Search

ABSTRACTWith the recent increase in studies on spatial heterogeneity, geographically weighted (GW) models have become an essential set of local techniques, attracting a wide range of users from different domains. In this study, we demonstrate a newly developed standalone GW software, GWmodelS using a community-level house price data set for Wuhan, China. In detail, a number of fundamental GW models are illustrated, including GW descriptive statistics, basic and multiscale GW regression, and GW principle component analysis. Additionally, functionality in spatial data management and batch mapping are presented as essential supplementary activities for GW modeling. The software provides significant advantages in terms of a user-friendly graphical user interface, operational efficiency, and accessibility, which facilitate its usage for users from a wide range of domains.

Abstract Many cities are facing challenges caused by the increasing use of motorised transport and Hanoi, Vietnam, is no exception. The proliferation of petrol powered motorbikes has caused serious problems of congestion, pollution, and road safety. This paper reports on a new survey dataset that was created as part of the Urban Transport Modelling for Sustainable Well-Being in Hanoi (UTM-Hanoi) project. The survey of nearly 30,000 respondents gathers data on households’ demographics, perceptions, opinions and stated behaviours. The data are informative in their own right and have also been used to experiment with multi-scale spatial statistics, synthetic population generation and machine learning approaches to predicting an individual’s perceptions of potential government policies. The paper reports on the key findings from the survey and conducts a technical validation to contrast the outcomes to similar datasets that are available.

ABSTRACTAs an established spatial analytical tool, Geographically Weighted Regression (GWR) has been applied across a variety of disciplines. However, its usage can be challenging for large datasets, which are increasingly prevalent in today’s digital world. In this study, we propose two high-performance R solutions for GWR via Multi-core Parallel (MP) and Compute Unified Device Architecture (CUDA) techniques, respectively GWR-MP and GWR-CUDA. We compared GWR-MP and GWR-CUDA with three existing solutions available in Geographically Weighted Models (GWmodel), Multi-scale GWR (MGWR) and Fast GWR (FastGWR). Results showed that all five solutions perform differently across varying sample sizes, with no single solution a clear winner in terms of computational efficiency. Specifically, solutions given in GWmodel and MGWR provided acceptable computational costs for GWR studies with a relatively small sample size. For a large sample size, GWR-MP and FastGWR provided coherent solutions on a Personal Computer (PC) with a common multi-core configuration, GWR-MP provided more efficient computing capacity for each core or thread than FastGWR. For cases when the sample size was very large, and for these cases only, GWR-CUDA provided the most efficient solution, but should note its I/O cost with small samples. In summary, GWR-MP and GWR-CUDA provided complementary high-performance R solutions to existing ones, where for certain data-rich GWR studies, they should be preferred.

Rising global demand for palm oil has created environmental pressures related to deforestation, burning, and peat exploitation, which in turn drives increased greenhouse gas (GHG) emissions. GHG emissions in oil palm (OP) production are known to vary spatially. However, temporal changes across contrasting management and soil types, are less well studied. This paper quantifies spatiotemporal GHG emissions across contrasting regions, management types, and soil types for the period 1990–2019 to assess the potential for reducing emission. The study focusses on Indonesia, as the biggest producer of OP, and in particular on the North Sumatra and Riau provinces, where OP is intensively produced. GHG inventories in 5 year time steps were constructed to investigate the change in drivers of emissions using spatial data, resampled to a 500 m grid. Total GHG emissions were found to have increased in both regions due to expanding OP production. However, results show a reduction in emissions flux from 1.98 to 1.15 Ton C _eq . ha ^−1 yr ^−1 in North Sumatra and 9.63–2.67 Ton C _eq . ha ^−1 yr ^−1 in Riau over the study period. This reduced flux was linked to the decreased deforestation and burning activities, together with increased biomass increment from lower carbon stock area conversion to OP. In both provinces, smallholder plantations emitted fewer emissions than industrial ones, and production on organic soils resulted in consistently higher emissions than on mineral soils. In North Sumatra, emissions under all management and soil types were found to decrease. In Riau, however, GHG emissions on organic soils regardless of management types, remained high. Our findings emphasise that potential for low-emissions OP production is attainable by reducing emissions per unit area through an improved understanding of GHG emissions spatiotemporal variability and their drivers. These contribute to reinforcing ongoing government regulations and guiding the industry towards low-emission OP productions.

Geodemographic classifications are useful tools for segmenting populations and have many applications but are not suitable for measuring neighbourhood change over time. There is a need for an approach that uses data of a higher spatiotemporal resolution to capture the fundamental dimensions of processes driving local changes. Data primitives are measures that capture the fundamental drivers of neighbourhood processes and therefore offer a suitable route. In this article, three types of gentrification are conceptualised, and four key data primitives are applied to capture them in a case study region in Yorkshire, England. These areas are visually validated according to their temporal properties to confirm the presence of gentrification and are then assigned to a high-level gentrification type. Ensemble modelling is then used to predict the presence, type, and temporal properties of gentrification across the rest of England. The results show an alignment of the spatial extent of gentrification types with previous gentrification studies throughout the country but may have made an overprediction in London. The periodicities of (1) residential, (2) rural, and (3) transport-led gentrification also vary throughout the country, but regardless of type, gentrification in areas within close proximity to one another have differing velocities such that they peak and complete within similar times. These temporal findings offer new, more timely tools for authorities in devising schedules of interventions and for understanding the intricacies of neighbourhood change.

In this paper we consider some of the issues of working with big data and big spatial data and highlight the need for an open and critical framework. We focus on a set of challenges underlying the collection and analysis of big data. In particular, we consider 1) inference when working with usually biased big data, challenging the assumed inferential superiority of data with observations, n, approaching N, the population n -> N. We also emphasise 2) the need for analyses that answer questions of practical significance or with greater emphasis on the size of the effect, rather than the truth or falsehood of a statistical statement; 3) the need to accept messiness in your data and to document all operations undertaken on the data because of this, in support of openness and reproducibility paradigms; and 4) the need to explicitly seek to understand the causes of bias, messiness etc in the data and the inferential consequences of using such data in analyses, by adopting critical approaches to spatial data science. In particular we consider the need to place individual data science studies in a wider social and economic contexts, along with the role of inferential theory in the presence of big data, and issues relating to messiness and complexity in big data.

Geographically weighted regression (GWR) is an inherently exploratory technique for examining process non-stationarity in data relationships. This paper develops and applies a hyper-local GWR which extends such investigations further. The hyper-local GWR simultaneously optimizes both local model selection (which covariates to include in each local regression) and local kernel bandwidth specification (how much data should be included locally). These are evaluated using a measure of model fit. The hyper-local GWR approach evaluates different kernel bandwidths at each location and selects the most parsimonious local regression model. By allowing models and bandwidths to vary locally, this approach extends and refines the one-size-fits-all "whole map model" and "constant bandwidth calibration" under standard GWR. The results provide an alternative, complementary and more nuanced interpretation of localized regression. The method is illustrated using a case study modeling soil total nitrogen (STN) and soil total phosphorus (STP) from data collected at 689 locations in a watershed in Northern China. The analysis compares linear regression, standard GWR, and hyper-local GWR models of STN and STP and highlights the different locations at which covariates are identified as significant predictors of STN and STP by the different GWR approaches and the spatial variation in bandwidths. The hyper-local GWR results indicate that the STN processes are more non-stationary and localized than found via a standard application of GWR. By contrast, the results for STP are more confirmatory (i.e., similar) between the two GWR approaches providing extra assurance about the nature of the moderate non-stationary relationships observed. That is, a standard GWR may underestimate localized spatial heterogeneity where it is strongly present (as in the STN case study) and may overestimate it where spatial homogeneity is present (as in the STP case study). The overall benefits of hyper-local GWR are discussed, particularly in the context of the original investigative aims of GWR. A hyper-local approach provides a useful counter view of local regression modeling to that found with standard GWR. Where spatial non-stationarity exists, the hyper-local GWR provides a more spatially nuanced indication of the localization than a standard GWR analysis and can be used to suggest the direction of further analyses and investigations. Some areas of further work are suggested.

Spatial data are used in many scientific domains including analyses of Ecosystem Services (ES) and Natural Capital (NC), with results used to inform planning and policy. However, the data spatial scale (or support) has a fundamental impact on analysis outputs and, thus, process understanding and inference. The Modifiable Areal Unit Problem (MAUP) describes the effects of scale on analyses of spatial data and outputs, but it has been ignored in much environmental research, including evaluations of land use with respect to ES and NC. This paper illustrates the MAUP through an ES optimisation problem. The results show that MAUP effects are unpredictable and nonlinear, with discontinuities specific to the spatial properties of the case study. Four key recommendations are as follows: (1) The MAUP should always be tested for in ES evaluations. This is commonly performed in socio-economic analyses. (2) Spatial aggregation scales should be matched to process granularity by identifying the aggregation scale at which processes are considered to be stable (stationary) with respect to variances, covariances, and other moments. (3) Aggregation scales should be evaluated along with the scale of decision making (e.g., agricultural field, farm holding, and catchment). (4) Researchers in ES and related disciplines should up-skill themselves in spatial analysis and core paradigms related to scale to overcome the scale blindness commonly found in much research.

Heavy geographic patterning to the 2016 Brexit vote in UK and Trump vote in US has resulted in numerous ecological analyses of variations in area-level voting behaviours. We extend this work by employing modelling approaches that permit regionally-specific associations between outcome and explanatory variables. We do so by generating a large number of regional models using penalised regression for variable selection and coefficient evaluation. The results reinforce those already published in that we find associations in support of a 'left-behind' reading. Multivariate models are dominated by a single variable-levels of degree-education. Net of this effect, 'secondary' variables help explain the vote, but do so differently for different regions. For Brexit, variables relating to material disadvantage, and to a lesser extent structural-economic circumstances, are more important for regions with a strong industrial history than for regions that do not share such a history. For Trump, increased material disadvantage reduces the vote both in global models and models built mostly for Southern states, thereby undermining the 'left-behind' reading. The reverse is nevertheless true for many other states, particularly those in New England and the Mid-Atlantic, where comparatively high levels of disadvantage assist the Trump vote and where model outputs are more consistent with the UK, especially so for regions with closer economic histories. This pattern of associations is exposed via our regional modelling approach, application of penalised regression and use of carefully designed visualization to reason over 100+ model outputs located within their spatial context. Our analysis, documented in an accompanying github repository, is in response to recent calls in empirical Social and Political Science for fuller exploration of subnational contexts that are often controlled out of analyses, for use of modelling techniques more robust to replication and for greater transparency in research design and methodology.

This paper reviews geodemographic classifications and developments in contemporary classifications. It develops a critique of current approaches and identifiea a number of key limitations. These include the problems associated with the geodemographic cluster label (few cluster members are typical or have the same properties as the cluster centre) and the failure of the static label to describe anything about the underlying neighbourhood processes and dynamics. To address these limitations, this paper proposed a data primitives approach. Data primitives are the fundamental dimensions or measurements that capture the processes of interest. They can be used to describe the current state of an area in a multivariate feature space, and states can be compared over multiple time periods for which data are available, through for example a change vector approach. In this way, emergent social processes, which may be too weak to result in a change in a cluster label, but are nonetheless important signals, can be captured. As states are updated (for example, as new data become available), inferences about different social processes can be made, as well as classification updates if required. State changes can also be used to determine neighbourhood trajectories and to predict or infer future states. A list of data primitives was suggested from a review of the mechanisms driving a number of neighbourhood-level social processes, with the aim of improving the wider understanding of the interaction of complex neighbourhood processes and their effects. A small case study was provided to illustrate the approach. In this way, the methods outlined in this paper suggest a more nuanced approach to geodemographic research, away from a focus on classifications and static data, towards approaches that capture the social dynamics experienced by neighbourhoods.

Volunteered Geographical Information (VGI) and social media can provide information about real-time perceptions, attitudes and behaviours in urban green space (UGS). This paper reviews the use of VGI and social media data in research examining UGS. The current state of the art is described through the analysis of 177 papers to (1) summarise the characteristics and usage of data from different platforms, (2) provide an overview of the research topics using such data sources, and (3) characterise the research approaches based on data pre-processing, data quality assessment and improvement, data analysis and modelling. A number of important limitations and priorities for future research are identified. The limitations include issues of data acquisition and representativeness, data quality, as well as differences across social media platforms in different study areas such as urban and rural areas. The research priorities include a focus on investigating factors related to physical activities in UGS areas, urban park use and accessibility, the use of data from multiple sources and, where appropriate, making more effective use of personal information. In addition, analysis approaches can be extended to examine the network suggested by social media posts that are shared, re-posted or reacted to and by being combined with textual, image and geographical data to extract more representative information for UGS analysis.

This paper describes the development and application of a novel and generic framework for parsimonious soil-water interaction models to predict the risk of agro-chemical runoff. The underpinning models represent two scales to predict runoff risk in fields and the delivery of mobilized pesticides to river channel networks. Parsimonious field and landscape scale runoff risk models were constructed using a number of pre-computed parameters in combination with live rainfall data. The precomputed parameters included spatially-distributed historical rainfall data to determine long term average soil water content and the sensitivity of land use and soil type combinations to runoff. These were combined with real-time live rainfall data, freely available through open data portals and APIs, to determine runoff risk using SCS Curve Numbers. The rainfall data was stored to provide antecedent, current and future rainfall inputs. For the landscape scale model, the delivery risk of mobilized pesticides to the river network included intrinsic landscape factors. The application of the framework is illustrated for two case studies at field and catchment scales, covering acid herbicide at field scale and metaldehyde at landscape scale. Web tools were developed and the outputs provide spatially and temporally explicit predictions of runoff and pesticide delivery risk at 1 km2 resolution. The model parsimony reflects the driving nature of rainfall and soil saturation for runoff risk and the critical influence of both surface and drain flow connectivity for the risk of mobilized pesticide being delivered to watercourses. The novelty of this research lies in the coupling of live spatially-distributed weather data with precomputed runoff and delivery risk parameters for crop and soil types and historical rainfall trends. The generic nature of the framework supports the ability to model the runoff and field-to-channel delivery risk associated with any in-field agricultural application assuming application rate data are available.

The debate over the conceptual constructs of landscape aesthetics, specifically whether landscape quality is inherently related to landscape physical characteristics or is subjectively “in the eye of the beholder,” has continued for years. Solutions accommodating both the biophysical and perceptual aspects of landscapes are thus desirable for landscape planners and policymakers. In response to policy shifts that emphasise both expert and public landscape perspectives, this study investigates the relationships between formal and informal landscape evaluations. It analyses crowdsourced data describing landscape aesthetic quality (Scenic-Or-Not) and authoritative landscape quality assessments (the Landscape Assessment Decision Making Process (LANDMAP) of Wales). Some agreement was found regarding landforms most likely to be perceived as scenic or unattractive by experts and non-experts, which aligns with previous landscape perception studies. However, contested landscape typologies are identified formal and informal landscape aesthetic evaluations are compared. Several limitations and implications for current formal landscape assessment paradigms (GIS based and vertical) are discussed and several approaches for capturing on-the-ground perceptions are suggested including recent extensions to GIS derived viewsheds (e.g., vertical voxel viewsheds).

A series of non-spatial and spatial hedonic models of feeding and replacement cattle prices at video auctions in Uruguay (2002 to 2009) were specified with predictors measuring marketing conditions (e.g., steer price), cattle characteristics (e.g., breed) and agro-ecological factors (e.g., soil productivity, water characteristics, pasture condition, season). Results indicated that cattle prices produced under extensive production systems were influenced by all of predictor categories, confirming that found previously. Although many of the agro-ecological predictors were inherently spatial in nature, the incorporation of spatial effects into the estimation of the hedonic model itself, through either a spatially-autocorrelated error term or allowing the regression coefficients to vary spatially and at different scales, was able to provide greater insight into the cattle price process. Through the latter extension, using a multiscale geographically weighted regression, which was the most informative and most accurate model, relationships between cattle price and predictors operated at a mixture of global, regional, local and highly local spatial scales. This result is considered a key advance, where uncovering, interpreting, and utilizing such rich spatial information can help improve the geographical provenance of Uruguayan beef and is critically important for maintaining Uruguay’s status as a key exporter of beef with respect to the health and safety benefits of natural, open-sky, grass-fed production systems.

Land change is a key topic in research on global environmental change, and the restoration of degraded land is the core component of the global Land Degradation Neutrality target under the UN 2030 Agenda for Sustainable Development. In this study, remote-sensing-derived land-use data were used to characterize the land-change processes in China’s Loess Plateau, which is experiencing large-scale ecological restoration. Geographically Weighted Regression was applied to capture the spatiotemporal variations in land change and driving-force relationships. First, we explored land-use change in the Loess Plateau for the period 1990−2015. Grassland, cropland and forestland were dominant land cover in the region, with a total percentage area of 88%. The region experienced dramatic land-use transitions during the study period: degraded grassland and wetland, expansion of cropland and built-up land and weak restoration of forestland during 1990−2000; and increases in grassland, built-up land, forestland and wetland, concurrent with shrinking cropland during 2000−2015. A Geographically Weighted Regression (GWR) analysis revealed altitude to be the common dominant factor associated with the four major land-use types (forestland, grassland, cropland and built-up land). Altitude and slope were found to be positively associated with forestland, while being negatively associated with cropland in the high, steep central region. For both forestland and grassland, temperature and precipitation behaved in a similar manner, with a positive hotspot in the northwest. Altitude, slope and distance to road were all negatively associated with built-up land across the region. The GWR captured the spatial non-stationarity on different socioeconomic driving forces. Spatial heterogeneity and temporal variation of the impact of socioeconomic drivers indicate that the ecological restoration projects positively affected the region’s greening trend with hotspots in the center and west, and also improved farmer well-being. Notably, urban population showed undesired effects, expressed in accelerating grassland degradation in central and western regions for 1990−2000, hindering forestland and grassland restoration in the south during 2000−2015, and highlighting the long-term sustainability of the vegetation restoration progress. Such local results have the potential to provide a methodological contribution (e.g., nesting local-level approaches, i.e., GWR, within land system research) and spatially explicit evidence for context-related and proactive land management (e.g., balancing urbanization and ecological restoration processes and advancing agricultural development and rural welfare improvement).

This study describes the integration and analysis of travel smart card data (SCD) with points of interest (POIs) from social media for a case study in Shenzhen, China. SCD ticket price with tap-in and tap-out times was used to identify different groups of travellers. The study examines the temporal variations in mobility, identifies different groups of users and characterises their trip purpose and identifies sub-groups of users with different travel patterns. Different groups were identified based on their travel times and trip costs. The trip purpose associated with different groups was evaluated by constructing zones around metro station locations and identifying the POIs in each zone. Each POI was allocated to one of six land use types, and each zone was allocated a set of land use weights based on the number of POI check-ins for the POIs in that zone. Trip purpose was then inferred from trip time linked to the land use at the origin and destination zones using a novel “land use change rate” measure. A cluster analysis was used to identify sub-groups of users based on individual temporal travel patterns, which were used to generate a novel “boarding time profile”. The results show how different groups of users can be identified and the differences in trip times and trip purpose quantified between and within groups. Limitations of the study are discussed and a number of areas for further work identified, including linking to socioeconomic data and a deeper consideration of the timestamps of POI check-ins to support the inference of dynamic and multiple land uses at one location. The methods and metrics developed by this research use social media POI data to semantically contextualise information derived from the SCD and to overcome the drawbacks and limitations of traditional travel survey data. They are novel and generalizable to other studies. They quantify spatiotemporal mobility patterns for different groups of travellers and infer how their purposes of their journeys change through the day. In so doing, they support a more nuanced and detailed view of who, where, when and why people use city spaces.

Volunteered geographical information (VGI) and citizen science have become important sources data for much scientific research. In the domain of land cover, crowdsourcing can provide a high temporal resolution data to support different analyses of landscape processes. However, the scientists may have little control over what gets recorded by the crowd, providing a potential source of error and uncertainty. This study compared analyses of crowdsourced land cover data that were contributed by different groups, based on nationality (labelled Gondor and Non-Gondor) and on domain experience (labelled Expert and Non-Expert). The analyses used a geographically weighted model to generate maps of land cover and compared the maps generated by the different groups. The results highlight the differences between the maps how specific land cover classes were under- and over-estimated. As crowdsourced data and citizen science are increasingly used to replace data collected under the designed experiment, this paper highlights the importance of considering between group variations and their impacts on the results of analyses. Critically, differences in the way that landscape features are conceptualised by different groups of contributors need to be considered when using crowdsourced data in formal scientific analyses. The discussion considers the potential for variation in crowdsourced data, the relativist nature of land cover and suggests a number of areas for future research. The key finding is that the veracity of citizen science data is not the critical issue per se. Rather, it is important to consider the impacts of differences in the semantics, affordances and functions associated with landscape features held by different groups of crowdsourced data contributors.

This paper develops and applies a novel method for inferring land cover/land use (LCLU) change that combines direct multi-date classification with measures from a change vector analysis. The model predicts change directly rather than the land cover at either time, although these could be inferred. Unsupervised classifications of bi-temporal imagery were manually labeled and used to generate reference data for class-to-class changes. These were used to train a Random Forest model with inputs from the bi-temporal image bands and change vector measures (change vector direction, angle and the spectral angle) and used to generate a predicted surface of land cover change for a case study in the Pearl River Delta, China. The overall accuracy of LCLU change prediction was 96% and specific class-to-class changes had errors rates of 0–12.8%. Some errors were related the semi-automated labeling of the training data. The spectral angle variables and Near Infra-Red image bands for both years were found to be strong predictors of change. Odd ratios were used to quantify regional differences in land cover change rates in urban and peri-urban areas. The regional differences and origins of the observed errors are discussed, along with some areas of further work. The key contributions of this paper are the focus on change rather than LCLU through the construction of a model to predict changes directly and the development of an approach that provides quick, effective and informative analysis of LCLU change in support of policy and planning in rapidly urbanizing areas.

There is currently a lack of in-situ environmental data for the calibration and validation of remotely sensed products and for the development and verification of models. Crowdsourcing is increasingly being seen as one potentially powerful way of increasing the supply of in-situ data but there are a number of concerns over the subsequent use of the data, in particular over data quality. This paper examined crowdsourced data from the Geo-Wiki crowdsourcing tool for land cover validation to determine whether there were significant differences in quality between the answers provided by experts and non-experts in the domain of remote sensing and therefore the extent to which crowdsourced data describing human impact and land cover can be used in further scientific research. The results showed that there was little difference between experts and non-experts in identifying human impact although results varied by land cover while experts were better than non-experts in identifying the land cover type. This suggests the need to create training materials with more examples in those areas where difficulties in identification were encountered, and to offer some method for contributors to reflect on the information they contribute, perhaps by feeding back the evaluations of their contributed data or by making additional training materials available. Accuracies were also found to be higher when the volunteers were more consistent in their responses at a given location and when they indicated higher confidence, which suggests that these additional pieces of information could be used in the development of robust measures of quality in the future.

Citizens are increasingly becoming an important source of geographic information, sometimes entering domains that had until recently been the exclusive realm of authoritative agencies. This activity has a very diverse character as it can, amongst other things, be active or passive, involve spatial or aspatial data and the data provided can be variable in terms of key attributes such as format, description and quality. Unsurprisingly, therefore, there are a variety of terms used to describe data arising from citizens. In this article, the expressions used to describe citizen sensing of geographic information are reviewed and their use over time explored, prior to categorizing them and highlighting key issues in the current state of the subject. The latter involved a review of ~100 Internet sites with particular focus on their thematic topic, the nature of the data and issues such as incentives for contributors. This review suggests that most sites involve active rather than passive contribution, with citizens typically motivated by the desire to aid a worthy cause, often receiving little training. As such, this article provides a snapshot of the role of citizens in crowdsourcing geographic information and a guide to the current status of this rapidly emerging and evolving subject.

Data fusion represents a powerful way of integrating individual sources of information to produce a better output than could be achieved by any of the individual sources on their own. This paper focuses on the data fusion of different land cover products derived from remote sensing. In the past, many different methods have been applied, without regard to their relative merit. In this study, we compared some of the most commonly-used methods to develop a hybrid forest cover map by combining available land cover/forest products and crowdsourced data on forest cover obtained through the Geo-Wiki project. The methods include: nearest neighbour, naive Bayes, logistic regression and geographically-weighted logistic regression (GWR), as well as classification and regression trees (CART). We ran the comparison experiments using two data types: presence/absence of forest in a grid cell; percentage of forest cover in a grid cell. In general, there was little difference between the methods. However, GWR was found to perform better than the other tested methods in areas with high disagreement between the inputs.

This paper describes and illustrates methods for quantifying regional differences in land use/land cover changes. A series of approaches are used to analyse differences in land cover change from data held in change matrices. These are contingency tables and are commonly used in remote sensing to describe the spatial coincidence of land cover recorded over two time periods. Comparative analyses of regional change are developed using odds ratios to analyse data in two regions. These approaches are extended using generalised linear models to analyse data for three or more regions. A generalised Poisson regression model is used to generate a comparative index of change based on differences in change likelihoods. Mosaic plots are used to provide a visual representation of statistically surprising land use losses and gains. The methods are explored using a hypothetical but tractable dataset and then applied to a national case study of coastal land use changes over 50 years conducted for the National Trust. The suitability of the different approaches to different types of problem and the potential for their application to land cover accuracy measures are briefly discussed.

Regular monitoring of expanding impervious surfaces areas (ISAs) in urban areas is highly desirable. MODIS data can meet this demand in terms of frequent observations but are lacking in spatial detail, leading to the mixed land cover problem when per-pixel classifications are applied. To overcome this issue, this research develops and applies a spatio-temporal sub-pixel model to estimate ISAs on an annual basis during 2001–2013 in the Jakarta Metropolitan Area, Indonesia. A Random Forest (RF) regression inferred the ISA proportion from annual 23 values of MODIS MOD13Q1 EVI and reference data in which such proportion was visually allocated from very high-resolution images in Google Earth over time at randomly selected locations. Annual maps of ISA proportion were generated and showed an average increase of 30.65 km2/year over 13 years. For comparison, a series of RF per-pixel classifications were also developed from the same reference data using a Boolean class constructed from different thresholds of ISA proportion. Results from per-pixel models varied when such thresholds change, suggesting difficulty of estimation of actual ISAs. This research demonstrated the advantages of spatio-temporal sub-pixel analysis for annual ISAs mapping and addresses the problem associated with definitions of thresholds in per-pixel approaches.