Toward estimation of map accuracy without a probability test sample

The time and effort required of probability sampling for accuracy assessment of large-scale land cover maps often means that probability test samples are not collected. Yet, map usefulness is substantially reduced without reliable accuracy estimates. In this article, we introduce a method of estimating the accuracy of a classified map that does not utilize a test sample in the usual sense, but instead estimates the probability of correct classification for each map unit using only the classification rule and the map unit covariates. We argue that the method is an improvement over conventional estimators, though it does not eliminate the need for probability sampling. The method also provides a new and simple method of constructing accuracy maps. We illustrate some of problems associated with accuracy assessment of broad-scale land cover maps, and our method, with a set of nine Landsat Thematic Mapper satellite image-based land cover maps from Montana and Wyoming, USA.     

A geostatistical approach for mapping thematic classification accuracy and evaluating the impact of inaccurate spatial data on ecological model predictions

Spatial information in the form of geographical information system coverages and remotely sensed imagery is increasingly used in ecological modeling. Examples include maps of land cover type from which ecologically relevant properties, such as biomass or leaf area index, are derived. Spatial information, however, is not error-free: acquisition and processing errors, as well as the complexity of the physical processes involved, make remotely sensed data imperfect measurements of ecological attributes. It is therefore important to first assess the accuracy of the spatial information being used and then evaluate the impact of such inaccurate information on ecological model predictions. In this paper, the role of geostatistics for mapping thematic classification accuracy through integration of abundant image-derived (soft) and sparse higher accuracy (hard) class labels is presented. Such assessment leads to local indices of map quality, which can be used for guiding additional ground surveys. Stochastic simulation is proposed for generating multiple alternative realizations (maps) of the spatial distribution of the higher accuracy class labels over the study area. All simulated realizations are consistent with the available pieces of information (hard and soft labels) up to their validated level of accuracy. The simulated alternative class label representations can be used for assessing joint spatial accuracy, i.e., classification accuracy regarding entire spatial features read from the thematic map. Such realizations can also serve as input parameters to spatially explicit ecological models; the resulting distribution of ecological responses provides a model of uncertainty regarding the ecological model prediction. A case study illustrates the generation of alternative land cover maps for a Landsat Thematic Mapper (TM) subscene, and the subsequent construction of local map quality indices. Simulated land cover maps are then input into a biogeochemical model for assessing uncertainty regarding net primary production (NPP).     

Multi-lag cluster designs for estimating the semivariogram for sediments affected by effluent discharges offshore in San Diego

Maps are useful tools for understanding, managing, and protecting the marine environment, yet few useful and statistically defensible maps of environmental quality and aquatic resources have been developed in near-coastal regions. Current environmental management efforts, such as ocean monitoring by sewage dischargers, routinely sample areas of potential impact using sparse sampling grids. Heterogeneous oceanic conditions often make extrapolation from these grids to non-sampled locations questionable. Although rarely applied in coastal monitoring, kriging offers a more rigorous statistical approach to mapping and allows confidence intervals to be calculated for predictions. Its usefulness relies on accurate models of the spatial variability through estimating the semivariogram. Many optimal designs for estimating the semivariogram have been proposed, but these designs are often difficult to implement in practice. In this paper, we present simple design strategies for augmenting existing monitoring designs with the goal of estimating the semivariogram. In particular, we investigate a multi-lag cluster design strategy, where clusters of sites, spaced at various lag distances, are placed around fixed stations on an existing sampling grid. We find that these multi-lag cluster designs provide improved accuracy in estimating the parameters of the semivariogram. Based on simulation study findings, we apply a multi-lag cluster enhancement to the monitoring grid for the City of San Diego’s Point Loma Wastewater Treatment Plant as part of a special study to map chemical contaminants in sediments around its sewage outfall.     

Detection of current and potential hazelnut plantation areas in Trabzon, North East Turkey using GIS and RS

Monitoring agricultural products requires the periodic determination of land cover and the production of land use policies in an optimum way. The hazelnut is one of the important Turkish agricultural exports and Turkey provides 77% of the world's hazelnuts. In Turkey, hazelnut production exceeds the demand; new regulations have been enacted to create new land use policies. By putting into practice regulations restricting hazelnut plantation areas, a more efficient and productive hazelnut harvest policy could be created. Therefore, more information on existing land cover is required to determine optimum (or ideal) potential hazelnut areas (PHA) and to forecast future crop production. The principle aim of this study is to create a methodology for determining existing PHA, using Geographic information system (GIS) and remote sensing (RS) techniques regarding to support hazelnut policy developers and economists. This study was basically carried out in the province of Trabzon, which is one of the most important hazelnut production areas in Turkey. Landsat ETM+ image was used to generate a current land cover classification. Using the supervised classification method, overall accuracy was determined to be 84.7%. Suitable hazelnut areas were determined according to criteria settled by government regulations.     

Modeled historical land use and land cover for the conterminous United States

The landscape of the conterminous United States has changed dramatically over the last 200 years, with agricultural land use, urban expansion, forestry, and other anthropogenic activities altering land cover across vast swaths of the country. While land use and land cover (LULC) models have been developed to model potential future LULC change, few efforts have focused on recreating historical landscapes. Researchers at the US Geological Survey have used a wide range of historical data sources and a spatially explicit modeling framework to model spatially explicit historical LULC change in the conterminous United States from 1992 back to 1938. Annual LULC maps were produced at 250-m resolution, with 14 LULC classes. Assessment of model results showed good agreement with trends and spatial patterns in historical data sources such as the Census of Agriculture and historical housing density data, although comparison with historical data is complicated by definitional and methodological differences. The completion of this dataset allows researchers to assess historical LULC impacts on a range of ecological processes.     

Use of land facets to design linkages for climate change

Least-cost modeling for focal species is the most widely used method for designing conservation corridors and linkages. However, these linkages have been based on current species' distributions and land cover, both of which will change with large-scale climate change. One method to develop corridors that facilitate species' shifting distributions is to incorporate climate models into their design. But this approach is enormously complex and prone to error propagation. It also produces outputs at a grain size (km2) coarser than the grain at which conservation decisions are made. One way to avoid these problems is to design linkages for the continuity and interspersion of land facets, or recurring landscape units of relatively uniform topography and soils. This coarse-filter approach aims to conserve the arenas of biological activity rather than the temporary occupants of those arenas. In this paper, we demonstrate how land facets can be defined in a rule-based and adaptable way, and how they can be used for linkage design in the face of climate change. We used fuzzy c-means cluster analysis to define land facets with respect to four topographic variables (elevation, slope angle, solar insolation, and topographic position), and least-cost analysis to design linkages that include one corridor per land facet. To demonstrate the flexibility of our procedures, we designed linkages using land facets in three topographically diverse landscapes in Arizona, USA. Our procedures can use other variables, including soil variables, to define land facets. We advocate using land facets to complement, rather than replace, existing focal species approaches to linkage design. This approach can be used even in regions lacking land cover maps and is not affected by the bias and patchiness common in species occurrence data.     

A Bayesian approach to multi-source forest area estimation

In efforts such as land use change monitoring, carbon budgeting, and forecasting ecological conditions and timber supply, there is increasing demand for regional and national data layers depicting forest cover. These data layers must permit small area estimates of forest area and, most importantly, provide associated error estimates. This paper presents a model-based approach for coupling mid-resolution satellite imagery with plot-based forest inventory data to produce estimates of probability of forest and associated error at the pixel-level. The proposed Bayesian hierarchical model provides access to each pixel’s posterior predictive distribution allowing for a highly flexible analysis of pixel and multi-pixel areas of interest. The paper presents a trial using multiple dates of Landsat imagery and USDA Forest Service Forest Inventory and Analysis plot data. The results describe the spatial dependence structure within the trial site, provide pixel and multi-pixel summaries of probability of forest land use, and explore discretization schemes of the posterior predictive distributions to forest and non-forest classes. Model prediction results of a holdout set analysis suggest the proposed model provides high classification accuracy, 88%, for the trial site.

Two neighborhood-free plot designs for adaptive sampling of forests

Adaptive cluster sampling (ACS) has the potential of being superior for sampling rare and geographically clustered populations. However, setting up an efficient ACS design is challenging. In this study, two adaptive plot designs are proposed as alternatives: one for fixed-area plot sampling and the other for relascope sampling (also known as variable radius plot sampling). Neither includes a neighborhood search which makes them much easier to execute. They do, however, include a conditional plot expansion: at a sample point where a predefined condition is satisfied, sampling is extended to a predefined larger cluster-plot or a larger relascope plot. Design-unbiased estimators of population total and its variance are derived for each proposed design, and they are applied to ten artificial and one real tree position maps to estimate density (number of trees per ha) and basal area (the cross-sectional area of a tree stem at breast height) per hectare. The performances—in terms of relative standard error (SE%)—of the proposed designs and their non-adaptive alternatives are compared. The adaptive plot designs were superior for the clustered populations in all cases of equal sample sizes and in some cases of equal area of sample plots. However, the improvement depends on: (1) the plot size factor; (2) the critical value (the minimum number of trees triggering an expansion); (3) the subplot distance for the adapted cluster-plots, and (4) the spatial arrangement of the sampled population. For some spatial arrangements, the improvement is relatively small. The adaptive designs may be particularly attractive for sampling in rare and compactly clustered populations with critical value of 1, subplot distance equal to the diameter of initial circular plots, or plot size factor of 2.5 for an initial basal area factor of 2.     

Methods to summarize change among land categories across time intervals

Time-series maps have become more detailed in terms of numbers of categories and time points. Our paper proposes methods for raster datasets where detailed analysis of all categorical transitions would be initially overwhelming. We create two measurements: Incidents and States. The former is the number of times a pixel’s category changes across time intervals; the latter is the number of categories that a pixel represents across time points. The combinations of Incidents and States summarize change trajectories. We also describe categorical transitions in terms of annual flow matrices, which quantify the additional information generated by intermediate time points within the temporal extent. Our approach summarizes change at the pixel and landscape levels in ways that communicate where and how categories transition over time. These methods are useful to detect hotspots of change and to consider whether the apparent changes are real or due to map error.     

Effects of preference heterogeneity among landowners on spatial conservation prioritization

The participation of private landowners in conservation is crucial to efficient biodiversity conservation. This is especially the case in settings where the share of private ownership is large and the economic costs associated with land acquisition are high. We used probit regression analysis and historical participation data to examine the likelihood of participation of Danish forest owners in a voluntary conservation program. We used the results to spatially predict the likelihood of participation of all forest owners in Denmark. We merged spatial data on the presence of forest, cadastral information on participation contracts, and individual‐level socioeconomic information about the forest owners and their households. We included predicted participation in a probability model for species survival. Uninformed and informed (included land owner characteristics) models were then incorporated into a spatial prioritization for conservation of unmanaged forests. The choice models are based on sociodemographic data on the entire population of Danish forest owners and historical data on their participation in conservation schemes. Inclusion in the model of information on private landowners’ willingness to supply land for conservation yielded at intermediate budget levels up to 30% more expected species coverage than the uninformed prioritization scheme. Our landowner‐choice model provides an example of moving toward more implementable conservation planning.     

Trees and rural households’ adaptation to local environmental change in the central highlands of Ethiopia

This study assesses the role of trees in adaptation strategies of rural households to local environmental change in the central highlands of Ethiopia. Change in tree cover was assessed by producing Land Use and Land Cover (LULC) maps using satellite remote sensing images, and household survey was conducted to generate socioeconomic data. The results show that tree cover has increased over the last 30 years, mostly in the form of eucalyptus woodlots around homesteads. Eucalyptus reportedly helps households pass through livelihood shocks and provide protection against negative effects of climate change and variability. Despite some concerns on the part of local agricultural experts that planting eucalyptus may affect future food production, farmers are converting their croplands into eucalyptus woodlots. We conclude that land use planning and proper allocation of land resource is important to improve local livelihoods while also supporting adaptation of communities to local environmental change in general and climate change in particular.     

Using earth observation data from multiple sources to map rare habitats in a coastal conservation area

Kenfig NNR (National Nature Reserve) is a coastal sand dune system in south Wales, UK. The site is an important location for the conservation of the fen orchidLiparis loeselii, a significant proportion of the UK population is found solely on the site. Approaches to the mapping and monitoring of the habitats at Kenfig NNR using EO (Earth Observation) methods are investigated. Typical airborne EO missions over such sites produce more than a single source of EO data; these may include various optical imaging sensors with different spectral ranges, film cameras and ranging devices to measure topography. Conservation managers are thus presented with the problem of which sources of data to use when producing a land cover map of the site of interest. Using a data set gathered over the Kenfig NNR site, we investigate land cover mapping methods for conservation. The land cover types of interest typically cover small areas within a much larger site so they present a hard problem for the EO data and associated classification methods to solve. Land cover classifications produced from the data sets provide a set of competing hypotheses of land cover type for the site. Methods we use to resolve this competition between the data sets include voting methods, data fusion methods and a method utilising fuzzy logic to aggregate information. This paper is intended to act as an introduction to some of the issues involved in using EO data for habitat mapping in highly heterogeneous coastal dune environments and to present some preliminary results of the performance of each method.     

Modeling and interpreting the accuracy assessment error matrix for a doubly classified map

This paper considers two maps having the same spatial extent and the same mapping categories but where each map is subject to classification error. An overlay of the maps yields a (dis)similarity matrix whose (i, j)-entry is the areal proportion placed into category i by the first map and into category j by the second map. A parametric model, called the latent truth model, is proposed which specifies the dissimilarity matrix in terms of the true (but unknown) proportions for the mapping categories as well as the unknown error rates for the two maps. The number of parameters in the model exceeds the degrees of freedom in the dissimilarity matrix. However, a method of regularization is applied to effectively reduce the dimension of the parameter space and to permit model fitting. From the fitted model, one obtains estimates for the true mapping proportions as well as estimated error matrices for each of the maps. Accuracy assessment characteristics for each map (such as user's accuracy, producer's accuracy, overall accuracy, and the kappa coefficient) can be computed from the estimated error matrices. Methods are illustrated with two landcover maps of Wicomico County, Maryland.     

Land cover changes in the Caroni Swamp Ramsar Site, Trinidad (1942 and 2007): implications for management

Caroni Swamp Ramsar Site, the largest mangrove dominated wetland in Trinidad and Tobago, continues to be impacted by human activities. This study examines changes in land cover and land use from 1942 to 2007 using remote sensing technology, geographic information systems (GIS) and extensive field surveys. Land cover maps were produced for 1942, 1957, 1986, 1994, 2003 and 2007 from aerial photos and high resolution satellite imagery. Caroni Swamp’s hydrology was altered in the 1920’s to facilitate rice cultivation. This resulted in the formation of large tracts of freshwater marsh. From 1942 to 1957, freshwater marsh and agriculture lands increased, but after this period there was a steady decline in both, as freshwater was diverted away from the wetland and salt water intruded further inland. Although mangrove forest was cleared for built development, its coverage has consistently increased in the Swamp from 1957, with the exception of 2003 when there was a decrease by less than 100 ha. This is in contrast to most areas in the tropics where mangrove coverage continue to decline. In this case, the mangrove trees are out-competing/shading marsh vegetation, causing shift in the wetland communities. In the Caroni Ramsar Site, the natural wetland communities generally increased from 1942 to 2003, but declined in 2007, as built development more than doubled. The paper provides spatial coverage, and quantifies land cover from 1942 to 2007. It also identifies reasons for the changes in land cover and uses, and the implications for management.     

A regression tree-based method for integrating land-cover and land-use data collected at multiple scales

As data sets of multiple types and scales proliferate, it will be increasingly important to be able to flexibly combine them in ways that retain relevant information. A case in point is Amazonia, a large, data-poor region where most whole-basin data sets are limited to understanding land cover interpreted through a variety of remote sensing techniques and sensors. A growing body of work, however, indicates that the future state of much of Amazonia depends on the land use to which converted areas are put, but land use in the tropics is difficult to assess from remotely sensed data alone. An earlier paper developed new snapshots of agricultural land use in this region using a statistical fusion of satellite data and agricultural census data, an underutilized ancillary data source available across Amazonia. The creation of these land-use maps, which have the spatial detail of a satellite image and the attribute information of an agricultural census, required the development of a new statistical technique for merging data sets at different scales and of fundamentally different data types. Here we describe and assess this nonlinear technique, which reinterprets existing land cover classifications by determining what categories are most highly related to the polygon land-use data across the study area. Although developed for this region, the technique appears to hold broad promise for the systematic fusion of multiple data sets that are closely related but of different origins. The figures in the printed version of this article appear in black and white. Color figures are available from the author upon request.     

Satellite detection of bird communities in tropical countryside.

The future of biodiversity hinges partly on realizing the potentially high conservation value of human-dominated countryside. The characteristics of the countryside that promote biodiversity preservation remain poorly understood, however, particularly at the fine scales at which individual farmers tend to make land use decisions. To address this problem, we explored the use of a rapid remote sensing method for estimating bird community composition in tropical countryside, using a two-step process. First, we asked how fine-grained variation in land cover affected community composition. Second, we determined whether the observed changes in community composition correlated with three easily accessible remote sensing metrics (wetness, greenness, and brightness), derived from performing a tasseled-cap transformation on a Landsat Enhanced Thematic Mapper Plus image. As a comparison, we also examined whether the most commonly used remote sensing indicator in ecology, the Normalized Difference Vegetation Index (NDVI), correlated with community composition. We worked within an agricultural landscape in southern Costa Rica, where the land comprised a complex and highly heterogeneous mosaic of remnant native vegetation, pasture, coffee cultivation, and other crops. In this region, we selected 12 study sites (each < 60 ha) that encompassed the range of available land cover possibilities in the countryside. Within each site, we surveyed bird communities within all major land cover types, and we conducted detailed field mapping of land cover. We found that the number of forest-affiliated species increased with forest cover and decreased with residential area across sites. Conversely, the number of agriculture-affiliated species using forest increased with land area devoted to agricultural and residential uses. Interestingly, we found that the wetness and brightness metrics predicted the number of forest- and agriculture-affiliated species within a site as well as did detailed field-generated maps of land cover. In contrast, NDVI and the closely correlated greenness metric did not correlate with land cover or with bird communities. Our study shows the strong potential of the tasseled-cap transformation as a tool for assessing the conservation value of countryside for biodiversity.     

Predation risk and the territory value of cover: an experimental study

In an experiment on territorial brown trout (Salmo trutta), we addressed the novel hypothesis that protective cover increases the value of a territory in relation to the perceived level of predation risk. We predicted that territory holders should invest more resources defending territories with cover than territories without cover, and that defence should increase as predation risk increases. First, trout were allowed to establish ownership in territories with or without overhead cover. Second, predation risk was manipulated by simulating aerial predator attacks in half of the territories of each type, whereupon the preference for cover was estimated. Third, owners of the four types of territories were staged in dyadic contests against size-matched intruders. Territory owners showed a preference for cover, which increased further after simulated predator attacks. In subsequent contests, conflicts over territories with cover were settled faster than conflicts over territories without cover, which may suggest that the value of cover increases the motivational asymmetry between owner and intruder. Consistent with our hypothesis, owners of territories with cover were much more aggressive if they had been subjected to predator attacks the day before the conflict. These results suggest that territory owners are able to estimate the value of protective cover in response to variation in the level of predation risk in the habitat.Communicated by J. Krause     

A distance dependent contagion function for vector-based data

Landscape pattern is of primary interest to landscape ecologists and landscape metrics are used to quantify landscape pattern. Metrics are commonly defined and calculated on raster-based land cover maps. One metric is the contagion, existing in several versions, e.g., unconditional and conditional, used as a measure of fragmentation. However, mapped data is sometimes in vector-based format or there may be no mapped data but only a point sample. In this study a definition of contagion for such cases is investigated. The metric is an extension of the usual contagion, based on pairs of points at varying distances and gives a function of the distance. In this study the extended contagion is calculated for vector-based delineated real landscapes and for simulated ones. Both unconditional and conditional contagions are studied using two classification systems. The unconditional contagion function was decreasing and convex, with upper and lower limits highly correlated to the Shannon diversity index, thus carrying only area proportion information. The spatial information lies in the speed by which the function converges to the lower limit; using a proxy function this can be expressed by a single parameter b, with high values for fragmented landscapes. No proxy function was found for the conditional contagion, for which only qualitative information was found. The extended contagion is applicable both in patch mosaic models of landscapes and in gradient-based models, where landscape characteristics change continuously without distinct borders between patches. The extended contagion can be useful in sample based surveys where there no map of the entire landscape is available.     

Optimizing model for land use/land cover retrieval from remote sensing imagery based on variable precision rough sets

The suitable spectral mode in remote sensing is often desirable to facilitate the inversion of ecological environment and landscape. This paper put forward an optimizing model based on variable precision rough sets (VPRS) for the land cover discrimination in wetland inventory. In the case study of Lake Baiyangdian which has important ecological functions to the northern China, this model is established successfully according to the domain-experts knowledge. The procedure is as follows. First step is data collection, including remote-sensing data (e.g., Landsat-5 TM bands), the digitized relief maps, and statistical yearbooks. Second, the remote sensing imagery (RSI) and relief maps are co-registered into the same resolution. Third, a condition set, including various attributes is derived from spectral bands, band math or ratio indices based on previous studies, at the same time, the decision set is derived from true land types after investigation and validation. Then, the remote sensing decision table (RSDT) is constructed by linking condition set with decision set according to the sequential pixels in RSI. Fourth, we create one forward greedy searching algorithm based on VPRS to handle this RSDT. After adjusting parameters such as β and knowledge granularity diameter (KGD), we obtain the stable optimized results. Comparative experiments and evaluation show that the discrimination or retrieval accuracy of VPRS model is satisfying (overall accuracy: 87.32% and KHAT: 0.84) and better than original data. Moreover, data dimension has been decreased dramatically (from 12 to 3) and key attributes found by the model may be useful for specific retrieval in wetland inventories.