Impact of data integration technique on historical land-use/land-cover change: Comparing historical maps with remote sensing data in the Belgian Ardennes

Historical reconstructions of land-use/cover change often require comparing maps derived from different sources. The objective of this study was to measure land-use/cover changes over the last 225 years at the scale of a Belgian landscape, Lierneux in Ardennes, on the basis of a heterogeneous time series of land cover data. The comparability between the land-cover maps was increased following a method of data integration by map generalisation. Two types of time series were built by integrating the maps either by reference to the initial map of the time series or by pair of successive maps. Land-cover change detection was performed on the initial time series without data integration and on the two types of integrated time series. Results reveal that land cover and landscape structure have been subject to profound changes in Lierneux since 1775, with an annual rate of change at the landscape level of up to 1.40%. The major land-cover change processes observed are expansion of grasslands-croplands and reforestation with coniferous species, leading to amore fragmented landscape structure. The annual rates of land-cover change estimated from integrated data are significantly different from the annual rates of change estimated without a prior integration of the data. There is a trade-off between going as far back in time as possibleversus performing change detection as accurately as possible. This revised version was published online in July 2006 with corrections to the Cover Date.     

Impact of error on landscape pattern analyses performed on land-cover change maps

Researchers have emphasized the value of linking observed patterns of land-cover change to the processes driving changes in land-use to explain the dynamics of a land change system. The association of pattern and process requires an accurate quantification of the spatial characteristics of land-cover change. The objective of this research is to assess the impact of error on the accuracy of landscape pattern analyses performed on maps of change. Simulation was used to develop of a series of error-free and error-perturbed change maps, which varied with respect to the pattern of change occurring between the time-1 and time-2 land-cover maps and the patterns of error associated with the time-1 and time-2 land-cover maps. A variety of change and error patterns were examined. The error-free and error-perturbed change maps were compared by calculating landscape pattern metrics, which revealed the degree to which error altered the pattern of change. The introduction of error notably changed the structure of the persistent and transitioning classes, with metrics indicating a more fragmented and variable landscape under error. Agreement between the error-free and error-perturbed maps improved when a greater amount of change occurred within the time-series, change was concentrated at the boundaries of land-cover classes and when time-2 errors were increasingly correlated to their time-1 counterparts. These results have several implications for change pattern analyses given the fundamental nature of land-cover change.     

Land-cover and structural changes in a western Norwegian cultural landscape since 1865, based on an old cadastral map and a field survey

Many studies of land-cover and structural changes in cultural landscapes have used historical maps as a source for information about past land-cover. All transformations of historical maps onto modern coordinate systems are however burdened with difficulties when it comes to accuracy. We show that a detailed land survey of the present landscape may enable transformation of an old cadastral map directly onto the present terrain with very high accuracy. The detailed resulting map enabled us to locate remnants of semi-natural grasslands and man-made structures with continuity from 1865 and to test hypotheses about relationships between landscape changes and landscape characteristics. The main land-cover change 1865–2002 was decrease of arable fields, and addition of three new land-cover classes: horticultural, orchard and abandoned areas. Of the 330 man-made structures present in 1865, only 58 remained in 2002, while 63 new structures had been built after 1865. We found that semi-natural grasslands with continuity since 1865 were situated on ground with significantly lower production capacity than mean 1865 production capacity. The man-made structures with continuity since 1865 were also associated with areas with significantly lower production capacity than the 1865 mean, situated in significantly steeper terrain but not further from the hamlet. Our study illustrates the potential of digitised and accurately transformed historical cadastral maps combined with detailed field surveys for analysis of land-cover and structural changes in the cultural landscape.     

Estimating landscape pattern metrics from a sample of land cover

Although landscape pattern metrics can be computed directly from wall-to-wall land-cover maps, statistical sampling offers a practical alternative when complete coverage land-cover information is unavailable. Partitioning a region into spatial units and then selecting a subset (sample) of these units introduces artificial patch edge and patch truncation effects that may lead to biased sample-based estimators of landscape pattern metrics. The bias and variance of sample-based estimators of status and change in landscape pattern metrics were evaluated for four 120-km × 120-km test regions with land cover provided by the 1992 and 2001 National Land-Cover Data of the United States. Bias was generally small for both the estimators of status and estimators of change in landscape pattern, but exceptions to this favorable result exist and it is advisable to assess bias for the specific metrics and region of interest in any given application. A 10-km × 10-km sample block generally yielded larger biases but smaller variances for the estimators relative to a 20-km × 20-km sample block. Stratified random sampling improved precision of the estimators relative to simple random sampling. The methodology developed to determine properties of sample-based estimators can be readily extended to evaluate other landscape pattern metrics, regions, and sample block sizes.     

Global analysis and simulation of land-use change associated with urbanization

A combination of rapid population growth and an accelerating demographic shift from rural to urbanized habitats has resulted in urbanization becoming an increasingly global phenomenon. Two alternate hypotheses describing urban landscape trajectories suggest urbanization is either leading to more homogeneous global patterns or urbanization has dichotomous trajectories of increasing dispersal or coalescence. To better understand the global variation in urban land-cover patterns and trajectories we described the variation in urban landscape structure for 120 cities distributed throughout the world assessed at circa 1990 and 2000. We coupled these data to a low-dimensional neighborhood based model of urban growth using a data-model fusion approach. Trajectories of urban growth were assessed using both the original data and model projections to 2030. The patterns of landscape change were related to both the rate of growth and income. The historical patterns of change showed a trend of increasing landscape complexity and this trend was projected to continue. Urban rate of growth was closely related to the change in several landscape metrics. Income was associated with landscape dynamics and this effect interacted with city size. Large cities were less sensitive to the income effect than small cities. Along with changes to the magnitude of each metric, the overall variation in metrics between years generally exhibited a decrease in variability and this variability was projected to continue decreasing. These findings supported the hypothesis that urban landscapes are becoming more homogeneous and that the dispersal-coalescing dichotomy represent endpoints rather than alternate states of urban growth.     

Analysing land-cover changes in relation to environmental variables in Hesse,Germany

Land-use and land-cover changes affect ecological landscape functions and processes. Hence, landscape ecologists have a central interest in a comprehensive understanding of such changes. Our study focuses on the relationships between environmental conditions and agricultural land-cover changes. We present a method to (i) characterise the major spatial-temporal processes of land-cover changes, (ii) identify the correlations between environmental attributes and land-cover changes and (iii) derive potential environmental drivers of land-cover changes in a German marginal rural landscape. The method was applied to study land-cover dynamics from 1945 to 1998 in the districts of Erda, Steinbrücken and Eibelshausen, situated in the marginal rural landscape of the Lahn-Dill Highlands, Germany. We employed land-cover data gained by the interpretation of multi-temporal aerial photographs. Various environmental variables were introduced into the analyses. We identified physical landscape attributes (elevation, slope, aspect, available water capacity and soil texture) and structural landscape dimensions (patch size, patch shape and distance between patch and nearest settlement). With the aid of GIS, K-means partitioning and canonical correspondence analysis, we investigated land-cover trajectory types, land-cover transitions at individual time intervals and their relationships to these environmental variables. Our results show that, between 1945 and 1998, land-cover changes correlated with the physical attributes of the underlying landscape. On the other hand, the structural landscape dimensions correlated with land cover only in periods of minor land-cover changes (1972–98). Greater diversity of physical landscape attributes is correlated with greater land-cover dynamics. Besides the important influence of socio-economic factors, land-cover changes in the study areas took place within the relatively stable physical constraints of the underlying landscape.This revised version was published online in May 2005 with corrections to the Cover Date.     

Analysis of land-cover transitions based on 17th and 18th century cadastral maps and aerial photographs

This paper explores the possibility of using non-geometric cadastral maps from the 17th and 18th century together with aerial photographs from 1945 and 1981 to analyse land-cover change in south-east Sweden. Habitats rich in plant species in the European rural landscape seem to be correlated with a long continuity of management. Accurate spatial data from historical data sources are fundamental to understand patterns of vegetation and biodiversity in the present-day landscape. However, traditional methods for rectification of non-geometric maps using corresponding points from orthophotos or modern maps are not satisfying, as internal inaccuracies will remain in the maps. This study presents a method to rectify the maps by local warping, thereby eliminating geometrical irregularities. Further, the land-cover changes were calculated and presented as transition matrices. The extent of arable fields and grasslands were analysed in relation to soil characteristics and continuity of management. The results show a dynamic relation between grassland and arable field, albeit the overall proportions remained almost the same between 17th and 18th centuries: 60% grassland to 32% arable field. The most substantial changes in land-cover were prior to 1945. Today there is 18% grasslands left in the study area, while 56% of the land-cover is arable field. Approximately 8% of present-day land-cover is semi-natural grassland 300 years of age or more. Compared to 300 years ago there is only 1% grassland left on peat and 2% on clay. In contrast, grassland covers associated with bare bedrock have been fairly stable in size. All semi-natural grasslands with a long continuity of management were situated on shallow soils, less than 50 cm depth. The major conclusions from this study are that (i) correctly rectified, old maps are very useful to address questions of land-cover changes in historical time, (ii) general trends in land use over 300 years in this hemi-boreal landscape seem to underestimate the full dynamics of land use change, and (iii) only a small proportion of the semi-natural grassland area had a 300 year continuity of management.     

Old-growth forest landscape transitions from pre-European settlement to present

We conducted a multi-temporal spatial analysis of forest cover for a 9600 ha landscape in northern Wisconsin, U.S.A., using data from pre-European settlement (1860s), post-settlement (1931), and current (1989) periods. Using GIS we have shown forest landscape changes and trajectories that have been generally described in aggregate for the norther Great Lake States region. We created the pre-European settlement map from the witness tree data of the original federal General Land Office survey notes. The 1931 cover was produced from the Wisconsin Land Economic Inventory, and the 1989 cover map was based on color infrared photography. We used GIS to analyze 1) land area occupied by different forest types at different dates, 2) temporal transitions between dates and their driving proceses, and 3) successional trajectories with landforms and spatial associations of forest types. Over the 120 year period, forest cover has changed from a landscape dominated by old-growth hemlock (Tsuga canadensis) and hardwood forests (Acer saccharum, Betula alleghaniensis) to largely second-growth hardwoods and conifers. The former dominant hemlock is largely eliminated from the landscape. From 1860 to 1931, large-scale disturbances associated with logging were the dominant processes on the landscape. Early successional forest types covered much of the landscape by the 1930s. From 1931 to 1989, succession was the dominant process driving forest transitions as forest types succeeded to a diverse group of upland hardwood and conifer forest types. If successional trajectories continue, a more homogeneous landscape may develop comprised of both a northern hardwood type dominated by sugar maple, and a boreal conifer/hardwood forest.     

Quantifying Spatiotemporal Patterns of Urban Land-use Change in Four Cities of China with Time Series Landscape Metrics

This paper provides a dynamic inter- and intra-city analysis of spatial and temporal patterns of urban land-use change. It is the first comparative analysis of a system of rapidly developing cities with landscape pattern metrics. Using ten classified Landsat Thematic Mapper images acquired from 1988 to 1999, we quantify the annual rate of urban land-use change for four cities in southern China. The classified images were used to generate annual maps of urban extent, and landscape metrics were calculated and analyzed spatiotemporally across three buffer zones for each city for each year. The study shows that for comprehensive understanding of the shapes and trajectories of urban expansion, a spatiotemporal landscape metrics analysis across buffer zones is an improvement over using only urban growth rates. This type of analysis can also be used to infer underlying social, economic, and political processes that drive the observed urban forms. The results indicate that urban form can be quite malleable over relatively short periods of time. Despite different economic development and policy histories, the four cities exhibit common patterns in their shape, size, and growth rates, suggesting a convergence toward a standard urban form.     

Neutral models for the analysis of broad-scale landscape pattern

The relationship between a landscape process and observed patterns can be rigorously tested only if the expected pattern in the absence of the process is known. We used methods derived from percolation theory to construct neutral landscape models,i.e., models lacking effects due to topography, contagion, disturbance history, and related ecological processes. This paper analyzes the patterns generated by these models, and compares the results with observed landscape patterns. The analysis shows that number, size, and shape of patches changes as a function of p, the fraction of the landscape occupied by the habitat type of interest, and m, the linear dimension of the map. The adaptation of percolation theory to finite scales provides a baseline for statistical comparison with landscape data. When USGS land use data (LUDA) maps are compared to random maps produced by percolation models, significant differences in the number, size distribution, and the area/perimeter (fractal dimension) indices of patches were found. These results make it possible to define the appropriate scales at which disturbance and landscape processes interact to affect landscape patterns.     

Detecting human-driven deviations from trajectories in landscape composition and configuration

While landscape trajectories are increasingly used for tracking change in processes such as agricultural intensification and urbanization, analyses that combine environmental and human disturbances remain scarce. The aim of this study was to investigate the relationship between Shannon evenness, a measure of landscape composition, and spatial contagion, a measure of landscape configuration, within sixteen Canadian regions covering a gradient of land-uses and human disturbances: natural, semi-natural, urban, and agricultural. The agricultural regions showed generally lower variation in contagion and evenness and overall lower contagion values (smaller patches), leading to steeper contagion-evenness slopes than in the other region categories. In addition, the sampled agricultural regions were much more similar to each other than were the sampled regions within each of the other three region categories. These results indicate that the process of agricultural development (at least in western Canada) leads to a reduction in pattern variation and an alteration of the expected relationships among pattern metrics in agricultural regions. This possibility is supported by a neutral model of patch dynamics, suggesting that the characteristic scale of disturbances is a generic structuring process of landscape trajectories.     

Influence of biophysical factors and differences in Ojibwe reservation versus Euro-American social histories on forest landscape change in northern Wisconsin,USA

Landscape ecology studies have demonstrated that past modifications of the landscape frequently influence its structure, highlighting the utility of integrating historical perspectives from the fields of historical ecology and environmental history. Yet questions remain for historically-informed landscape ecology, especially the relative influence of social factors, compared to biophysical factors, on long-term land-cover change. Moreover, methods are needed to more effectively link history to ecology, specifically to illuminate the underlying political, economic, and cultural forces that influence heterogeneous human drivers of land-cover change. In northern Wisconsin, USA, we assess the magnitude of human historical forces, relative to biophysical factors, on land-cover change of a landscape dominated by eastern white pine (Pinus strobus L.) forest before Euro-American settlement. First, we characterize land-cover transitions of pine-dominant sites over three intervals (1860–1931; 1931–1951; 1951–1987). Transition analysis shows that white pine was replaced by secondary successional forest communities and agricultural land-covers. Second, we assess the relative influence of a socio-historical variable (“on-/off-Indian reservation”), soil texture (clay and sand), and elevation on land-cover transition. On the Lake Superior clay plain, models that combine socio-historical and biophysical variables best explain long-term land-cover change. The socio-historical variable dominates: the magnitude and rate of land-cover change differs among regions exposed to contrasting human histories. Third, we developed an integrative environmental history-landscape ecology approach, thereby facilitating linkage of observed land-cover transitions to broader political, economic, and cultural forces. These results are relevant to other landscape investigations that integrate history and ecology.     

Landscape patterns from mathematical morphology on maps with contagion

The perceived realism of simulated maps with contagion (spatial autocorrelation) has led to their use for comparing landscape pattern metrics and as habitat maps for modeling organism movement across landscapes. The objective of this study was to conduct a neutral model analysis of pattern metrics defined by morphological spatial pattern analysis (MSPA) on maps with contagion, with comparisons to phase transitions (abrupt changes) of patterns on simple random maps. Using MSPA, each focal class pixel on a neutral map was assigned to one of six pattern classes—core, edge, perforated, connector, branch, or islet—depending on MSPA rules for connectivity and edge width. As the density of the focal class (P) was increased on simple random maps, the proportions of pixels in different pattern classes exhibited two types of phase transitions at threshold densities (0.41 ≤ P ≤ 0.99) that were predicted by percolation theory after taking into account the MSPA rules for connectivity and edge width. While there was no evidence of phase transitions on maps with contagion, the general trends of pattern metrics in relation to P were similar to simple random maps. Using an index P for comparisons, the effect of increasing contagion was opposite that of increasing edge width.     

Statistical models of landscape pattern metrics,with applications to regional scale dynamic forest simulations

Forest managers in Canada need to model landscape pattern or spatial configurationoverlarge (100,000 km²) regions. This presents a scalingproblem, as landscape configuration is measured at a high spatial resolution,but a low spatial resolution is indicated for regional simulation. We present astatistical solution to this scaling problem by showing how a wide range oflandscape pattern metrics can be modelled from low resolution data. Our studyarea comprises about 75,000 km² of boreal mixedwoodforest in northeast Alberta, Canada. Within this area we gridded a sample of 84digital forest cover maps, each about 9500 ha in size, to aresolution of 1 ha and used FRAGSTATS to compute a suite oflandscape pattern metrics for each map. We then used multivariate dimensionreduction techniques and canonical correlation analysis to model therelationship between landscape pattern metrics and simpler stand table metricsthat are easily obtained from non-spatial forest inventories. These analyseswere performed on four habitat types common in boreal mixedwood forests: youngdeciduous, old deciduous, white spruce, and mixedwood types. Using only threelandscape variables obtained directly from stand attribute tables (totalhabitatarea, and the mean and standard deviation of habitat patch size), ourstatistical models explained more than 73% of the joint variation in fivelandscape pattern metrics (representing patch shape, forest interior habitat,and patch isolation). By PCA, these five indices captured much of the totalvariability in the rich set of landscape pattern metrics that FRAGSTATS cangenerate. The predictor variables and strengths of association were highlyconsistent across habitat classes. We illustrate the potential use of suchstatistical relationships by simulating the regional, cumulative effects ofwildfire and forest management on the spatial arrangement of forest patches,using non-spatial stand attribute tables.This revised version was published online in May 2005 with corrections to the Cover Date.     

Historic range of variability in landscape structure in subalpine forests of the Greater Yellowstone Area,USA

A measure of the historic range of variability (HRV) in landscape structure is essential for evaluating current landscape patterns of Rocky Mountain coniferous forests that have been subjected to intensive timber harvest. We used a geographic information system (GIS) and FRAGSTATS to calculate key landscape metrics on two ∼130,000-ha landscapes in the Greater Yellowstone Area, USA: one in Yellowstone National Park (YNP), which has been primarily shaped by natural fires, and a second in the adjacent Targhee National Forest (TNF), which has undergone intensive clearcutting for nearly 30 years. Digital maps of the current and historical landscape in YNP were developed from earlier stand age maps developed by Romme and Despain. Maps of the TNF landscape were adapted from United States Forest Service Resource Information System (RIS) data. Key landscape metrics were calculated at 20-yr intervals for YNP for the period from 1705-1995. These metrics were used to first evaluate the relative effects of small vs. large fire events on landscape structure and were then compared to similar metrics calculated for both pre- and post-harvest landscapes of the TNF. Large fires, such as those that burned in 1988, produced a structurally different landscape than did previous, smaller fires (1705-1985). The total number of patches of all types was higher after 1988 (694 vs. 340-404 before 1988), and mean patch size was reduced by almost half (186 ha vs. 319-379 ha). The amount of unburned forest was less following the 1988 fires (63% vs. 72-90% prior to 1988), yet the number of unburned patches increased by nearly an order of magnitude (230 vs. a maximum of 41 prior to 1988). Total core area and mean core area per patch decreased after 1988 relative to smaller fires (∼73,700 ha vs. 87,000-110,000 ha, and 320 ha vs. 2,123 ha, respectively). Notably, only edge density was similar (17 m ha⁻¹ after 1988) to earlier landscapes (9.8-14.2 m ha⁻¹).Three decades of timber harvesting dramatically altered landscape structure in the TNF. Total number of patches increased threefold (1,481 after harvest vs. 437 before harvest), and mean patch size decreased by ∼70% (91.3 ha vs. 309 ha). None of the post-harvest landscape metrics calculated for the TNF fell within the HRV as defined in YNP, even when the post-1988 landscape was considered. In contrast, pre-harvest TNF landscape metrics were all within, or very nearly within, the HRV for YNP. While reference conditions such as those identified by this study are useful for local and regional landscape evaluation and planning, additional research is necessary to understand the consequences of changes in landscape structure for population, community, ecosystem, and landscape function. This revised version was published online in July 2006 with corrections to the Cover Date.     

Multiscale ecological assessment of remote sensing images

In landscape ecology, the importance of map extent and resolution on the value of landscape indices is widely discussed, but the information content of the map, mostly derived from remote sensing images, is not. In this study, we sought (1) to understand the influence of changes in maps’ spatial and spectral resolution of agricultural landscape elements, taking hedgerow networks as a case study, and (2) to explore how predictions of species distribution might be affected by maps’ resolutions, taking two carabid species as a case study. To do so, we compared maps from different remote sensors, derived two landscape characterization variables from the maps related to patterns known to drive ecological processes, and analyzed their predictive power on biological data distribution to assess the information content of these maps. The results show that (1) the use of several methods, including landscape metrics, was useful to assess map validity; (2) the spatial resolution of satellite images is not the only important factor; changes in spectral resolution significantly alter maps; (3) the relevant definition of “hedgerow” to construct functional maps is species and process specific; thus the different maps are not either good or bad, but rather provide complementary information; (4) the more a species responds to network structure and over small areas, the less the different maps can be substitutable one to another.     

Neutral models for testing landscape hypotheses

Neutral landscape models were originally developed to test the hypothesis that human-induced fragmentation produces patterns distinctly different from those associated with random processes. Other uses for neutral models have become apparent, including the development and testing of landscape metrics to characterize landscape pattern. Although metric development proved to be significant, the focus on metrics obscured the need for iterative hypothesis testing fundamental to the advancement of the discipline. We present here an example of an alternative neutral model and hypothesis designed to relate the process of landscape change to observed landscape patterns. The methods and program, QRULE, are described and options for statistical testing outlined. The results show that human fragmentation of landscapes results in a non-random association of land-cover types that can be describe by simple statistical methods. Options for additional landscape studies are discussed and access to QRULE described in the hope that these methods will be employed to advance our understanding of the processes that affect the structure and function in human dominated landscapes.     

A new approach for rescaling land cover data

The resolution of satellite imagery must often be increased or decreased to fill data gaps or match preexisting project requirements. It is well known that a change in resolution introduces systematic errors of size, shape, location and amount of contiguous land cover types. Nevertheless, robust methods for rescaling landscape data are frequently required to assess patterns of landscape change through time and over large areas. We developed a new method for rescaling spatial data that allows map resolution (grain size) to be either increased or decreased while holding the total proportion of land cover types constant. The method uses a weighted sampling net of variable resolution to sample an existing map and then randomly selects from the frequency of cover types derived from this sample to assign the cover type for the corresponding location in the rescaled map. The properties of the sampling net had a variable effect on measures of landscape pattern with the characteristic patch size (S) the most robust metric and the number of clusters (A) the most variable. A comparison of up-scaled and down-scaled maps showed that this process is not symmetrical, producing different errors for increases versus decreases in grain size. Rescaling Landsat (30 m) imagery to the 10 m resolution of SPOT imagery for four National Park units within Maryland and Virginia resulted in errors due to rescaling that were small (1–2%) relative to the total error (∼11%) associated with these images. The new rescaling method is general because it provides a single method for increasing or decreasing resolution, can be applied to maps with multiple land cover types, allows grid geometry to be transformed (i.e., square to hexagonal grids), and provide a more consistent basis for landscape comparisons when maps must be derived from multiple sources of classified imagery.     

Quantifying the effect of grazing and shrub-clearing on small scale spatial pattern of vegetation

Disturbances such as grazing, invading species, and clear-cutting, often act at small spatial scales, and means for quantifying their impact on fine scale vegetation patterns are generally lacking. Here we adopt a set of landscape metrics, commonly used for quantifying coarse scale fragmentation, to quantify fine scale fragmentation, namely the fine scale vegetation structure. At this scale, patches often consist of individual plants smaller than 1 m², requiring the grain of the analysis to be much smaller. We used balloon aerial photographs to map fine details of Mediterranean vegetation (pixel size <0.04 m) in experimental plots subjected to grazing and clear-cutting and in undisturbed plots. Landscape metrics are sensitive to scale. Therefore, we aggregated the vegetation map into four coarser scales, up to a resolution of 1 m, and analyzed the effect of scale on the metrics and their ability to distinguish between different disturbances. At the finest scale, six of the seven landscape metrics we evaluated revealed significant differences between treated and undisturbed plots. Four metrics revealed differences between grazed and control plots, and six metrics revealed differences between cleared and control plots. The majority of metrics exhibited scaling relations. Aggregation had mixed effects on the differences between metric values for different disturbances. The control plots were the most sensitive to scale, followed by grazing and clearing. We conclude that landscape metrics are useful for quantifying the very fine scale impact of disturbance on woody vegetation, assuming that the analysis is based on sufficiently high spatial resolution data.