An evaluation of sampling strategies to improve precision of estimates of gross change in land use and land cover

Statistical sampling offers a cost-effective, practical alternative to complete-coverage mapping for the objective of estimating gross change in land cover over large areas. Because land cover change is typically rare, the sampling strategy must take advantage of design and analysis tools that enhance precision. Using two populations of land cover change in the eastern United States, we demonstrate that the choice of sampling unit size and use of a survey sampling regression estimator can significantly improve precision with only a minor increase in cost.     

Quantifying urban land use and impact on soils in Egypt using diurnal satellite imagery of the Earth surface

Conversion of agricultural land to urban use represents a potential loss of agricultural productivity, especially in areas where arable land is in short supply. Using derived products from both daytime (Landsat sensor data) and night-time imaging systems (U.S. Air Force Defense Meteorological Satellite Program's Operational Linescan System (DMSP/OLS)) we examined the impacts of urbanization on soils in Egypt; a country with very limited agricultural land. We concluded that urban land cover types to occupy 3.7% of the total area of Egypt and that over 30% of the soils most suitable for agriculture are under urban land cover. Analysis of multiyear historical DMSP/OLS data sets (digitized from paper images) proved unreliable for long-term urban growth estimates.     

The stability of global estimates from confusion matrices

Abstract

This Letter responds to a recent Letter by Hay (1988) which describes a method for correcting interpreted land cover areas based on field survey in the form of a confusion matrix. It is shown how Hay's and an alternative direct method are both stable and sensitive only when there is high contingency between all interpreted and surveyed classes.     

Land cover inventory in the Netherlands using remote sensing; application in a soil and groundwater vulnerability assessment system

Abstract

Information on actual land cover is necessary for various applications, such as soil and groundwater protection studies and hydrological studies. Therefore, a decision to produce a national land cover data base of the Netherlands, using satellite images, was made. The first version of the data base is now available for the whole of the Netherlands. Prior to the supervised classification the area was stratified in more or less homogeneous areas. Because cost, time and logistics required for a random sampling of the entire country were prohibitive, a mixed quantitative/qualitative classification accuracy assessment procedure was proposed. Classification performances were quantitatively assessed by comparing the classification results with digitized ground reference maps using a Geographical Information System (GIS). This offers a flexible method for locating the incorrectly labelled pixels and determining the possible reasons thereof. Classification accuracy of the land cover types which do not change much in time was also qualitatively assessed, using aerial photographs and topographical maps. The land cover data derived from remote sensing images can be readily combined with other digitized geographical data bases (e.g. soil maps).

The results of the proposed classification and validation procedure are presented for a test site situated in a stratum in the south of the Netherlands. It is shown how the land cover data are applied in a soil and groundwater vulnerability assessment system.     

Predictive models for remotely-sensed data acquisition in Indonesia

Abstract

Indonesian spatio-temporal cloud cover distribution was quantified with the aid of Geostationary Meteorological Satellite (GMS) and Landsat data. For all land areas iterative interactive factorial analyses grouped GMS-derived pixels with similar cloud cover profiles into 18 classes. Statistics of Landsat and SPOT images, grouped by class, were used to quantify temporal profiles of probability of acquiring remotely-sensed data with 10 per cent, 20 per cent and 30 per cent cloud cover for any Indonesian land area. Analysis of the patio-temporal characteristics of local climatic conditions permitted one to explain these profiles and to verify the validity of their seasonal variations for long periods. These profiles were fitted with a seventh-order polynomial for use in computer simulation of predictive models of remotely-sensed data acquisition.     

Deriving terrain and textural information from stereo RADARSAT data for mountainous land cover mapping

A new procedure is proposed for land cover classification in a mountainous area using stereo RADARSAT-1 data. The method integrates a few types of information that can be extracted from the same stereo RADARSAT images: (1) the Digital Elevation Model (DEM) generated from the stereo RADARSAT images; (2) terrain information (elevation, slope and aspect) extracted from the derived DEM; and (3) textural information derived from the same RADARSAT images. An Artificial Neural Network (ANN) classifier is applied for the land cover classification. Performance of the proposed method is evaluated using a mountainous study area in Southern Argentina, where there is a lack of up-to-date information for environmental monitoring. The results show that the integration of textural and terrain information can greatly improve the accuracy of the classification using the ANN classifier. It demonstrates that stereo RADARSAT images provide valuable data sources for land cover mapping, especially in mountainous areas where cloud cover is a problem for optical data collection and topographical data are not always available.     

Seasat SAR identification of dry climate urban land cover †

Abstract

Digitally processed Seasat SAR imagery of the Denver Colorado area was examined to assess its potential for mapping urban land cover and the compatibility of SAR derived classes with those described in the U.S. Geological Survey classification system. The entire scene was interpreted to generate a small-scale land cover map. In addition, six subscene enlargements representative of urban land cover categories extant in the area were used as test sites for detailed analysis of land cover types. Two distinct approaches were employed and compared in examining the imagery—a visual interpretation of black-and-white positive transparencies and an automated-machine/visual interpretation. The latter used the Image 100 interactive image analysis system to generate land cover classes by density level slicing of the image frequency histogram.     

Use of cartographic data and Landsat TM images to determine land use change in the vicinity of Mexico City

Evaluation of change in land use is important for planning further development in populated areas. Here we attempt to determine the growth of urban areas in the vicinity of Mexico City, using a 1993 Landsat Thematic Mapper (TM) image and cartographic data contained in maps published by the Instituto Nacional de Estadistica Geografia e Informatica (INEGI 1975, 1983). The area occupied by urban areas in 1975 and 1983 was quantified using raster images generated by scanning the maps. Supervised classification processes were applied to a 1993 Landsat TM image in bands 1, 2, 3, 4, 5 and 7, of the area of Chalco. The image was pre-processed and then processed to enhance the spectral response of the surface materials. The different land cover types that characterise distinct land uses in the study area were identified in the image and an overall classification accuracy of 82% was estimated using aerial photographs from the Chalco area. The resulting evaluation of the land use changes in the Chalco urban area was plotted, and a growth greater than 14% per year was estimated.     

Moorland soil erosion and spectral reflectance

Abstract

The dramatic increase in erosion on Glaisdale Moor in the North York Moors is mapped using aerial photographs for 1973, 1978 and 1983. The erosion provides a focus for a quantitative analysis of the relationships between spectral reflectance and ground variables in areas of peat, bare soil and vegetation.     

Advances in classification for land cover mapping using SPOT HRV imagery

Abstract

High-resolution data from the HRV (High Resolution Visible) sensors onboard the SPOT-1 satellite have been utilized for mapping semi-natural and agricultural land cover using automated digital image classification algorithms. Two methods for improving classification performance are discussed. The first technique involves the use of digital terrain information to reduce the effects of topography on spectral information while the second technique involves the classification of land-cover types using training data derived from spectral feature space. Test areas in Snowdonia and the Somerset Levels were used to evaluate the methodology and promising results were achieved. However, the low classification accuracies obtained suggest that spectral classification alone is not a suitable tool to use in the mapping of semi-natural cover types.     

Cloud cover distribution in Indonesia

Abstract

Indonesian spatio-temporal cloud cover distribution was quantified to allow planners to forecast probabilities for remote sensing data acquisition. The original data consisted of four randomly chosen Geostationary Meteorological Satellite displays per month from 1981 to 1985, which were analysed with a microcomputer after a 1 °15'pixel size digitization. Iterative interactive factorial analyses combined with a parallelepiped classifier clustered data by grouping pixels with similar profiles. For all classes, pixel profiles were linearly combined to provide a small number of classes with high pixel-class correlations. A 0–7 correlation threshold led to 18 classes for all land areas. Statistics of Landsat and SPOT images, grouped by class, were used to verify, calibrate and improve class profiles. This led to quantified temporal profiles of probability of acquiring remotely-sensed data with a given cloud cover percentage for any Indonesian land and marine area.     

Mapping and monitoring land cover in the Willandra Lakes World Heritage Region

Abstract

Landsat digital data is used to map land cover in the Willandra Lakes World Heritage Region in the semi-arid region of New South Wales. An unsupervised classification using a migrating means clustering algorithm followed by a maximum-likelihood classification procedure was found to be the most discriminating method of classifying land cover. The results of applying three change-detection routines to multitemporal data sets of selected sites within the region are presented and discussed.     

Multi-spectral classification of snow using NOAA AVHRR imagery

Abstract

Problems of accurate discrimination between snow and cloud, together with the detection of the snow pack boundary, have handicapped the use of satellite data in operational snow-cover mapping systems. A technique, involving an unsupervised clustering procedure, is described which allows the removal of cloud areas using NOAA-9 Advanced Very High Resolution Radiometer (AVHRR) channel-1, channel-3 and channel-4 data in conditions of recent snow lie and a difference channel (channel-2 —channel-1 with channel-3 and channel-4) during periods of advanced snow melt. Accurate delineation of snow extent is provided by the techniques if these specified snow conditions are taken into account. A method for the identification of areas of marginal snow melt is also presented, based on comparisons with Landsat Thematic Mapper data. The classifications also enable the determination of snow areas influenced by cloud shadows and conifer forest in addition to separating areas of differing snow depth and percentage cover.     

Production of land-use and land-cover maps of central Guangdong Province of China from LANDSAT MSS imagery†

Abstract

The problems of mapping the land use and land cover of a large area of central Guangdong Province of China from LANDSAT MSS data were examined with reference to the manual and digital approaches. Based on intensive field study of a test site in the study area, we discovered the importance of topographic effects, slopes, seasons, cropping system and the intensity of land use in affecting the accuracy of the resultant maps. It was concluded that visual interpretation was essential in providing the level of reference required for the image interpreter to perform the digital analysis satisfactorily. In view of the coarse spatial resolution of the LANDSAT MSS data it is recommended that the most straightforward digital analysis involving the use of the supervised approach with the minimum Euclidean distance classification and an iterative selection of training areas be adopted for the land-use/land-cover mapping of the whole study area to achieve an accuracy of 80 per cent for eight level I and fifteen level II categories of the land-use and land-cover classification scheme.     

Cover Airborne MSS data processing for forest classification

Abstract

Four SPOT HRV images of the same area of East Anglia, acquired between February and September 1986, have been evaluated at the National Remote Sensing Centre for their potential use in agricultural land cover mapping. Spectral coincidence plots were used in feature selection. Information from single images contained a high level of spectral confusion between cover types. Vegetation index images and original data were used in supervised maximum likelihood classification. Higher classification accuracies were achieved using the original data than the vegetation indices. An overall classification accuracy of 71 per cent for 10 land cover types was improved to 88 per cent by reducing the number of classes. Although the imagery acquired for the study did not correspond well to key dates in the crop calendar, the broad land cover categories, cereal crops, field crops (sugar beet and vegetables), grass land and broadleaved woodlands could be mapped from SPOT. Using vegetation indices from the whole scene, a map of land cover has been produced for an administrative district within the scene. Comparison with simulated Thematic Mapper data indicates greater crop discrimination is provided in the mid-infrared part of the spectrum.     

Super resolution land cover mapping of hyperspectral images using the deep image prior-based approach

ABSTRACT

Due to the instantaneous field-of-view (IFOV) of the sensor and diversity of land cover types, some pixels, usually named mixed pixels, contain more than one land cover type. Soft classification can predict the portion of each land cover type in mixed pixels in the absence of spatial distribution. The spatial distribution information in mixed pixels can be solved by super resolution mapping (SRM). Typically, SRM involves two steps: soft class value estimation, which is similar to the image super resolution of image restoration, and land cover allocation. A new SRM approach utilizes a deep image prior (DIP) strategy combined with a super resolution convolutional neural network (SRCNN) to estimate fine resolution fraction images for each land cover type; then, a simple and efficient classifier is used to allocate subpixel land cover types under the constraint of the generated fine fraction images. The proposed approach can use prior information of input images to update network parameters and no longer require training data. Experiments on three different cases demonstrate that the subpixel classification accuracy of the proposed DIP-based SRM approach is significantly better than the three conventional SRM approaches and a transfer learning-based neural network SRM approach. In addition, the DIP-SRM approach performs very robustly about small-area objects within multiple land cover types and significantly reduces soft classification uncertainty. The results of this paper provide an extension for utilizing SRCNN to address SRM issues in hyperspectral images.     

Interpreting image-based methods for estimating the signal-to-noise ratio

The signal-to-noise ratio (SNR) of remotely sensed imagery has been estimated directly using a variety of image-based methods such as the Homogeneous Area (HA) and Geostatistical (GS) methods. However, previous research has shown that such estimates may be dependent on land cover type. We examine this dependence on land cover type using Compact Airborne Spectrographic Imager (CASI) imagery of an agricultural region in Falmouth, Cornwall. The SNR was estimated using the GS method for six different land covers and a range of wavelengths. Large differences in the SNR existed between land cover types. It follows that single estimates of SNR (e.g. for one land cover) should not be associated with an image (as a whole). It is recommended that either (i) each statistic is reported per land cover type per wavelength or (ii) that an image of local statistics is reported per wavelength. The regression of noise on signal can be used to separate independent noise (intercept) from signal-dependent noise (slope). Variation in the noise and SNR estimates can be used to (i) allow more accurate prediction of the SNR and (ii) provide information on uncertainty.     

Urban built-up land change detection with road density and spectral information from multi-temporal Landsat TM data

In this article, Landsat TM images acquired during the same season from both 1984 and 1997 were analysed for urban built-up land change detection in Beijing, China, where great changes have taken place during the recent decades. To reduce the spectral confusion between urban 'built-up' and rural 'non built-up' land cover categories, we propose a new structural method based on road density combined with spectral bands for change detection. The road density represents one type of structural information while the multiple Landsat TM bands represent spectral information. Road density maps for both dates were produced using a gradient direction profile analysis (GDPA) algorithm and then integrated with spectral bands. Results from the spectral-structural postclassification comparison (SSPCC) and spectral-structural image differencing (SSID) methods were evaluated and compared with spectral-only change detection methods. The proposed SSPCC method greatly reduced spectral confusion and increased the accuracy of land cover classification compared with spectral classification, which in turn improved the change detection results. This article also shows that the SSID change detection result complemented spectral band differencing by detecting areas with greater structural changes, some of which were missed, by spectral band differencing.     

Characterization and classification of South American land cover types using satellite data

Abstract

Various methods are compared for carrying out land cover classifications of South America using multitemporal Advanced Very High Resolution Radiometer data. Fifty-two images of the normalized difference vegetation index (NDVI) from a 1-year period are used to generate multitemporal data sets. Three main approaches to land cover classification are considered, namely the use of the principal components transformed images, the use of a characteristic curves procedure based on N DVI values plotted against time, and finally application of the maximum likelihood rule to multitemporal data sets. Comparison of results from training sites indicates that the last approach yields the most accurate results. Despite the reliance on training site figures for performance assessment, the results are nevertheless extremely encouraging, with accuracies for several cover types exceeding 90 per cent.     

Comparison of multi-temporal NOAA-AVHRR and SPOT-XS satellite data for mapping land-cover dynamics in the west African Sahel

Abstract

Multi-resolution and multi-temporal remote sensing data (SPOT-XS and AVHRR) were evaluated for mapping local land cover dynamics in the Sahel of West Africa. The aim of this research was to evaluate the agricultural information that could be derived from both high and low spatial resolution data in areas where there is very often limited ground information. A combination of raster-based image processing and vector-based geographical information system mapping was found to be effective for understanding both spatial and spectral land-cover dynamics. The SPOT data proved useful for mapping local land-cover classes in a dominantly recessive agricultural region. The AVHRR-LAC data could be used to map the dynamics of riparian vegetation, but not the changes associated with recession agriculture. In areas where there was a complex mixture of recession and irrigated agriculture, as well as riparian vegetation, the AVHRR data did not provide an accurate temporal assessment of vegetation dynamics.