LANDSCAPE PATTERN AND CHANGE THROUGH INTEGRATION OF REMOTE SENSING AND STONE WALL FEATURE IDENTIFICATION

Stone walls are relics of an agricultural civilization that once flourished in New England. By identifying the locations of both historical and present day stone walls, compositions of post-agricultural landscapes common across the New England region can be assessed with inclusion of historic human-land use interactions. I selected the town of New Shoreham, known as Block Island, as the study site for this thesis. Block Island is located approximately 14.5 km south of the Rhode Island mainland. The Island has rich land use history which resulted in an extensive network of stone walls still present across the landscape. Through visual image interpretation of 0.5 ft (0.1524 m) resolution orthophotography collected in the spring of 2011 and a historical topographic map from 1900, I created two datasets of stone walls containing total lengths of 260.6 km and 349.1 km, respectively. Analysis of these two datasets allowed for a temporal analysis to then creation three additional datasets containing stone walls between 1900 and 2011 which were matching, removed and built. The presence of stone walls on Block Island was quantified in connection to ancillary Geographic Information System (GIS) data, representing both natural and anthropogenic classifications of the landscape. The natural landscape is represented by land use and land cover (LULC) available for 1988, 1995, 2003/04 and 2011. Data of LULC were further quantified for land cover change frequency (LCCF); the number of land cover changes occurring within each 45 m pixel between 1988 and 2011. The anthropogenic landscape is distinguished by the parcel boundaries for New Shoreham as of 2013 and protected open space as of 2013. The 2011 dataset of stone walls was quantified for stone wall distribution among each land cover class for the temporal range, finding a higher abundance of stone walls within agricultural lands for 1988 and 1995 and urban lands from 1995 through 2011. The 2011 stone wall dataset was also quantified for distribution among each land LCCF class to find a higher proportion of stone walls contained within lands with the greatest frequency of land cover change. A strong relationship exists between the coincidence of stone walls and the boundaries of land parcels. Approximately 81% of parcels are in part bordered by a stone wall from the 2011 dataset. Additionally, over 50% of the lengths stone walls within the 5 datasets of stone walls are bordering parcel boundaries, with the more current datasets of 2011, matching and built having over 80% of their lengths adjacent to the boundaries of 2013 parcels. Lastly, at least 37% of the stone walls current as of 2011 are expected to remain untouched due to being contained within land designated as protected open space. Stone walls represent a human component, among the many broad factors which generate the composition of landscape mosaics. By utilizing abilities of GIS technologies to identify stone walls for a large geographic area, this research models initial exploration of the relationship between this historical feature and the landscape it continues to reside within. Additionally, this work adds justification to continue the integration of remote sensing technologies and human’s cultural histories in studying driving factors of land cover change and anthropogenic landscape characterization.


LIST OF TABLES
There are many unknowns pertaining to the land use history of New England prior to European settlement when the Native Americans occupied the lands, as well as what conditions truly define the natural environment prior to human alteration of the landscape Hammond, 2002). However, there is a clear distinction in the time period between pre-and post-intensive human-landscape interaction; when lands within New England were settled by Europeans and management of lands became well established on an annual basis. The documented record of human history dates back to include this relatively recent period of intensive human land use, allowing for studies which involve assessment of temporal trends in land development and alteration.
Geographic information systems (GIS) have emerged as useful tools in addressing landscape-level research questions (Turner et al., 1996). A GIS allows for integration of remotely sensed data in conjunction with ancillary data. Specifically, it is with use of a GIS that location-based data can be visualized and analyzed.
Collections of remotely sensed ground imagery have been acquired throughout the past century and continue to advance in both data volume and data quality.
Additionally, public interests in the field of historical landscape ecology continues to grow as seen through the pursuits of environmental organizations to use knowledge of past land use in conservation initiatives; in several cases relying on data available as a result of the synthesis between GIS and human history (Hammond 2002). The abilities of spatial data analysis will greatly enhance the underlying purpose of this research, by allowing for an integrative method between knowledge of historical human land use and the temporal characterization of land. More specifically, a GIS will assist in identification of stone walls and exploration of the connection between stone walls and the natural and anthropogenic landscape.
The scope of this study is to assess the ability of human land use interactions to persist. Stone walls are indicators of human land use and through their identification I in turn identify locations influenced by the era of human settlement and agriculture.
Additionally, by completing a temporal assessment of stone walls I am able to assess stone walls approximately two centuries after their initial mass creation. First, both current and historical distributions of stone walls were determined. These data were then be compared with a temporal compilation of the natural landscape as based on This integration will further study these historical features by assessing the spatial relationships between the temporal distribution of stone walls and the present and more recent landscape in which they reside.
Stone walls have previously been considered as a factor in studies pertaining to historical land use and change (Cronon, 2011) as well as landscape characterization (Wessels, 1997). However, few landscape studies solely focus on stone walls as features in their own right. Specifically, the temporal distribution of stone walls and the spatial connection between these features and the present day landscape. Through making use of GIS technologies, stone walls are able to be temporally identified with historical data and aerial imagery. This identification allowed for this research to focus on both temporal change to the distribution of stone walls and an initial assessment of the relationship between stone walls and the more recently characterized landscape.
The temporal pattern of stone walls can be determined through use of historical maps and remote sensing data.

2.
Present and more recent characterizations of the landscape can be assessed with inclusion of the temporal placement of stone walls.

History and Function of Stone Walls
In present day New England, stone walls are the most noticeable relics existing as evidence of the historical agricultural civilization that once flourished between the 18 th and 19 th centuries. Stone walls exist in New England as products of the integrated histories of nature and humans. There were several factors which lead to the formation of stone walls. Additionally, the function of these walls has changed over time.
Stone walls are composed of till stones. These stones are a product of New England's geologic setting. The Laurentide Ice Sheet retreated from Rhode Island about 20,000 years ago (Boothroyd, 2002). This glacier completely reworked the New England landscape, burying an abundance of ablation till under the surface at varying depths. Additionally, the New England climate has seasonal temperature variations which result in yearly cycles of ground freeze and thaw. The combination of these geologic and climatic factors allowed for the process of frost heaving to occur which results in the surfacing of buried till stones through swelling and settling of the surrounding soil.
Due to large scale agricultural practices under way in the late 1700s and early 1800s, these stones began to emerge on the soil surface at a high rate. This emerging was predominately a result of the forest clearing taking place which was the first major anthropogenic interaction to result in soil destabilization (Thorson, 2009).
However, use of plows enhanced the mechanisms causing stone surfacing by increasing the water holding capacity of the soil allowing for a greater magnitude of frost and thaw to take place.
Every spring these till stones emerged throughout fields and pastures in substantial quantities. To rid the fields of stones they were stacked around the boundaries of fields and properties and eventually formed into walls and fences. This was a process which was completed with use of tools to break and shape the stone, oxen to haul piles of rock and human labor to pick the rocks and form the walls. The overarching reason for the creation of stone walls was to serve as "linear landfills" (Thorson, 2009). At first, most stone walls were formed in conjunction with existing wood fencing. However, as settlements expanded and both resources and social mentalities changed, so did the function of stone walls.
After initial settlement people lived in communes where lands and property was shared, but after time people saw the value in personal property ownership and had the desire to clearly define their property boundary from that of their neighbors.
Additionally, there was a need to create fences for the purpose of keeping animals in the confines of their owner's fields and out of the fields of nearby farmers (Allport, 1990). Good fences make good neighbors became a common sentiment among New Englanders (Frost, 1914). Coinciding with these changes was also a reduction in the abundance of available forest resources, which at first were seemingly endless. Fences first built by the settlers were made mostly of wood. However, wood fences would easily rot and need to be replaced. By the mid-1800s forest abundance in New England was at a minimum of about 20% (Bellemare, 2002). Additionally, New England was dominated by fields, pastures and woodlots . Ultimately it was the combination of factors: the seemingly endless amount of available stones, the increased need to fence one's property from their neighbor, the breakdown of wood fences and the reduction of available forest materials that led to the increased reliance of stones to be shaped into stone fences.
As the function of stone walls increased so did the value placed on them. The blueprint for building a stone wall would vary based on its purpose. Some stone walls were formed with additional precision such as wall ends separating a path or property as well as those more likely to be seen by those visiting from out of town. Some walls served as property boundaries while others were built to hold sheep and cattle (Allport, 1990).
In a day an individual could build about 5 m of wall while a team could form up to 60 m (Thorson 2009). It has been estimated that in 1871 there were 406,422 km (252,539 miles) of stone walls existing within the northeast (Allport, 1994 andThorson, 2009). At this time farming was beginning to decline in New England and farmers were moving both to the mid-west where soils and equipment were better for intensive farming and into cities where the industrial revolution created opportunities (Jeon, 2014).
In the years since agriculture decline, old pastures and fields have been overtaken by second growth forest and expanding urbanization. However, the stone walls remained and the same walls remain to this day; except for those which were altered from natural mechanisms or due to human interference. These stone walls still identify distinctions in property boundaries and land use; the latter likely a result of the former. Additionally, stone walls have continued to be built but their function is generally for aesthetic purposes to highlight property (Allport, 1990). Stone walls also serve the function of creating their own environments at a local scale where they provide habitat and refuge for small mammals, repositories for nuts and seeds and a microclimate conditions for young low lying vegetation to settle (Thorson, 2009 andCollier, 2013). Stone walls are a part of a social value that exists for many native New Englanders. A value transcended from our ancestors that serve as a reminder of a time filled with challenges, hardship and most importantly opportunity and perseverance.

The Landscape and Anthropogenic Land Use
A landscape is a product of multiple factors pertaining to natural abiotic and biotic conditions as well as human interactions, specifically anthropogenic land use (Turner, 2001). The combination of these factors results in the landscape as a mosaic of patches. These patches can then be studied as based on their structure, function and change which is what landscape ecologists focus on for the purpose of assessing how the configuration of landscapes results in ecological processes over time (Turner, 2001).
Humans have long been an integrated part of the environment and human manipulation of the landscape has lasting effects Turner, 2001). The task of deconstructing a temporally rich and complex landscape and identifying change through time is fundamental to the understanding of past human activity (Lock et al., 2002). Additionally, studies in ecosystems must consider the legacy effects of historical human land use Motzkin, 1996).
Previous studies of historical anthropogenic land alteration as based on examination of records, documented recollections, and in situ research has led to a much fuller comprehension of the present day landscape configuration and how the land cover mosaic has transitioned over time as based on both natural and human disturbances . Investigations of land use history have increased knowledge on the development of vegetative land cover, response of vegetative communities to both novel and natural disturbances, and new perspectives to be used in landscape management (Foster, 1992).
Specifically, in New England, various studies have assessed consequences of historical agricultural land use through characterizing temporal structure and function of these landscapes. Studies focused on soil structure and chemical composition, vegetative composition, and resistance to disturbance, determined that past land use does influence compositions of subsequent landscapes (Foster, 2003 andFlinn, 2005).
While initial site conditions can be a defining factor in determination of land cover, land ownership can also play a crucial role by altering the spatial extent of land use (Foster, 1992). Social and economic considerations are among the most important drivers of landscape change, yet few studies have addressed both economic and environmental influences on landscape structure, and how land ownership may affect landscape dynamics (Turner, 1996).
When stone walls were originally placed on the landscape they formed borderers around fields and properties. This notation is still very evident within the study site. Through a simple overlay of the parcel boundaries in New Shoreham and present day aerial imagery, stone walls clearly coincide with these boundaries. This initial relationship makes clear that the interaction between nature and humans, which characterized the historical anthropogenic landscape, resulted in landscape alteration which has persisted in some capacity to the present day and promotes further investigation.

Remote Sensing
The acquisition of land cover imagery using remote sensing began with aerial images captured by planes pre-1900's and made huge advances in the 1970's with use of satellites to capture multi-band imagery of the globe. Commonly acquired are ground reflectance values in the red, green, blue and thermal bands of the electromagnetic spectrum but there are several other possibilities. Through postprocessing and rectification, data are delivered to the user as pixelated images in which each pixel's location is associated with its x-y location on the ground and the size of each pixel correlates with the resolution of the receiver which acquired the data. Collection of both aerial and satellite imagery has continued to advance in detail by means of increasing number of spectral bands, resolution and positional accuracy.
Remote sensing data acquisition has ultimately resulted in a collection spatial data containing several decades of land cover and equally essential recent high resolution data sets. The consortium of federal agencies, which produce high resolution imagery of the Earth's surface, do so in part to assist in studies which focus on land cover change (US EPA, 2014). Available data used by those in the field of remote sensing for landscape analysis, has greatly enhanced scientific understanding of environmental change.
The use of satellite-based remote sensing data has been determined to be a cost-effective approach to document changes over large geographic regions (Lunetta, 2004). This can be more recognized through review of temporal land cover change studies (Yang et al., 2014). In this study, classification and determination of landscape change occurred through use of the pre-classified land cover imagery derived from aerial photographs. "Aerial photographs provide the largest source of information available today for research of long-term vegetation dynamics, and are the only source of information on vegetation dynamics that combines high spatial resolution, large spatial extent, and long-term coverage (Kadmon et al., 1999)." In today's world of remote sensing, aerial photography is just one of the many sources of data which researchers can use for landscape studies. The integration of data captured through Light Detection and Ranging (LiDAR) as well as unmanned aerial vehicles (UAVs) into landscape analysis could have major implications on research findings; advancing the scope of studies both in depth and spatial extent.

Study Site: New Shoreham, Rhode Island
I selected the town of New Shoreham, also known as Block Island, as the study site. Currently, the Island is located 14.5 km south of the Rhode Island mainland.
Geologically, New Shoreham is a located just north of the Late Wisconsinan terminal moraine that retreated approximately 18,000 years ago (Boothroyd, 2000). While the Island had been inhabited by the Manisseans, a Niantic tribe of the Native Americans, for at least two centuries prior to European settlement, the first documentation of the Island was written by Giovanni da Verrazzano in 1524.The Island was officially settled in 1661 by a group of 16 men from the Massachusetts Bay Colony (Rosenzweig et al., 2000). Block Island is an ideal study site for assessing the connections between the historical anthropogenic land use and characterizations of the temporal landscape.
The general patterns of land use history on the Island was very reflective of mainland New England. This includes inhabitation by settlers from Europe, massive forest clearing and intensive agriculture and husbandry (Livermore, 1886). Combined factors of geologic history and human land use history resulted in the creation of stone walls on Block Island, just as in other areas which also contain stone walls throughout New England.
While New Shoreham does possess a similar characterization and history as the mainland, there are variations related to New Shoreham's island geography.
Initially, Block Island was covered in dense forest. However, these resources quickly dwindled for their use for fencing, building materials and fuel (Livermore, 1886). It is evident that these original forests never recovered so between initial loss of timber and 1750 resident were unsure of their future on the Island (Livermore, 1886). However, in 1750 peat became a valuable fuel source with coal becoming viable in 1846 (Livermore, 1886). Other main resources valuable to the productivity of the Island include sea week for fertilizer and the fisheries (Livermore, 1886 representative of the landscape of the southern New England region. In 1886 it was estimated that over 482 km (300 miles) of stone wall were contained on Block Island (Livermore, 1886).

Land Cover Data
The RIGIS database contains land cover maps for the years 1988, 1995, 2003/04 and 2011 derived from image interpretation and classification processes.
LULC data was classified with a minimum mapping unit (MMU) of 0.5 acres. The MMU refers to the smallest size area entity to be mapped as a discrete area (Saura, 2002). These land use and land cover maps were used in this study to complete a temporal analysis of land cover change. Data are characterized to the U.S. Geological Survey's classification system (Anderson et al., 1976). All four datasets are available as an Anderson Level III classification.

Additional Datasets
Determination of the distribution of historical stone walls was based on the information presented within a historical topographic map from 1900 provided by the All data were either downloaded in or re-projected to the NAD 1983 Rhode Island State Plane Foot Coordinate System (Table 1). The RIGIS database is an online service used to obtain data for this project. The RIGIS database is freely accessible to the public and allowed for analysis at the appropriate scales of this study.

Identification of Stone Walls
Methods to the practice of feature identification vary based on the quality of data utilized and purpose of identification. Common practice to identification involves digitization of features by manual delineation through user visualization and pattern recognition. Visualization of the data is a powerful way to utilize perception of the human eye for detection of features on the landscape, especially at the size of narrow linear, man-made features. Other method of automatic extraction were explored but since accuracy was a high priority for this study, delineation was adopted to identify and extract stone wall information contained within the dataset of 2011. Due to the high spatial accuracies of the 2011 orthophotography stone walls are able to be clearly visible in open fields, urban areas and under canopy ( Figure 5).

Stone Walls and Land Cover Change
To assess temporal distributions of stone walls on the landscape, determination Temporal frequency refers to the rate at which change events occur; ecosystem and/or anthropogenic (Lunetta, 2004). Land cover classification datasets for 1988, with an accuracy of <1meter differential correction. The approximate vertical and horizontal distance that stone walls were set away from the receiver was set for additional location accuracy. Most of the stone walls surveyed were located along roads where I was able to easily find and record their locations. Additionally, I was able to access several stone walls located in Lewis Farm which is contained in protected open space in the southwest corner of the Island. The software Terrasync Pro 5.6 was used to convert the features collected into a shapefile which could be exported and read by ESRI's ArcMap. An accuracy assessment was done to determine the accuracy of the 2011 stone wall dataset based on the stone walls identified in the field which were not within the dataset. However, since I was only able to survey a sampling of the stone walls, I was not able to determine which stone walls in the dataset are not located on the ground.

Temporal Distribution of Stone Walls
Amounts

Stone Walls and Temporal Land Cover
Land cover classifications at an AL1 (Table 3). In year to year trends are broken out into a bar graph and line graph to show that throughout the temporal range urban, forest and water increased while agriculture, bushland, wetland  (Table 4).
The land cover change on Block Island during the temporal range was assessed to determine total LCCF for each 45 m pixel. The assessment of LCCF determined that approximately 54% of the land had no change in AL1 coding, 44% of the land had changed once, 2% of the land changed twice and almost 0% of the land changed between each date assessed ( Table 5, Figure 19 and 20). Similar to the analysis with stone walls and LULC, the distribution of stone walls per LCCF class match up very well to the percent of land per LCCF class. However, a z test for comparison of percentages found significant the difference between the amount of stone walls and the amount of land within the LCCF class of 3, indicating a greater abundance of stone walls within this class (Table 5).
To assess LCCF at a local scale around the stone walls a comparison was done between buffered areas containing stone walls and absent of stone walls. There was no difference found between the magnitude of land change around the stone walls as based on a 15 m buffer and the magnitude of land change around 15 m buffers absent of stone walls (t-test, p= 0.520, Table 6, Figure 22). This buffer was somewhat arbitrarily chosen as based on the total land area of the Island and size buffers that would allow for a good representation of areas absent of stone walls to be compared.

Stone Walls and the Anthropogenic Landscape
The  (Table 9). It was not determined which stone walls were inaccurately included within the 2011 stone wall dataset. However, there were no stone walls in the dataset not also located on the ground in the specific areas surveyed.
The majority of the stone walls identified in both the field and within the 2011 stone wall dataset are contained within the AL1 classes of urban, agriculture and forest (Table 10). The majority of the stone walls identified in the field but missed within the 2011 stone wall dataset are also contained within the AL1 classes of urban, agriculture and forest (Table 10). Land use classes are based on the LULC data normalized to an Anderson Level 1 from 2011 ( Figure 14). Additionally, 12.12 km (45.09%) of the stone walls surveyed in the field are within 7.5 m of roads from the 2014 RIGIS roads layer emphasizing the ease of access to these walls as compared to walls on private lands and under canopy (Table 9).

Conclusions
Successful identification of stone walls for two dates allowed for creation a  (Figure 31).
Since the accuracy of the 1900 is not known, neither is the accuracies of the datasets for matching, built and removed. It is important to mention that over approximately 43% of stone walls within the dataset of built stone walls are located parallel to roads. It is very likely that these stone walls were in existence in 1900 but were not distinguishable on the topographic map. This point would also lead to inaccuracies of the other stone wall datasets as well.
The distribution of stone walls per land cover class to that of the percent of land per land cover class is very similar. However, there is instances in which the percentages of stone walls compared to the percentages of land per land cover class is significant, suggesting that there is more stone walls within agricultural lands from and land cover that is found in this study. While it is logical for there to be more stone walls contained within agricultural and urban lands than water, wetland and barren, ultimately Block Island is completely covered in stone walls. This makes it difficult to synthesize these findings into a conclusion; emphasizing the need to expand the scope of this study in both methodology and selection of sites.
At a local scale, the LCCF of areas containing stone walls was tested against areas absent of stone walls to quantify if stone walls were located in areas with less land cover change throughout the temporal range. No significant difference was found when comparing the buffer zones containing stone walls and zones absent of stone walls. However, it must be considered that the frequency of land cover change within the time period of this study area was minimal as based on approximately 98% of the land being contained within the LCCF classes of either 0 or 1.
The While some relationships were found between the distributions of stone walls and both land cover classes and land change frequency for the date range, this study did not find a very strong relationship between stone walls and land cover overall.
This can be attributed to both the resolutions of the available data assessed and data not assessed which may or may not be available. Additionally, assessing the land cover classifications at a 45 m resolution may have been an oversimplification to these data.

Future Considerations
It is suggested that this study be used as a model to be expanded to other areas which also contain stone walls. It would be specifically interesting to assess areas with fewer stone walls and which have experienced a high magnitude of land cover change, to compare to those areas with opposite conditions. This would greatly enhance understanding of the relationships between stone walls and the temporal landscape.
Hence, expanding understanding of how historic land use interactions can persist. This model only uses an area in which stone walls are high abundant throughout and which has gone through minimal temporal land cover change in the more recent past. So while the buffer analysis did attempt to isolate areas containing stone walls and those in which stone walls were absent, the study site itself limited the size of a buffer to 15m. By assessing areas with different characteristics the range of influence stone walls exhibit can become understood and accurately modeled. Additionally, findings from this study can be more fully understood with integration of other spatial datasets and qualitative historical information. This will allow for both the purpose of the current and the origin of historical stone walls to be better addressed, as based on both environmental and social factors.
Other considerations include the incorporation of stone walls into landscape ecological studies assessing factors which influence land cover change as well as studies within other environmental fields. By identifying locations of stone walls a standalone dataset exists that can become easily accessible. Stone walls can be studied as small mammal habitats and corridors by wildlife biologists and areas for breeding of beetles and ticks by entomologists. Hydrologists can assess the ability for stone walls to influence overland flow and infiltration. Additionally, through field research it has been noticed that stone walls are commonly overtaken by shrubs and this could also be assessed in relation to growth and spread of invasive species.

Conservation of Stone Walls
Through this study the relationship between stone walls and landscape characterization and change has begun to be assessed. There is much potential for this relationship to continue to be analyzed and understood in more depth based on the abundance of stone walls located with the study site and the remaining New England and adjacent states which have stone walls. Stone wall conservation will be a result of human value placed on both stone walls and lands which happen to contain stone walls. As based on this study, there is a clear appreciation for both stone walls and lands. Additionally, the state of Rhode Island places specific value on stone wall conservation enacted through the RI General Law § 45-2-39.1 and RI General Law § 11-41-32 which give penalty to theft of a stone wall and RI General Law §44-3-43 which gives tax exemption to owners of certain types of historic stone walls (RI Gen L § 45-2-39.1 (2013), RI Gen L § 11-41-32 (2014) and RI Gen L § 44-3-43 (2014)).
This leaves us in a positon to suggest that the conservation of stone walls be considered within these studies because the preservation of these features will allow for not only the continued study of stone walls, but also for persistence of the relationships studies find between stone walls and their environment. Through more specific assessments of other factors relating to the conservation of stone walls including government regulations and future projections of land cover change, the persistence of stone walls can be better understood.     Comparison of the amount of both land and stone walls contained within each land cover change frequency class. Classes were determined by calculating the amount of times the land changed between 1988 and 2011 resulting in a range of 0-3. Change values were determined based on 45 m resolution LULC datasets. A two tailed z test for 2 population proportions was used to compare the proportion of stone walls within a given land cover change frequency class to the proportion of land within the same class.

TABLES
* Statistically significant at P<0.05 ` Table 6: The Frequency of Land Cover Change: Comparing areas with stone walls and absent of stone walls 15 meter buffers were created around points located on stone walls from the 2011 dataset. 15 meter buffers were created around an equal number of random points generated which do not overlap with the stone walls. Land cover change frequency information was obtained for each buffer. Areas per land cover change frequency unit were calculated and weighted by multiplying the area by the unit value (0-3). A two-sample t-test was performed on the weighted arrays. Differences in area are attributed to overlapping of buffers around random points which were then dissolved.           Stone walls which were present in the 1900 dataset and the 2011 dataset of stone walls.

Figure 9: Stone Walls Built Between 1900 and 2011
Stone walls which were not present in the 1900 dataset and present in the 2011 dataset of stone walls.

Figure 10: Stone Walls Removed Between 1900 and 2011
Stone walls which were present in the 1900 dataset and not present in the 2011 dataset of stone walls.

Figure 11: 1988 Anderson Level 1 Land Cover Classification
Pre-classified land cover from RIGIS normalized to an Anderson Level 1 Classification with 7 cover classes and 45 m pixel resolution.

Figure 12: 1995 Anderson Level 1 Land Cover Classification
Pre-classified land cover from RIGIS normalized to an Anderson Level 1 Classification with 7 cover classes and 45 m pixel resolution.

Figure 13: 2003/04 Anderson Level 1 Land Cover Classification
Pre-classified land cover from RIGIS normalized to an Anderson Level 1 Classification with 7 cover classes and 45 m pixel resolution.

Figure 14: 2011 Anderson Level 1 Land Cover Classification
Pre-classified land cover from RIGIS normalized to an Anderson Level 1 Classification with 7 cover classes and 45 m pixel resolution.

Figure 20: Land Cover Change Frequency (1988-2011)
Bar graph representing the percentages of land per each land change frequency unit.

Figure 21: Stone Walls per Land Cover Change Frequency Class (1988-2011)
Bar graph representing the percentages of stone walls from the 2011 dataset within each land change frequency unit.

Figure 22: Land Use Change Frequency around Stone Walls
15 meter buffers were created around 3,135 points located on stone walls from the 2011 dataset. 3,135 random point were created 30 meters away from stone walls and 15 meters from the edge of the New Shoreham boundary. 15 meters buffers were created around the random points. Points from both sets were used to extract land change frequency information to determine if there is a difference between the frequencies of land cover change around stone walls compared to areas without stone walls.