Thematic accuracy of the NLCD 2001 land cover for the conterminous United States

The land-cover thematic accuracy of NLCD 2001 was assessed from a probability-sample of 15,000 pixels. Nationwide, NLCD 2001 overall Anderson Level II and Level I accuracies were 78.7% and 85.3%, respectively. By comparison, overall accuracies at Level II and Level I for the NLCD 1992 were 58% and 80%. Forest and cropland were two classes showing substantial improvements in accuracy in NLCD 2001 relative to NLCD 1992. NLCD 2001 forest and cropland user's accuracies were 87% and 82%, respectively, compared to 80% and 43% for NLCD 1992. Accuracy results are reported for 10 geographic regions of the United States, with regional overall accuracies ranging from 68% to 86% for Level II and from 79% to 91% at Level I. Geographic variation in class-specific accuracy was strongly associated with the phenomenon that regionally more abundant land-cover classes had higher accuracy. Accuracy estimates based on several definitions of agreement are reported to provide an indication of the potential impact of reference data error on accuracy. Drawing on our experience from two NLCD national accuracy assessments, we discuss the use of designs incorporating auxiliary data to more seamlessly quantify reference data quality as a means to further advance thematic map accuracy assessment.     

Thematic accuracy of the 1992 National Land-Cover Data for the eastern United States: Statistical methodology and regional results

The accuracy of the 1992 National Land-Cover Data (NLCD) map is assessed via a probability sampling design incorporating three levels of stratification and two stages of selection. Agreement between the map and reference land-cover labels is defined as a match between the primary or alternate reference label determined for a sample pixel and a mode class of the mapped 3×3 block of pixels centered on the sample pixel. Results are reported for each of the four regions comprising the eastern United States for both Anderson Level I and II classifications. Overall accuracies for Levels I and II are 80% and 46% for New England, 82% and 62% for New York/New Jersey (NY/NJ), 70% and 43% for the Mid-Atlantic, and 83% and 66% for the Southeast.     

Updating the 2001 National Land Cover Database land cover classification to 2006 by using Landsat imagery change detection methods

The recent release of the U.S. Geological Survey (USGS) National Land Cover Database (NLCD) 2001, which represents the nation's land cover status based on a nominal date of 2001, is widely used as a baseline for national land cover conditions. To enable the updating of this land cover information in a consistent and continuous manner, a prototype method was developed to update land cover by an individual Landsat path and row. This method updates NLCD 2001 to a nominal date of 2006 by using both Landsat imagery and data from NLCD 2001 as the baseline. Pairs of Landsat scenes in the same season in 2001 and 2006 were acquired according to satellite paths and rows and normalized to allow calculation of change vectors between the two dates. Conservative thresholds based on Anderson Level I land cover classes were used to segregate the change vectors and determine areas of change and no-change. Once change areas had been identified, land cover classifications at the full NLCD resolution for 2006 areas of change were completed by sampling from NLCD 2001 in unchanged areas. Methods were developed and tested across five Landsat path/row study sites that contain several metropolitan areas including Seattle, Washington; San Diego, California; Sioux Falls, South Dakota; Jackson, Mississippi; and Manchester, New Hampshire. Results from the five study areas show that the vast majority of land cover change was captured and updated with overall land cover classification accuracies of 78.32%, 87.5%, 88.57%, 78.36%, and 83.33% for these areas. The method optimizes mapping efficiency and has the potential to provide users a flexible method to generate updated land cover at national and regional scales by using NLCD 2001 as the baseline.     

Estimating area and map accuracy for stratified random sampling when the strata are different from the map classes

The results of an accuracy assessment are typically organized using an error matrix that displays the proportion of area correctly mapped for each class and the proportion of area misclassified. Stratified random sampling is commonly implemented to obtain the reference data used to estimate the error matrix. When the strata correspond exactly to the map classes, the formulas for estimating accuracy and area are well known. Nevertheless, applications arise in which the strata are different from the map classes, as for example when the stratification is based on the map class labels of one map but the sample is subsequently used to assess the accuracy of other maps. In this paper, the estimators required when the stratum label and map label do not match for all pixels are presented for the proportion of area of each class based on the reference classification and for overall, user’s, and producer’s accuracies. Standard error formulas are also presented. A numerical example is provided to illustrate the computations.     

Updating the 2001 National Land Cover Database Impervious Surface Products to 2006 using Landsat Imagery Change Detection Methods

A prototype method was developed to update the U.S. Geological Survey (USGS) National Land Cover Database (NLCD) 2001 to a nominal date of 2006. NLCD 2001 is widely used as a baseline for national land cover and impervious cover conditions. To enable the updating of this database in an optimal manner, methods are designed to be accomplished by individual Landsat scene. Using conservative change thresholds based on land cover classes, areas of change and no-change were segregated from change vectors calculated from normalized Landsat scenes from 2001 and 2006. By sampling from NLCD 2001 impervious surface in unchanged areas, impervious surface predictions were estimated for changed areas within an urban extent defined by a companion land cover classification. Methods were developed and tested for national application across six study sites containing a variety of urban impervious surface. Results show the vast majority of impervious surface change associated with urban development was captured, with overall RMSE from 6.86 to 13.12% for these areas. Changes of urban development density were also evaluated by characterizing the categories of change by percentile for impervious surface. This prototype method provides a relatively low cost, flexible approach to generate updated impervious surface using NLCD 2001 as the baseline.     

Spatial agreement between two land‐cover data sets stratified by agricultural eco‐regions

Two of the most widely used land‐cover data sets for the United States are the National Land‐Cover Data (NLCD) at 30‐m resolution and the Global Land‐Cover Characteristics (GLCC) at 1‐km nominal resolution. Both data sets were produced around 1992 and expected to provide similar land‐cover information. This study investigated the spatial distribution of NLCD within major GLCC classes at 1‐km unit over a total of 11 agricultural‐related eco‐regions across the continental United States. Our results exhibited that data agreement or relationship between the GLCC and NLCD was higher for the eco‐regions located in the corn belt plains with homogeneous or less complicated land‐cover distributions. The GLCC cropland primarily corresponded to NLCD row crops, pasture/hay and small grains, and was occasionally related to NLCD forest, grassland and shrubland in the remaining eco‐regions due to high land‐cover diversity. The unique GLCC classes of woody savanna and savanna were mainly related to the NLCD orchard and grassland, respectively, in the eco‐region located in the Central Valley of California. The GLCC urban/built‐up among vegetated areas strongly agreed to the NLCD urban for the eco‐regions in the corn belt plains. A set of sub‐class land‐cover information provided through this study is valuable to understand the degrees of spatial similarity for the major global vegetated classes. The sub‐class information from this study provides reference for substituting less‐detailed global data sets for detailed NLCD to support national environment studies.     

Sources of error in accuracy assessment of thematic land-cover maps in the Brazilian Amazon

Valid measures of map accuracy are critical, yet can be inaccurate even when following well-established procedures. Accuracy assessment is particularly problematic when thematic classes lie along a land-cover continuum, and boundaries between classes are ambiguous. In this study, we examined error sources introduced during accuracy assessment of a regional land-cover map generated from Landsat Thematic Mapper (TM) data in Rondônia, southwestern Brazil. In this dynamic, highly fragmented landscape, the dominant land-cover classes represent a continuum from pasture to second growth to primary forest. We used high spatial resolution, geocoded videography as a reference, and focused on second-growth forest because of its potential contribution to the regional carbon balance. To quantify subjectivity in reference data labeling, we compared reference data produced by five trained interpreters. We also quantified the impact of other error sources, including geolocation errors between the map and reference data, land-cover changes between dates of data collection, heterogeneous reference samples, and edge pixels.Interpreters disagreed on classification of almost 30% of the samples; mixed reference samples and samples located in transitional classes accounted for a majority of disagreements. Agreement on second-growth forest labels between any two interpreters averaged below 50%, while agreement on primary forest was over 90%. Greater than 30% of disagreement between map and reference data was attributed to geolocation error, and 2.4% of disagreement was attributed to change in land cover between dates. After geocorrection, 24% of remaining disagreements corresponded to reference samples with mixed land cover, and 47% corresponded to edge pixels on the classified map. These findings suggest that: (1) labels of continuous land-cover types are more subjective and variable than commonly assumed, especially for transitional classes; however, using multiple interpreters to produce the reference data classification increases reference data accuracy; and (2) validation data sets that include only non-mixed, non-edge samples are likely to result in overly optimistic accuracy estimates, not representative of the map as a whole. These results suggest that different regional estimates of second-growth extent may be inaccurate and difficult to compare.     

Sub-pixel land cover mapping for per-field classification

A method was developed to transform a soft land cover classification into hard land cover classes at the sub-pixel scale for subsequent per-field classification. First, image pixels were segmented using vector boundaries. Second, the pixel segments (ranked by area) were labelled with a land cover class (ranked by class typicality). Third, a hard per-field classification was generated by examining each polygon (representing a land cover parcel, or field) in its entirety (by grouping the fragments of the polygon contained within different image pixels) and assigning to it the modal land cover class. The accuracy of this technique was considerably higher than that of both a corresponding hard per-pixel classification and a perfield classification based on hard per-pixel classified imagery.     

Effects of landscape characteristics on land-cover class accuracy

The effects of patch size and land-cover heterogeneity on classification accuracy were evaluated using reference data collected for the National Land-Cover Data (NLCD) set accuracy assessment. Logistic regression models quantified the relationship between classification accuracy and these landscape variables for each land-cover class at both the Anderson Levels I and II classification schemes employed in the NLCD. The general relationships were consistent, with the odds of correctly classifying a pixel increasing as patch size increased and decreasing as heterogeneity increased. Specific characteristics of these relationships, however, showed considerable diversity among the various classes. Odds ratios are reported to document these relationships. Interaction between the two landscape variables was not a significant influence on classification accuracy, indicating that the effect of heterogeneity was not impacted by the sample being in a small or large patch. Landscape variables remained significant predictors of class-specific accuracy even when adjusted for regional differences in the mapping and assessment processes or landscape characteristics. The land-cover class-specific analyses provide insight into sources of classification error and a capacity for predicting error based on a pixel's mapped land-cover class, patch size and surrounding land-cover heterogeneity.     

RadarSat-2 全极化数据土地覆盖分类精度评价

RadarSat-2 全极化数据能否应用于土地覆盖分类需要大量的研究和论证。本文以NLCD 为参考数据,利用 SVM 分类器对1 景旧金山地区RadarSat-2 全极化数据进行土地覆盖分类实验,并从分类类别面积一致性、空间相似性两个方 面对分类结果进行分类精度评价。实验获得RadarSat-2 数据分类结果总体分类精度76.91%和kappa 系数0.65,表明RadarSat- 2 全极化数据用于土地覆盖分类分类精度较高,可以达到很好的分类质量。     

A cartographic and database approach for land cover/use mapping and generalization from remotely sensed data

Classification of remotely sensed data involves a set of generalization processes, i.e. reality is simplified to a set of a few classes that are relevant to the application under consideration. This article introduces an approach to image classification that uses a class hierarchy structure for mapping unit definition at different generalization levels. This structure is implemented as an operational relational database and allows querying of more detailed land cover/use information from a higher abstraction level, which is that viewed by the map user. Elementary mapping units are defined on the basis of an unsupervised classification process in order to determine the land cover/use classes registered in the remotely sensed data. Mapping unit composition at different generalization levels is defined on the basis of membership values of sampled pixels to land cover/use classes. Unlike fuzzy classifications, membership values are presented to the user at mapping unit level.     

Approaches for the production and evaluation of fuzzy land cover classifications from remotely-sensed data

Remote sensing is an attractive source of data for land cover mapping applications. Mapping is generally achieved through the application of a conventional statistical classification, which allocates each image pixel to a land cover class. Such approaches are inappropriate for mixed pixels, which contain two or more land cover classes, and a fuzzy classification approach is required. When pixels may have multiple and partial class membership measures of the strength of class membership may be output and, if strongly related to the land cover composition, mapped to represent such fuzzy land cover. This type of representation can be derived by softening the output of a conventional ‘hard’ classification or using a fuzzy classification. The accuracy of the representation provided by a fuzzy classification is, however, difficult to evaluate. Conventional measures of classification accuracy cannot be used as they are appropriate only for ‘hard’ classifications. The accuracy of a classification may, however, be indicated by the way in which the strength of class membership is partitioned between the classes and how closely this represents the partitioning of class membership on the ground. In this paper two measures of the closeness of the land cover representation derived from a classification to that on the ground were used to evaluate a set of fuzzy classifications. The latter were based on measures of the strength of class membership output from classifications by a discriminant analysis, artificial neural network and fuzzy c-means classifiers. The results show the importance of recognising and accommodating for the fuzziness of the land cover on the ground. The accuracy assessment methods used were applicable to pure and mixed pixels and enabled the identification of the most accurate land cover representation derived. The results showed that the fuzzy representations were more accurate than the ‘hard’ classifications. Moreover, the outputs derived from the artificial neural network and the fuzzy c-means algorithm in particular were strongly related to the land cover on the ground and provided the most accurate land cover representations. The ability to appropriately represent fuzzy land cover and evaluate the accuracy of the representation should facilitate the use of remote sensing as a source of land cover data.     

Validation of a large area land cover product using purpose-acquired airborne video

Large area land cover products generated from remotely sensed data are difficult to validate in a timely and cost effective manner. As a result, pre-existing data are often used for validation. Temporal, spatial, and attribute differences between the land cover product and pre-existing validation data can result in inconclusive depictions of map accuracy. This approach may therefore misrepresent the true accuracy of the land cover product, as well as the accuracy of the validation data, which is not assumed to be without error. Hence, purpose-acquired validation data is preferred; however, logistical constraints often preclude its use — especially for large area land cover products. Airborne digital video provides a cost-effective tool for collecting purpose-acquired validation data over large areas. An operational trial was conducted, involving the collection of airborne video for the validation of a 31,000 km² sub-sample of the Canadian large area Earth Observation for Sustainable Development of Forests (EOSD) land cover map (Vancouver Island, British Columbia, Canada). In this trial, one form of agreement between the EOSD product and the airborne video data was defined as a match between the mode land cover class of a 3 by 3 pixel neighbourhood surrounding the sample pixel and the primary or secondary choice of land cover for the interpreted video. This scenario produced the highest level of overall accuracy at 77% for level 4 of classification hierarchy (13 classes). The coniferous treed class, which represented 71% of Vancouver Island, had an estimated user's accuracy of 86%. Purpose acquired video was found to be a useful and cost-effective data source for validation of the EOSD land cover product. The impact of using multiple interpreters was also tested and documented. Improvements to the sampling and response designs that emerged from this trial will benefit a full-scale accuracy assessment of the EOSD product and also provides insights for other regional and global land cover mapping programs.     

A scalable approach to mapping annual land cover at 250 m using MODIS time series data: A case study in the Dry Chaco ecoregion of South America

Land use and land cover (LULC) maps from remote sensing are vital for monitoring, understanding and predicting the effects of complex human-nature interactions that span local, regional and global scales. We present a method to map annual LULC at a regional spatial scale with source data and processing techniques that permit scaling to broader spatial and temporal scales, while maintaining a consistent classification scheme and accuracy. Using the Dry Chaco ecoregion in Argentina, Bolivia and Paraguay as a test site, we derived a suite of predictor variables from 2001 to 2007 from the MODIS 250 m vegetation index product (MOD13Q1). These variables included: annual statistics of red, near infrared, and enhanced vegetation index (EVI), phenological metrics derived from EVI time series data, and slope and elevation. For reference data, we visually interpreted percent cover of eight classes at locations with high-resolution QuickBird imagery in Google Earth. An adjustable majority cover threshold was used to assign samples to a dominant class. When compared to field data, we found this imagery to have georeferencing error < 5% the length of a MODIS pixel, while most class interpretation error was related to confusion between agriculture and herbaceous vegetation. We used the Random Forests classifier to identify the best sets of predictor variables and percent cover thresholds for discriminating our LULC classes. The best variable set included all predictor variables and a cover threshold of 80%. This optimal Random Forests was used to map LULC for each year between 2001 and 2007, followed by a per-pixel, 3-year temporal filter to remove disallowed LULC transitions. Our sequence of maps had an overall accuracy of 79.3%, producer accuracy from 51.4% (plantation) to 95.8% (woody vegetation), and user accuracy from 58.9% (herbaceous vegetation) to 100.0% (water). We attributed map class confusion to limited spectral information, sub-pixel spectral mixing, georeferencing error and human error in interpreting reference samples. We used our maps to assess woody vegetation change in the Dry Chaco from 2002 to 2006, which was characterized by rapid deforestation related to soybean and planted pasture expansion. This method can be easily applied to other regions or continents to produce spatially and temporally consistent information on annual LULC.     

A hybrid approach to urban land use/cover mapping using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) images

A hybrid method that incorporates the advantages of supervised and unsupervised approaches as well as hard and soft classifications was proposed for mapping the land use/cover of the Atlanta metropolitan area using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) data. The unsupervised ISODATA clustering method was initially used to segment the image into a large number of clusters of pixels. With reference to ground data based on 1?:?40?000 colour infrared aerial photographs in the form of Digital Orthophoto Quarter Quadrangle (DOQQ), homogeneous clusters were labelled. Clusters that could not be labelled because of mixed pixels were clipped out and subjected to a supervised fuzzy classification. A final land use/cover map was obtained by a union overlay of the two partial land use/cover maps. This map was evaluated by comparing with maps produced using unsupervised ISODATA clustering, supervised fuzzy and supervised maximum likelihood classification methods. It was found that the hybrid approach was slightly better than the unsupervised ISODATA clustering in land use/cover classification accuracy, most probably because of the supervised fuzzy classification, which effectively dealt with the mixed pixel problem in the low-density urban use category of land use/cover. It was suggested that this hybrid approach can be economically implemented in a standard image processing software package to produce land use/cover maps with higher accuracy from satellite images of moderate spatial resolution in a complex urban environment, where both discrete and continuous land cover elements occur side by side.     

Landsat urban mapping based on a combined spectral-spatial methodology

Urban mapping using Landsat Thematic Mapper (TM) imagery presents numerous challenges. These include spectral mixing of diverse land cover components within pixels, spectral confusion with other land cover features such as fallow agricultural fields and the fact that urban classes of interest are of the land use and not the land cover category. A new methodology to address these issues is proposed. This approach involves, as a first step, the generation of two independent but rudimentary land cover products, one spectral-based at the pixel level and the other segment-based. These classifications are then merged through a rule-based approach to generate a final product with enhanced land use classes and accuracy. A comprehensive evaluation of derived products of Ottawa, Calgary and cities in southwestern Ontario is presented based on conventional ground reference data as well as inter-classification consistency analyses. Producer accuracies of 78% and 73% have been achieved for urban ‘residential’ and ‘commercial/industrial’ classes, respectively. The capability of Landsat TM to detect low density residential areas is assessed based on dwelling and population data derived from aerial photography and the 2001 Canadian census. For low population densities (i.e. below 3000 persons/km²), density is observed to be monotonically related to the fraction of pixels labeled ‘residential’. At higher densities, the fraction of pixels labeled ‘residential’ remains constant due to Landsat's inability to distinguish between high-rise apartment dwellings and commercial/industrial structures.     

Mapping U.S. forest biomass using nationwide forest inventory data and moderate resolution information

A spatially explicit dataset of aboveground live forest biomass was made from ground measured inventory plots for the conterminous U.S., Alaska and Puerto Rico. The plot data are from the USDA Forest Service Forest Inventory and Analysis (FIA) program. To scale these plot data to maps, we developed models relating field-measured response variables to plot attributes serving as the predictor variables. The plot attributes came from intersecting plot coordinates with geospatial datasets. Consequently, these models serve as mapping models. The geospatial predictor variables included Moderate Resolution Imaging Spectrometer (MODIS)-derived image composites and percent tree cover; land cover proportions and other data from the National Land Cover Dataset (NLCD); topographic variables; monthly and annual climate parameters; and other ancillary variables. We segmented the mapping models for the U.S. into 65 ecologically similar mapping zones, plus Alaska and Puerto Rico. First, we developed a forest mask by modeling the forest vs. nonforest assignment of field plots as functions of the predictor layers using classification trees in See5©. Secondly, forest biomass models were built within the predicted forest areas using tree-based algorithms in Cubist©. To validate the models, we compared field-measured with model-predicted forest/nonforest classification and biomass from an independent test set, randomly selected from available plot data for each mapping zone. The estimated proportion of correctly classified pixels for the forest mask ranged from 0.79 in Puerto Rico to 0.94 in Alaska. For biomass, model correlation coefficients ranged from a high of 0.73 in the Pacific Northwest, to a low of 0.31 in the Southern region. There was a tendency in all regions for these models to over-predict areas of small biomass and under-predict areas of large biomass, not capturing the full range in variability. Map-based estimates of forest area and forest biomass compared well with traditional plot-based estimates for individual states and for four scales of spatial aggregation. Variable importance analyses revealed that MODIS-derived information could contribute more predictive power than other classes of information when used in isolation. However, the true contribution of each variable is confounded by high correlations. Consequently, excluding any one class of variables resulted in only small effects on overall map accuracy. An estimate of total C pools in live forest biomass of U.S. forests, derived from the nationwide biomass map, also compared well with previously published estimates.     

Using genetic algorithms in sub-pixel mapping

In remotely sensed images, mixed pixels will always be present. Soft classification defines the membership degree of these pixels for the different land cover classes. Sub-pixel mapping is a technique designed to use the information contained in these mixed pixels to obtain a sharpened image. Pixels are divided into sub-pixels, representing the land cover class fractions. Genetic algorithms combined with the assumption of spatial dependence assign a location to every sub-pixel. The algorithm was tested on synthetic and degraded real imagery. Obtained accuracy measures were higher compared with conventional hard classifications.     

The first detailed land‐cover map of Socotra Island by Landsat/ETM+ data

Present study has produced first detailed land‐cover map of Socotra Island. A Landsat 7 ETM+ dataset was used as a main source of remotely sensed data. From numerous reference points (more than 250) coming from the ground data verification the set of training fields and the set of evaluation fields were digitised. As a classification method the supervised maximum likelihood classification without prior probabilities was used in combination with rule‐based post‐classification sorting, providing results of sufficient accuracy and subject resolution. Estimates of the area and degree of coverage of particular land‐cover classes within Socotra Island have brought excellent overview on state of island biotopes. Overall accuracy of the map achieved is more than 80%, 19 terrestrial land‐cover classes (including three types of Shrublands, three types of Woodlands, two types of Forests and Mangroves) have been distinguished. It consequently allows estimates of the current and potential occurrence of endemic plant populations, proposals of management and conservation plans and agro‐forestry planning.