Multibaseline and tomographic synthetic aperture radar (SAR) data are often affected by phase distortions known as phase screens. These distortions stem either from atmospheric effects or residual errors in platform motion. Calibrating and compensating for the phase screen is crucial to prevent spreading and defocusing in multidimensional tomographic imaging. Given the growing interest in artificial intelligence and deep learning, we aim to utilize their potential to develop a phase calibration process for SAR tomographic data. Our proposed framework is based upon a convolutional neural network (CNN) and generates training patches directly from the tomographic images under consideration, without relying on external references or resources. Once trained, the network effectively estimates phase distortions across the entire image; these are then used to calibrate the tomographic data. Experimental results from AfriSAR and UAVSAR tomographic datasets are included to showcase the effectiveness of the proposed solution.

In the above article [1], the correct reference associated with equation (2) on page 3, left column, is the below paper: 1) P. Imperatore and G. Fornaro, “Joint Phase-Screen Estimation in Airborne Multibaseline SAR Tomography Data Processing,” IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1–14, 2024, Art. no. 4412614, doi: 10.1109/TGRS.2024.3446186.

Image segmentation represents a fundamental step in analyzing very high-spatial-resolution (VHR) remote sensing imagery. Its objective is to partition an image into segments that best match with geo-objects. However, the diverse appearances of geospatial objects often lead to interobject homogeneity and intraobject heterogeneity. Existing segmentation methods often struggle to accurately segment geo-objects with varying shapes and scales. To address these challenges, we propose DeepMerge, a novel method that integrates deep learning and region adjacency graphs (RAGs) to accurately segment complete geo-objects in large VHR images. DeepMerge begins with an initial over-segmentation of the image and then iteratively merges similar regions to achieve complete geo-object segmentation. A deep learning model is employed to learn the similarity between adjacent superpixel pairs. This approach only requires labels indicating whether adjacent superpixels belong to the same geo-object eliminating the need for object-level annotations, enabling weakly supervised segmentation. A cross-scale module is incorporated to capture multiscale information, enhancing the representation of superpixels. In addition, the feature distances between neighboring super-pixels are deemed as scale parameters (thresholds) to control the merging procedure, thus yielding an interpretable, predictable, stable, and optimal scale parameter 0.5. DeepMerge can achieve high segmentation accuracy in a weakly supervised manner, which is validated on large-scale remote sensing images of 0.55-m resolution covering an area of 5660 km2. The experimental results demonstrate that DeepMerge achieves the highest F value (0.9552) and the lowest total error (TE) (0.0827), accurately segmenting geo-objects of varying sizes and outperforming all competing methods.

Polarimetric Synthetic Aperture Radar (PolSAR) images offer a rich repository of information, crucial for diverse applications ranging from classification to target identification. In the domain of PolSAR image classification, the Wishart classifier emerged as a prominent and widely employed technique. This classifier is often used to articulate the properties of polarimetric scattering types in images, providing valuable insight into various types of targets. With the growing interest in multidisciplinary Artificial Intelligence (AI) research, especially in computer vision and Natural Language Processing (NLP), our goal is to integrate this enthusiasm into polarimetric image analysis. We propose extending the Wishart classifier framework to include a free-form and open-ended Visual Question Answering (VQA) model. This model is designed to answer natural language questions related to PolSAR images, covering pixel details and scattering patterns. The objective is to provide accurate natural language responses that reflect real-world scenarios, such as assisting the visually impaired. Both questions and answers in this context are intentionally left open-ended to capture the complexity of inquiries in the polarimetric SAR images domain.

Clouds in optical images are inevitable and can adversely affect subsequent analysis. Given that SAR imagery remains unaffected by cloud cover, many cloud removal methods involve the fusion of SAR and optical imagery. Unfortunately, existing methods for cloud removal through SAR-optical image fusion are computationally intensive and time-consuming, limiting their practical application. To address these challenges, this paper proposes a novel multi-feature fusion network (MFFNet) for SAR-optical image fusion, aiming to remove clouds from optical images effectively. The proposed method was applied to global and all-season Sentinel-1 and Sentinel-2 images. Quantitative experiments demonstrate that MFFNet achieves high accuracy and efficiency. Specifically, our method obtains an SSIM value of 87.10 and a speed of 25.97 FPS.

Selective logging represents a primary driver of forest degradation, being early germs of the deforestation process. It negatively impacts the remaining forests, leading to biodiversity losses, and catalyzing in the long term the climate changes. It is a process related to spatial-temporal changes in the forest areas, and consequently can be monitored by means of satellite images. In this work, we evaluated the potential of image time series of both Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical data to detect selective logging using several variations of the Long ShortTerm Memory (LSTM) network. Particularly, we exploited complex-valued SAR images that contain both intensity and phase information to capture small changes caused by selective logging. The experiments using optical data achieved the highest accuracy of 98.21%, while those using SAR data reached a maximum accuracy of 72.68%. The results demonstrated the effectiveness of optical images and the potential of complex-valued SAR data for selective logging detection.

Field-Level cereal yield estimation using Machine Learning (ML) models poses a significant challenge especially when applied across large areas. A large sample size is required to represent the high yield variability caused by varying topographic and climatic conditions. To enhance ML-based prediction accuracy, we propose to decompose the complexity of agricultural landscape using landforms and agro-ecological zones and use these classes as spatially explicit constraints to partition field samples. We trained three ML models using remote sensing data to estimate wheat yield. When training ML models without the mentioned spatial constraints, we achieved an R2=0.58 and RMSE=840kg/ha. Training ML separately across various landform classes increase the accuracy. For instance, wheat yield cultivated in plain areas was predicted with R2=0.72, and RMSE=809kg/ha. These results emphasized the potential of training ML separately across main landform classes for improving the accuracy of yield predictions across diverse geographical contexts.

Glaciers are one of the essential climate variables. Tracking their areal changes over time is of high importance for monitoring the impacts of climate change and designing adaptation strategies. Mapping glaciers from optical remote sensing data might result in a very limited temporal resolution due to the absence of cloud-free imagery at the end of the ablation season. Synthetic aperture radar (SAR) solves this problem as it can operate in almost all weather conditions. Here, we present a deep learning strategy for glacier mapping based solely on Sentinel-1 SAR data in Svalbard. We test two options for integrating SAR image time series into deep learning models, namely, 3D convolutions and long short-term memory (LSTM) cells. Both proposed models achieve an intersection over union (IoU) of 0.964 on the test subset. Our results highlight the applicability of SAR data in glacier mapping with the potential to obtain glacier inventories with higher temporal resolution. We shared our dataset, code-base and pretrained models at https://github.com/konstantin-a-maslov/icemapper.

The use of street view imagery (SVI) and advanced urban visual intelligence technologies has revolutionized built environment auditing (BEA) practice, by enabling high-resolution BEA at large scales. This study reviewed 96 studies of BEA published before October 2023. The Google SVI was employed in 92.7% of the included studies. Manual processing of SVI was used in BEA in most studies (81.3%), while deep learning methods were mostly used in the remaining studies. Validated auditing tools were used in 71% of the studies. Streets were the most frequently audited objects (54.2%), followed by sidewalk (51%), traffic (49%), and land use (34.4%). Objective attributes exhibited higher reliability in BEA, compared to subjective attributes (e.g. neighborhood environmental perception). The Active Neighborhood Checklist and Microscale Audit of Pedestrian Streetscapes were the two most widely used SVI-based BEA tools. Several key areas for improving the accuracy and reliability of SVI-based BEA were identified: building standardized datasets of built environment features for more accurate auditing, combining multi-source SVI for more comprehensive assessments, and adapting auditing tools to the contexts in developing countries. This study would contribute to a deeper understanding of built environmental influences on health, and facilitate informed decision-making in urban planning and public health efforts.

Clouds in remote sensing optical images often obscure essential information. They may lead to occlusion or distortion of ground features, thereby affecting the subsequent analysis and extraction of target information. Therefore, the removal of clouds in optical images is a critical task in various applications. SAR-optical image fusion has achieved encouraging performance in the reconstruction of cloud-covered information. Such methods, however, are extremely time-consuming and computationally intensive, making them difficult to apply in practice. This letter proposes a novel Feature Pyramid Network (FPNet) that effectively reconstructs the missing optical information. FPNet enables the extraction and fusion of multi-scale features from the SAR image and the cloudy optical image, as the FPNet leverages the power of convolutional neural networks by merging the feature maps from different scales. It can learn useful features efficiently because it downsamples the input images while preserving important information, thus reducing the computational workload. Experiments are conducted on a benchmark global SEN12MS-CR dataset and a regional South Sudan dataset. Results are compared with those of state-of-the-art methods such as DSen2-CR and GLF-CR. The experimental results demonstrate that FPNet accomplishes superior performance in terms of accuracy and visual effects. Both the inference and training speeds of FPNet are fast. Specifically, it runs at 96 FPS and requires less than four hours to train a single epoch using SEN12MS-CR on two 2080ti GPUs. Therefore, it is suitable for applying to various study areas.

Extracting training datasets for supervised classification of Synthetic Aperture Radar (SAR) images is complicated, due to e.g. poor radiometric resolution, speckle noise, and lack of reference data. It is challenging to link radar scatterers in SAR images with the counterparts in the reference datasets registered in geographic coordinate systems. To address this issue, this paper proposes a method called Rdr-Code to radarcode geodetic reference datasets for creating SAR training datasets for machine learning applications. To assess the importance of building heights in radarcoding, we compared the assignment of height values by a minuscule pseudo height with the actual building heights derived from a Lidar-based DEM product. We used 30 PAZ SAR images in X-band, which were acquired between 2019 and 2021, over the north-west part of the Netherlands, and employed Top10NL and AHN as reference LULC polygon and height datasets respectively. The radarcoding accuracy was compared using nine buildings as references in the SAR coordinates. The radarcoding accuracy was 64.5% with the pseudo height and 84.5% with actual building heights. A trade-off between accurate building feature information and separation between close buildings was observed. We conclude that this is an effective way to radarcode reference datasets and can be used for crafting training datasets for machine learning methods.

Glacier mapping is essential for studying and monitoring the impacts of climate change. However, several challenges such as debris-covered ice and highly variable landscapes across glacierized regions worldwide complicate large-scale glacier mapping in a fully-automated manner. This work presents a novel hybrid CNN-transformer model (GlaViTU) for multi-regional glacier mapping. Our model outperforms three baseline models—SETR-B/16, ResU-Net and TransU-Net—achieving a higher mean IoU of 0.875 and demonstrates better generalization ability. The proposed model is also parameter-efficient, with approximately 10 and 3 times fewer parameters than SETR-B/16 and ResU-Net, respectively. Our results provide a solid foundation for future studies on the application of deep learning methods for global glacier mapping. To facilitate reproducibility, we have shared our data set, codebase and pretrained models on GitHub at https://github.com/konstantin-a-maslov/GlaViTU-IGARSS2023.

Grasping the roof structure of a building is a key part of building reconstruction. Directly predicting the geometric structure of the roof from a raster image to a vectorized representation, however, remains challenging. This paper introduces an efficient and accurate parsing method based upon a vision Transformer we dubbed Roof-Former. Our method consists of three steps: 1) Image encoder and edge node initialization, 2) Image feature fusion with an enhanced segmentation refinement branch, and 3) Edge filtering and structural reasoning. The vertex and edge heat map F1-scores have increased by 2.0% and 1.9% on the VWB dataset when compared to HEAT. Additionally, qualitative evaluations suggest that our method is superior to the current state-of-the-art. It indicates effectiveness for extracting global image information and maintaining the consistency and topological validity of the roof structure.

Synthetic aperture radar (SAR) systems can be designed with different polarimetric modalities. Most spaceborne SAR systems acquire dual-polarimetric data to meet various operational requirements. They are designed to capture more information about the Earth’s surface than single-pol systems and to cover a wider area than full-pol modalities. Dual-polarimetric data may not be as informative as fully polarimetric images. Several methods exist to augment dual-polarimetric images to take the capabilities of fully polarimetric data. Such methods, nevertheless, are either specific to special dual-pol modalities, i.e., compact modes, or rely on model assumptions that may not be valid in various scattering scenarios. In this article, a new framework for reconstructing fully polarimetric information from typical modalities of dual-pol data is proposed. The framework uses deep learning solutions to augment dual-polarimetric data without relying on model assumptions. Besides the specific architecture of the network used, which makes it efficient to extract distinctive features, a specific loss function is defined to account for the different scattering properties of the polarimetric data. Experiments on different real data show that the reconstruction performance of the proposed framework is superior to the conventional reconstruction method that widely experimented in the literature. Moreover, the pseudo-fully polarimetric data reconstructed by the proposed method closely match the actual fully polarimetric images acquired by radar systems, confirming the reliability and effectiveness of the proposed method.

This study investigates the PAZ SAR data in both VV and HH channels, contributing to PAZ product assessment. It focuses on deformation maps and deformation time series over a test site in Leeuwarden, the Netherlands. To optimize the estimates of the deformation time series and parameters of the temporal model, we developed and applied a model-backfeed method (MBF). This MBF iteratively re-introduces into phase unwrapping the best deformation model of every Constantly Coherent Scatterer, such as Persistent Scatterer (PS), determined by Multiple Hypothesis Testing (MHT). In this regard, distinct temporal behavior of individual scatterers is considered and modeled, thus improving the estimates of deformation time series. 24 co-polarimetric SAR data acquired between 2019 and 2021 are used for our test. The test result shows that PAZ SAR in HH offered 7% more PS than those in VV, and the VV mode identified a bit more PS along line-infrastructure like roads. Besides, by comparison and GNSS-based validation, we find that co-pol PAZ products are generally of a good quality. The MBF method increases the average ensemble coherence by 38% for VV and 35% for HH, and decreases the average spatio-temporal consistency by 2.3% for VV and HH and mitigates phase unwrapping errors, thereby optimize deformation estimates.

OBJECTIVE: Antimicrobial resistance (AMR) is a global threat to health and healthcare. In response to the growing AMR burden, research funding also increased. However, a comprehensive overview of the research output, including conceptual, temporal, and geographical trends, is missing. Therefore, this study uses topic modelling, a machine learning approach, to reveal the scientific evolution of AMR research and its trends, and provides an interactive user interface for further analyses. METHODS: Structural topic modelling (STM) was applied on a text corpus resulting from a PubMed query comprising AMR articles (1999-2018). A topic network was established and topic trends were analysed by frequency, proportion, and importance over time and space. RESULTS: In total, 88 topics were identified in 158,616 articles from 166 countries. AMR publications increased by 450% between 1999 and 2018, emphasizing the vibrancy of the field. Prominent topics in 2018 were Strategies for emerging resistances and diseases, Nanoparticles, and Stewardship. Emerging topics included Water and environment, and Sequencing. Geographical trends showed prominence of Multidrug-resistant tuberculosis (MDR-TB) in the WHO African Region, corresponding with the MDR-TB burden. China and India were growing contributors in recent years, following the United States of America as overall lead contributor. CONCLUSION: This study provides a comprehensive overview of the AMR research output thereby revealing the AMR research response to the increased AMR burden. Both the results and the publicly available interactive database serve as a base to inform and optimise future research.

The goal of this work is to investigate the potential of PRecursore IperSpettrale della Missione Applicativa (PRISMA) hyperspectral data to predict the concentration of four macronutrients (K, P, N, S) and four micronutrients (Ca, Fe, Mg, Zn) in final wheat production. All investigated nutrients are essential to improving human nutrition. The initial findings indicate accurate predictions for Zn, P, Mg, S, K, Ca and Fe (R2 ranging from 0.57 to 0.74). N was less accurately estimated (R2 of 0.49). We conclude that the foliar chemical properties and temporal dynamics as detected by hyperspectral data translate successfully to the target micro- and macronutrients composition of the wheat production.

Information on cultivated crops is relevant for a large number of food security studies. Different scientific efforts are dedicated to generate this information from remote sensing images by means of machine learning methods. Unfortunately, these methods do not take account of the spatial-temporal relationships inherent in remote sensing images. In our paper, we explore the capability of a 3D Fully Convolutional Neural Network (FCN) to map crop types from multi-temporal images. In addition, we propose the Intersection Over Union (IOU) loss function for increasing the overlap between the predicted classes and ground reference data. The proposed method was applied to identify soybean and corn from a study area situated in the US corn belt using multi-temporal Landsat images. The study shows that our method outperforms related methods, obtaining a Kappa coefficient of 91.8%. We conclude that using the IOU loss function provides a superior choice to learn individual crop types.

Synergizing SAR multi-sensors facilitates long-term deformation time series monitoring of radar scatterers on the Earth surface. Due to the disparity in e.g. radar wavelength, incidence angle, orbital direction, and polarization, however, there is no straightforward way to concatenate such time series from different SAR sensors. This study as an extension of [1] proposes the use of tie-point pairs, i.e. scatterers that are most likely reflected from a common ground target, aiming at integrating multi-sensor SAR data to monitor surface deformation without the loss of spatial resolution. Tie-point pairs are identified using geolocation uncertainty of radar scatterers. A probabilistic temporal model of tie-point pairs' time series is developed to link deformation time series from different sensors. We tested the proposed approaches in Groningen, The Netherlands, using 82 Radarsat-2 (C-band, July 2009 - June 2015) and 13 ALOS-2 (L-band, September 2014 - May 2020). Finally we identified 3315 tie-points with three different intersection types and determined their best temporal models. For those points, the maximum vertical subsidence velocity is up to 10 mm yr-1 between 2009 and 2020.

Roof shape information is essential for creating 3D building models. However, the automated extracting of roof structures from Earth observation data is a difficult task involving significant uncertainties caused by scene complexity and limited multi-source data coverage. This paper introduces the integrally-attracted wireframe parsing (IAWP) framework to reconstruct building rooflines as a planar graph from remotely sensed images with a single forward pass. We add global geometric line priors through the Hough transform into deep networks to better extract the linear geometric features. We perform experiments on the vectorizing world building (VWB) dataset. The investigated method improves the F-score metrics of corner points/edges by 0.1%/7.7% and 0.6%/1.1%, respectively. Visual comparison results also indicate that the HT-IHT block gives consistent improvements in terms of geometric regularity.

Building extraction from remote sensing images using convolutional neural networks (CNNs) has been an active research topic in recent years. Most results obtained by CNN-based algorithms, however, still have common issues with the precision of the delineation of building outlines and the separation of different buildings. Recently, efforts have been made towards the automation of building outline regularization. This paper employs a new instance segmentation framework named Hybrid Task Cascade (HTC) as baseline model, integrating detection and segmentation as a joint multi-stage processing. We further integrate regularization methods such as convex hull and Douglas-Peucker algorithm to obtain accurately segmented edges. The method is tested on the crowdAI benchmark dataset by comparing with alternative state-of-the-art models (i.e., Mask R-CNN). The results show that our method achieves better instance segmentation results and improves the results in terms of geometric regularity of building segments.

Spatial information of land values is fundamental for planners and policy makers. Individual appraisals are costly, explaining the need for predictive modelling. Recent work has investigated using Space Syntax to analyse urban access and explain land values. However, the spatial dependence of urban land markets has not been addressed in such studies. Further, the selection of meaningful variables is commonly conducted under non-spatialized modelling conditions. The objective of this paper is to construct a land value map using a geostatistical approach using Space Syntax and a spatialized variable selection. The methodology is applied in Guatemala City. We used an existing dataset of residential land value appraisals and accessibility metrics. Regression-kriging was used to conduct variable selection and derive a model for spatial prediction. The prediction accuracy is compared with a multivariate regression. The results show that a spatialized variable selection yields a more parsimonious model with higher prediction accuracy. New insights were found on how Space Syntax explains land value variability when also modelling the spatial dependence. Space Syntax can contribute with relevant spatialized information for predictive land value modelling purposes. Finally, the spatial modelling framework facilitates the production of spatial information of land values that is relevant for planning practice.

Recently developed urban air quality sensor networks are used to monitor air pollutant concentrations at a fine spatial and temporal resolution. The measurements are however limited to point support. To obtain areal coverage in space and time, interpolation is required. A spatio-temporal regression kriging approach was applied to predict nitrogen dioxide (NO2) concentrations at unobserved space-time locations in the city of Eindhoven, the Netherlands. Prediction maps were created at 25 m spatial resolution and hourly temporal resolution. In regression kriging, the trend is separately modelled from autocorrelation in the residuals. The trend part of the model, consisting of a set of spatial and temporal covariates, was able to explain 49.2% of the spatio-temporal variability in NO2 concentrations in Eindhoven in November 2016. Spatio-temporal autocorrelation in the residuals was modelled by fitting a sum-metric spatio-temporal variogram model, adding smoothness to the prediction maps. The accuracy of the predictions was assessed using leave-one-out cross-validation, resulting in a Root Mean Square Error of 9.91 μg m−3, a Mean Error of −0.03 μg m−3 and a Mean Absolute Error of 7.29 μg m−3. The method allows for easy prediction and visualization of air pollutant concentrations and can be extended to a near real-time procedure.

Classification of very high resolution (VHR) satellite images faces two major challenges: 1) inherent low intra-class and high inter-class spectral similarities and 2) mismatching resolution of available bands. Conventional methods have addressed these challenges by adopting separate stages of image fusion and spatial feature extraction steps. These steps, however, are not jointly optimizing the classification task at hand. We propose a single-stage framework embedding these processing stages in a multiresolution convolutional network. The network, called FuseNet, aims to match the resolution of the panchromatic and multispectral bands in a VHR image using convolutional layers with corresponding downsampling and upsampling operations. We compared FuseNet against the use of separate processing steps for image fusion, such as pansharpening and resampling through interpolation. We also analyzed the sensitivity of the classification performance of FuseNet to a selected number of its hyperparameters. Results show that FuseNet surpasses conventional methods.

This letter proposes an associative hierarchical conditional random field (AHCRF) model to improve the classification accuracy of high-resolution remote sensing images. It considers segmentation quality of superpixels, avoids a time-consuming selection of optimal scale parameters, and alleviates the problem of classification accuracy sensitive to undersegmentation errors that is present in traditional object-oriented classification methods. The model is built on a graph hierarchy, including the pixel layer as a base layer and multiple superpixel layers derived from a mean shift presegmentation. It extracts clustered features of pixels for superpixels at each layer and then defines the potentials of the AHCRF model. We suggest a weighted version of the interlayer potential using the size of a superpixel as a measure to reflect segmentation quality. In this way, erroneously labeled pixels of a superpixel are penalized. Experiments are presented using a part of the downsampled Vaihingen data from the ISPRS benchmark data set. Results confirm that our model shows more than 80% overall classification accuracy and is superior to the original AHCRF model and comparable to other models. It also alleviates the choosing of suitable segmentation parameters.

Classification of very high-resolution (VHR) satellite images has three major challenges: 1) inherent low intraclass and high interclass spectral similarities; 2) mismatching resolution of available bands; and 3) the need to regularize noisy classification maps. Conventional methods have addressed these challenges by adopting separate stages of image fusion, feature extraction, and postclassification map regularization. These processing stages, however, are not jointly optimizing the classification task at hand. In this paper, we propose a single-stage framework embedding the processing stages in a recurrent multiresolution convolutional network trained in an end-to-end manner. The feedforward version of the network, called FuseNet, aims to match the resolution of the panchromatic and multispectral bands in a VHR image using convolutional layers with corresponding downsampling and upsampling operations. Contextual label information is incorporated into FuseNet by means of a recurrent version called ReuseNet. We compared FuseNet and ReuseNet against the use of separate processing steps for both image fusions, e.g., pansharpening and resampling through interpolation and map regularization such as conditional random fields. We carried out our experiments on a land-cover classification task using a Worldview-03 image of Quezon City, Philippines, and the International Society for Photogrammetry and Remote Sensing 2-D semantic labeling benchmark data set of Vaihingen, Germany. FuseNet and ReuseNet surpass the baseline approaches in both the quantitative and qualitative results.

Permanent scatterer interferometry is a multitemporal interferometric synthetic aperture radar technique that produces high-accuracy ground deformation measurement. A high density of permanent scatterer (PS) is required to provide accurate results. In natural environments with low PS density, distributed scatterers (DSs) could serve as additional coherent observations. This paper introduces a polarimetric scattering property-based adaptive filtering method that preserves PS candidates and filters DS candidates. To further increase the coherence estimate of DS candidates, the technique includes a complex coherence decomposition that adaptively selects the most stable scattering mechanisms, thus improving pixel coherence estimation. The proposed method was evaluated on 11 quad-polarized ALOS PALSAR images and 21 dual-polarized Sentinel-1 images acquired over San Fernando Valley, CA, USA, and Groningen, The Netherlands, respectively. The application of this method increased the number of coherent pixels by almost a factor of eight compared with a single-polarization channel. This paper concludes that a coherence estimate can be significantly improved by applying scattering property-based adaptive filtering and coherence matrix decomposition and accurate displacement measurements can be achieved.

Convolutional neural networks (CNNs), widely studied in the domain of computer vision, are more recently finding application in the analysis of high-resolution aerial and satellite imagery. In this paper, we investigate a deep feature learning approach based on CNNs for the detection of informal settlements in Dar es Salaam, Tanzania. This information is vital for decision making and planning of upgrading processes. Distinguishing the different urban structure types is challenging because of the abstract semantic definition of the classes as opposed to the separation of standard land-cover classes. This task requires the extraction of complex spatial-contextual features. To this aim, we trained a CNN in an end-to-end fashion and used it to classify informal and formal settlements. Our experimental results show that CNNs outperform state of the art methods using hand-crafted features. We conclude that CNNs are able to effectively learn the spatial-contextual features for accurately discriminating formal and informal settlements.

This letter investigates fully convolutional networks (FCNs) for the detection of informal settlements in very high resolution (VHR) satellite images. Informal settlements or slums are proliferating in developing countries and their detection and classification provides vital information for decision making and planning urban upgrading processes. Distinguishing different urban structures in VHR images is challenging because of the abstract semantic definition of the classes as opposed to the separation of standard land-cover classes. This task requires extraction of texture and spatial features. To this aim, we introduce deep FCNs to perform pixel-wise image labeling by automatically learning a higher level representation of the data. Deep FCNs can learn a hierarchy of features associated to increasing levels of abstraction, from raw pixel values to edges and corners up to complex spatial patterns. We present a deep FCN using dilated convolutions of increasing spatial support. It is capable of learning informative features capturing long-range pixel dependencies while keeping a limited number of network parameters. Experiments carried out on a Quickbird image acquired over the city of Dar es Salaam, Tanzania, show that the proposed FCN outperforms state-of-the-art convolutional networks. Moreover, the computational cost of the proposed technique is significantly lower than standard patch-based architectures.

This study aims to characterize the spatial arrangement of land cover features and integrate the spatial arrangement information with other commonly used land use indicators. The characterization is conducted at object level, corresponding to land cover objects. At the local urban level, a VHR image is dominated by buildings. Therefore, the characterization of spatial arrangement of land cover elements is mainly conducted on building objects. Since building objects have different functional properties in urban areas, we classify them into a set of different types according to their geometrical, morphological, and contextual properties. The spatial arrangement is characterized by considering the composition of different building types. Regarding the integration of land use indicators and spatial arrangement information, we construct a Bayesian network, in which the spatial arrangement is served as a latent variable, and the land use indicators calculated according to existing studies are treated as the nodes of this Bayesian network. This is followed by urban land use classification. We applied our proposed method to a subset of a Pleiades image for an urban area of Wuhan, China. We conclude that our proposed method can provide an effective means for urban land use extraction.

There is a rapid diminishing ice extent in the Arctic but an opposite trend in Antarctic. The observed Antarctic sea ice extent to be expanding at a statistically significant rate 16.5±3.5×103km2yr-1(IPCC 2013), reported by the IPCC AR5, whereas the trend is statistically insignificant at a rate of 5.6±9.2×103km2yr-1in IPCC AR4 (IPCC 2007). The trends are derived from a long time series of passive microwave data, covering both the South Pole and the North Pole regions almost daily since 1970s. The sea ice concentration (proportion of ice area in a pixel) has been retrieved from the brightness temperature of passive microwave data, and the sea ice extent is defined as the area of ice that has an ice concentration no less than 15% (to avoid weather filtering issues near the ice edge). In an conventional way, the daily extents are averaged to monthly mean values, on the basis of those, monthly deviations are derived and linear regression model is applied to determine the rate (Parkinson and Cavalieri 2012). During this process, uncertainties from sensor transitions, processing method update, the addition of new data sources and the choice for a statistical method can all influence the final result (Cavalieri et al. 2012; Eisenman et al. 2014).

The recognition of spatial patterns of trees from satellite images is important for forestry, horticulture, and wildlife management. As a spatial optimization problem, we used variogram matching as the objective function for simulated tree arrangements and their simulated reproductions. We developed a variogram difference-based spatial simulated annealing (VDBSSA) method and applied it in reproducing alternate simulated tree arrangements in several orchards in India. Regular, linear, and sparse configurations with clearly separated objects could be distinguished. QuickBird 2 panchromatic images (0.6-m resolution) were binarized, followed by generating simulated images of a similar appearance. The performance of VDBSSA for alternate arrangements with the same spatial structure was explained by pattern characteristics. The density of objects inversely influenced the number of possible alternate arrangements. A sparse configuration gave more choices of alternate spatial arrangements than a dense configuration. VDBSSA is valuable for generating a pattern of well identifiable objects from a remote sensing image.

This paper proposes spatial distribution pattern-based subpixel mapping (SPMS) as a novel subpixel mapping (SPM) strategy. It separately considers spatial distribution patterns of different types of geographical objects. Initially, it classifies geographical objects into areal, linear, and point patterns according to their spatially geometric characteristics. For the different patterns, SPMS uses the vectorial boundary-based SPM algorithm with the spatial dependence assumption to deal with areal objects, the linear template matching-based SPM algorithm for linear objects, and the spatial pattern consistency matching-based SPM algorithm for point objects. The three patterns are integrated to generate a subpixel map. An artificially created image and two remotely sensed images were used to evaluate the performance of SPMS. The results were compared with a traditional hard classifier and seven existing SPM methods. The experimental results demonstrated that SPMS performed better than the hard classification and traditional SPM methods, particularly when dealing with linear and point objects.

Geographical information systems support the application of statistical techniques to map spatially referenced crop data. To do this in the optimal way, errors and uncertainties have to be minimized that are often associated with operations on the data. This paper applies a spatial statistical approach to upscale crop yields from the field level toward the scale of Burkina Faso. Observed yields were related to the Normalized Difference Vegetation Index derived from SPOT-VEGETATION. The objective was to quantify the uncertainties at the subsequent steps. First, we applied a point pattern analysis to examine uncertainties due to the sampling network of field surveys in the country. Second, geographically weighted regression kriging (GWRK) was applied to upscale the yield observations and to quantify the corresponding uncertainty. The proposed method was demonstrated with the mapping of sorghum yields in Burkina Faso and results were compared with those from regression kriging (RK) and kriging with external drift using a local kriging neighborhood (KEDLN). The proposed method was validated with independent yield observations obtained from field surveys. We observed that the lower uncertainty range value increased by 39%, and the upper uncertainty range value decreased by 51%, when comparing GWRK with RK and KEDLN. Moreover, GWRK reduced the prediction error variance as compared to RK (20 vs. 31) and to KEDLN (20 vs. 39). We found that climate and topography had a major impact on the country’s sorghum yields. Further, the financial ability of farmers influenced the crop management and, thus, the sorghum crop yields. We concluded that GWRK effectively utilized information present in the covariate datasets and improved the accuracies of both the regional-scale mapping of sorghum yields and was able to quantify the associated uncertainty.

This paper introduces a weighted total least squares (WTLS)-based estimator into image registration to deal with the coordinates of control points (CPs) that are of unequal accuracy. The performance of the estimator is investigated by means of simulation experiments using different coordinate errors. Comparisons with ordinary least squares (LS), total LS (TLS), scaled TLS, and weighted LS estimators are made. A novel adaptive weight determination scheme is applied to experiments with remotely sensed images. These illustrate the practicability and effectiveness of the proposed registration method by collecting CPs with different-sized errors from multiple reference images with different spatial resolutions. This paper concludes that the WTLS-based iteratively reweighted TLS method achieves a more robust estimation of model parameters and higher registration accuracy if heteroscedastic errors occur in both the coordinates of reference CPs and target CPs.

Canopy leaf area index (LAI) is a quantitative measure of canopy foliar area. LAI values can be derived from Moderate Resolution Imaging Spectroradiometer (MODIS) images. In this paper, MODIS pixels from a heterogeneous forest located in The Netherlands were decomposed using the linear mixture model using class fractions derived from a high-resolution aerial image. Gaussian Bayesian networks (GBNs) were applied to improve the spatio-temporal estimation of LAI by combining the decomposed MODIS images with a spatial version of physiological principles predicting growth (3PG) model output at different moments in time. Results showed that the spatial-temporal output obtained with the GBN was 40% more accurate than the spatial 3PG, with a root-mean-square error below 0.25. We concluded that the GBNs improved the spatial estimation of LAI values of a heterogeneous forest by combining a spatial forest growth model with satellite imagery.

Remote-sensing-based drought monitoring methods provide fast and useful information for a sustainable management strategy of drought impact over a region. Common pixel-based monitoring methods are limited in the analysis of the dynamics of this impact at regional scale. For instance, these hardly allow us to quantify the movement of drought in space and time and to compare drought with rainfall deficits without losing the variability of these events within a region. This study proposed an object-based approach that allowed us to visualize and quantify the spatio-temporal movement of drought impact on vegetation, called vegetative drought, in a region. The GIS software Dynomap was used to extract and track objects. Measures of distance and angle were used for determining the speed and direction of vegetative drought and rainfall deficit objects, calculated from the National Oceanic and Atmospheric Administration's (NOAA's) normalized difference vegetation index and rainfall estimates data. The methods were applied to the two rainy seasons during the drought year 1999 in East Africa. Results showed that vegetative drought objects moved into the southwestern direction at an average angle of −138.5° during the first season and −144.5° during the second season. The speed of objects varied between 38 km dekad−1 and 185 km dekad−1 during the first season and between 33 km dekad−1 and 144 km dekad−1 during the second season, reflecting the rate of spread between dekads. Vegetative drought objects close to rainfall deficit objects showed similar trajectories and sometimes regions overlapped. This indicated that the two events are related. We conclude that a spatiotemporal relationship existed between the two types of events and that this could be quantified.

This paper introduces geometric active contours and level-set evolution for the semi-automatic monitoring of urban trees in a sequence of very high resolution images. In our implementation active contours locally identify tree crown image regions using a localized data energy term and competition of energy forces between adjacent contours. In a first step, a surface fitting operation generates a set of ellipses which are used as the initial state of contour optimization. Contours evolve simultaneously using a Gaussian localized energy term which considers the energy of the evolving contour and its neighboring contours. We incorporate prior information into the multitemporal analysis, as contours are propagated and optimized through the sequence of images. We report the implementation and obtained results using a set of two and four aerial images of a residential area in New York, USA and Enschede, The Netherlands, respectively.

Wireless sensor network (WSN) applications require efficient, accurate and timely data analysis in order to facilitate (near) real-time critical decision-making and situation awareness. Accurate analysis and decision-making relies on the quality of WSN data as well as on the additional information and context. Raw observations collected from sensor nodes, however, may have low data quality and reliability due to limited WSN resources and harsh deployment environments. This article addresses the quality of WSN data focusing on outlier detection. These are defined as observations that do not conform to the expected behaviour of the data. The developed methodology is based on time-series analysis and geostatistics. Experiments with a real data set from the Swiss Alps showed that the developed methodology accurately detected outliers in WSN data taking advantage of their spatial and temporal correlations. It is concluded that the incorporation of tools for outlier detection in WSNs can be based on current statistical methodology. This provides a usable and important tool in a novel scientific field.

The leaf area index (LAI) is a biophysical variable related to atmosphere-biosphere exchange of CO2. One way to obtain LAI value is by the Moderate Resolution Imaging Spectroradiometer (MODIS) biophysical products. In this paper, we use this product to improve the physiological principles predicting growth model within a Gaussian Bayesian network (GBN) setup. The MODIS time series, however, contains gaps caused by persistent clouds, cloud contamination, and other technique problems. We used the Expectation Maximization (EM) algorithm to estimate these missing values. During a period of 26 successive months, the EM algorithm is applied to four different cases: successively and not successively missing values during two different winter seasons, successively and not successively missing values during one spring season, and not successively missing values during the full study. Results show that the maximum value of the averaged absolute error between the original values and those estimated equals 0.16. This low value indicates that the estimated values well represent the original values. Moreover, the root mean square error of the GBN output reduces from 1.57 to 1.49 when performing the EM algorithm to estimate the not successively missing values. We conclude that the EM algorithm within a GBN can adequately handle missing MODIS LAI values and improves the estimation of the LAI.

We develop a change detection model of tree crowns in urban areas addressing the gradual transition of vegetation in space and the fuzziness of tree crowns in remote sensing images. In the model, crown memberships are estimated for two images using soft classifiers. Next, Gaussian surface fitting is applied on the crown membership images to obtain tree crown objects parameters. Tree crowns are then modeled as both: elliptical two dimensional objects and as fuzzy regions. Bi-temporal changes are established by assessing differences in overlap of the modeled objects. The change model was successfully tested in a residential area of The Netherlands using a Quick- Bird and an aerial image of 2006 and 2009 respectively.

Estimating the contribution of the forests to carbon sequestration is commonly done by applying forest growth models. Such models inherently use field observations such as leaf area index (LAI), whereas a relevant information is also available from remotely sensed images. This paper aims to improve the LAI estimated from the forest growth model [physiological principals predicting growth (3-PG)] by combining these values with the LAI derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite imagery. A Bayesian networks (BNs) approach addresses the bias in the 3-PG model and the noise of the MODIS images. A novel inference strategy within the BN has been developed in this paper to take care of the different structures of the inaccuracies in the two data sources. The BN is applied to the Speulderbos forest in The Netherlands, where the detailed data were available. This paper shows that the outputs obtained with the BN were more accurate than either the 3-PG or the MODIS estimate. It was also found that the BN is more sensitive to the variation of the LAI derived from MODIS than to the variation of the LAI 3-PG values. In this paper, we conclude that the BNs can improve the estimation of the LAI values by combining a forest growth model with satellite imagery.

Investigations in data quality and uncertainty modeling are becoming key topics in geoinformation science. This paper models a collection of outcomes from a standard segmentation algorithm as a random set. It quantifies extensional uncertainties of extracted objects using statistical characteristics of random sets. The approach is applied to a synthetic data set and vegetation patches in the Poyang Lake area in China. These patches are of interest as they have both sharp and vague boundaries. Results show that random sets provide useful spatial information on uncertainties using their basic parameters like the mean, level sets, and variance. The number of iterations to achieve a stable covering function and the sum of the variances are good indicators of boundary sharpness. The coefficient of variation has a positive relation with the degree of uncertainty. An asymmetry ratio reflects the uneven gradual changes along different directions where broad boundaries exist. This paper shows that several characteristics of extensional uncertainty of segmented objects can be quantified numerically and spatially by random sets.

Domestic fires at the city level, being causes for casualties and causing significant material damages, are stored as a point pattern in a GIS. In this paper we apply a statistical point pattern analysis to derive major causes from related layers of information. We fit a G-function to analyse neighbourhood relations and a Strauss process for inferring causal relations. Using open-source software we find significant differences in patterns and explaining factors between the different parts of the day, in particular for different building types and income groups. We conclude that a quantitative spatial model can be fitted and that this provides a useful opportunity for fire brigades to improve planning their efforts.

The paper describes a novice method for sampling geochemicals to characterize mine tailings. We model the spatial relationships between a multi-element signature and, as covariates, abundance estimates of secondary iron-bearing minerals in mine tailings dumps. The covariates of interest, are readily, but less accurately obtainable by using airborne hyperspectral data and estimated through spectral unmixing. Via simulated annealing an optimal prospective sampling scheme for a new unvisited area is derived based on the variogram model of a previously sampled area.

This paper explores the effects of class separability in Markov-random-field-based superresolution mapping (SRM). We propose to account for class separability by means of controlling the balance tuned by a smoothness parameter between the prior and the likelihood terms in the posterior energy function. A generally applicable procedure estimates the optimal smoothness parameter, based on local energy balance analysis. The study shows how the optimal value of the smoothness parameter depends quantitatively and monotonically upon the class separability and the scale factor. Effects are studied on an image synthesized from an agricultural scene with field boundary subpixels. We varied systematically the class separability, the scale factor, and the smoothness parameter values. The accuracy of the resulting land-cover-map image is assessed by means of the kappa statistic at the fine-resolution scale and the class area proportion at the coarse-resolution scale. Performance is compared with a hard and a soft classification of the coarse-resolution image. We demonstrate that an optimal value of the smoothness parameter exists for each combination of scale factor and class separability. This allows us to reach a high classification accuracy (kappa = 0.85) even for poorly separable classes, i.e., with a transformed divergence equal to 0.5 and a scale factor equal to 10. The developed procedure agrees with the empirical data for the optimal smoothness parameter. The study shows that SRM is now applicable to a larger set of images with class separability ranging from poor to excellent.

Many spatial phenomena exhibit vagueness. Representation of such phenomena requires vague objects. In previous work, we provided definitions for vague objects: vague points, vague lines, and vague regions. Each of these objects is presented as a fuzzy set in ℝ2that satisfies well-defined properties. In this paper, we propose a number of geometric measures for vague objects, using the concept of mass distribution. The membership function of a vague object can be seen as a mass distribution. According to this view, a crisp object is a body with constant density, and a vague object is a body of varying density. We provide mathematical definitions for length of a vague line, area of a vague region, centroid of a vague object, as well as a measure for the vagueness of an object. The length of a vague line and the area of a vague region are indeed the mass of the vague line and of the vague region, respectively. Both metrics give an average of the values of the corresponding crisp metric on the α-cuts of the vague object. The centroid of a vague object is its centre of mass associated with a membership degree. The last metric functions as a measure of the degree of vagueness for a vague object.

Vagueness is often present in spatial phenomena. Representing and analysing vague spatial phenomena requires vague objects and operators, whereas current GIS and spatial databases can only handle crisp objects. This paper provides mathematical definitions for vague object types and operators. The object types that we propose are a set of simple types, a set of general types, and vague partitions. The simple types represent identifiable objects of a simple structure, i.e. not divisible into components. They are vague points, vague lines, and vague regions. The general types represent classes of simple type objects. They are vague multipoint, vague multiline, and vague multiregion. General types assure closure under set operators. Simple and general types are defined as fuzzy sets in ℝ2 satisfying specific properties that are expressed in terms of topological notions. These properties assure that set membership values change mostly gradually, allowing stepwise jumps. The type vague partition is a collection of vague multiregions that might intersect each other only at their transition boundaries. It allows for a soft classification of space. All types allow for both a finite and an infinite number of transition levels. They include crisp objects as special cases. We consider a standard set of operators on crisp objects and define them for vague objects. We provide definitions for operators returning spatial types. They are regularized fuzzy set operators: union, intersection, and difference; two operators from topology: boundary and frontier; and two operators on vague partitions: overlay and fusion. Other spatial operators, topological predicates and metric operators, are introduced giving their intuition and example definitions. All these operators include crisp operators as special cases. Types and operators provided in this paper form a model for a spatial data system that can handle vague information. The paper is illustrated with an application of vague objects in coastal erosion.

Meteosat satellites with the Spinning Enhanced Visible and Infrared Imagery (SEVIRI) sensor onboard provide remote-sensing images nowadays every 15 min. This paper investigates and applies image-mining methods to make an optimal use of images. It develops a simple, time-efficient, and generic model to facilitate pattern discovery and analysis. The focus of this paper is to develop a model for monitoring and analyzing forest fires in space and time. As an illustration, a diurnal cycle of fire in Portugal on July 28, 2004 was analyzed. Kernel convolution characterized the hearth of the fire as an object in space. Objects were extracted and tracked over time automatically. The results thus obtained were used to make a linear model for fire behavior with respect to vegetation and wind characteristics as explanatory variables. This model may be useful for predicting hazards at an almost real-time basis. The research illustrates how image mining improves information extraction from the Meteosat SEVIRI images

Object hierarchy is often ignored when collecting and classifying geographical objects. Object attributes are defined on the basis of uncertain parameters that may change in space and time. In this paper, we consider fuzzy decision trees for classification and a Bayesian hierarchical model for modeling and handling uncertainty. The study is illustrated with dynamic geographical objects from a coastal management application in the northern part of The Netherlands. Hierarchical modeling is applied to obtain posterior distributions for several boundary regions. The posterior distributions yield lower and upper limits of membership functions describing boundaries between object classes. In this way, a proper fuzzy decision tree for the coastal management application is built, which includes the inherent dynamic uncertainty.

This paper presents a method for estimating the partial abundance of spectrally similar minerals in complex mixtures. The method requires formulation of a linear function of individual spectra of individual minerals. The first and second derivatives of each of the different sets of mixed spectra and the individual spectra are determined. The error is minimized by means of simulated annealing. Experiments were made on several different mixtures of selected endmember, which could plausibly occur in real situations. The variance of the differences between the first derivatives of the observed spectrum and the first derivatives of the endmember spectra gives the most precise estimates for the partial abundance of each endmember. We conclude that the use of first-order derivatives provides a valuable contribution to unmixing procedures provided that the signal-to-noise ratio is at least 50 : 1

An ontological approach in GIS serves as a framework for the conceptualization of processes in the real world. In this paper, we examine an application in coastal change in the Netherlands, whereby beaches are subject to artificial nourishment to offset the effect of severe erosion. The use of ontologies helps to define two scenarios: SI determined by the regulations from the Ministry for Public Works; SII grounded on the abilities from an existing spatial dataset. A comparison between SI and SII shows that 72.8% of the objects suitable and unsuitable for nourishment are correctly classified. A higher overlap is found in areas where actual beach nourishments were carried out. Inaccuracies in attributes influence the determination of the objects. A sensitivity analysis applied to altitude illustrates a significant increase of objects suitable for nourishment for both scenarios, when altitude is decreased within the lower limit of the root mean square error for the 95% confidence interval. Moreover, the sensitivity of altitude shows that artificial boundaries for beach nourishment objects are not reasonable and consequently should be treated as vague objects.

Thematic maps prepared from remotely sensed images require a statistical accuracy assessment. For this purpose, the /spl kappa/-statistic is often used. This statistic does not distinguish between whether one unit is classified as another, or vice versa. In this paper, the Bradley-Terry (BT) model is applied for accuracy assessment. This model compares categories pairwise. The probability of one class over another class is estimated as well as the expected values of class pixels. The study is illustrated with an Advanced Spaceborne Thermal Emission and Reflection Radiometer image from the Netherlands, to which a maximum-likelihood classification with the Euclidean distance is applied. An error matrix is generated using an IKONOS image from the same area as ground truth. It is shown to which degree the BT model extends the /spl kappa/-statistic. A comparison with the Mahalanobis distance is made. Standardization is carried out to overcome problems emerging from the fact that a common BT model does not include the number of correctly classified pixels. The study shows how the BT model serves as an alternative to the usual /spl kappa/-statistic.

Variability in per cell classification accuracy is predominantly modelled with land-cover class as the explanatory variable, i.e. with users' accuracies from the error matrix. Logistic regression models were developed to include other explanatory variables: heterogeneity in the 3×3 window around a cell, the size of the patch and the complexity of the landscape in which a cell is located. It was found that per cell, the probability of correct classification was significantly (α=0.05) higher for cells with a less heterogeneous neighbourhood, for cells part of larger patches and for cells in regions with a less heterogeneous landscape. To validate the models, a leave-one-out procedure was applied in which the absolute difference between the actual and the model-estimated number of cells correctly classified was summarized over 55 regions in the Netherlands. The sum of differences reduced from 60.9 to 48.1 after adding the variables ‘patch size’ and ‘landscape dominance’ to the land-cover class model. Spatial variability thus modelled therefore led to a substantial improvement in the estimation of the per cell classification accuracy.

This research addresses existential uncertainty of spatial objects derived from satellite sensor imagery. An image segmentation technique is applied at various values of splitting and merging thresholds. We test the hypothesis that objects occurring at many segmentation steps have less existential uncertainty than those occurring at only a few steps.

This paper applies generalised linear statistical techniques in a GIS to analyse wildlife data from a Kenyan wildlife reserve and its surrounding areas. Attention focuses on the spatial distribution of elephant during nine successive surveys, analysing their temporal and spatial relationship to 12 environmental covariates. A principal component analysis identifies five major determining factors, thereby reducing dimensionality in the data, while a simple spatial analysis procedure, suitable for wildlife data obtained from airborne surveys, quantfies clustering for different animal species. The number of explanatory variables appearing in abundance models is found to be subject to large variations during successive surveys with a minimum and maximum of four and eight variables, respectively. Species from highly clustered populations are found to have over 20 times more observations within short distances compared to the rest. The study concludes that a combination of generalised linear modelling and GIS gives deeper insight into the dynamics of wildlife species in and around well-defined nature reserves.

The paper describes an iterative extension to maximum a posteriori (MAP) supervised classification methods. A posteriori probabilities per class are used for classification as well as to obtain class area estimates. From these, an updated set of prior probabilities is calculated and used in the next iteration. The process converges to statistically correct area estimates. The iterative process can be combined effectively with a stratification of the image, which is made on the basis of additional map data. Moreover, it relies on the sample sets being representative. Therefore, the method is shown to be well applicable in combination with an existing GIS. The paper gives a description of the procedure and provides a mathematical foundation. An example is presented to distinguish residential, industrial, and greenhouse classes. A significant improvement of the classification was obtained.

Geographical Information Systems can be used for processing spatial data to assess the risks of environmental contamination. Their use depends upon the amount and quality of available data, the models for interpolation and calculation of health risk, data processing procedures and interactivity. In this study, it is shown that interactive use of GIS is important for obtaining physically relevant results. Three forms of interactivity are distinguished: interactivity by means of user interfacing, interactivity requiring additional information outside GIS and interactivity when changing the use of GIS. Three stages in which interaction with GIS are crucial can be distinguished: application of geostatistics, choice of appropriate models, and decision making. This study focuses on three cases in The Netherlands dealing with soil contamination and soil stress analysis and with implications for risk assessment in which interactivity within GIS is analysed. Emphasis is given to contour volumes of polluted soil, to combine GIS with deterministic models and to apply land use scenarios. Finally, there is a discussion of how some forms of interactivity could be replaced by expert systems.

Methods are needed for monitoring the propagation of errors when spatial models are driven by quantitative data stored in raster geographical information systems. This paper demonstrates how the standard stochastic theory of error propagation can be extended and applied to continuously differentiable arithmetic operations (quantitative models) for manipulating gridded map data. The statistical methods have been programmed using the Taylor series expansion to approximate the models. Model inputs are (a) model coefficients and their standard errors and (b) maps of continuous variables and the associated prediction errors, which can be obtained by optimal interpolation from point data. The model output is a map that is accompanied by a map of prediction errors. The relative contributions of the errors in the inputs (model coefficients, maps of individual variables) can be determined and mapped separately allowing judgments to be made about subsequent survey optimization. The methods are illustrated by two case studies.