsh, water quality and other resources of importance to sustainable development.

Work Cited

Lawrence, P.L. and Beazley, K. 1994. Analysis of Land Cover and Land Use Change in the Long Point Area from 1955 to 1990 Using Aerial Photography Long Point Environmental Folio Series. (Nelson, J.G. and Lawrence, P.L. eds). Technical Note 2. Heritage Resources Centre, University of Waterloo, Waterloo, Ontario.
McLellan, A.G. (ed). 1971. Waterloo County Selected Geographical Essays Department of Geography, University of Waterloo, Waterloo, Ontario.
Yeung, C.L. 1993. Analysis of Land Use/Land Cover Change of the Long Point Region from 1974 to 1984 Using Landsat MSS Images Long Point Environmental Folio Series. (Nelson, J.G. and Lawrence, P.L. eds). Technical Note 1. Heritage Resources Centre, University of Waterloo, Waterloo, Ontario.