A simulated 'sandbox' for exploring the modifiable areal unit problem in aggregation and disaggregation.

We present a spatial testbed of simulated boundary data based on a set of very high-resolution census-based areal units surrounding Guadalajara, Mexico. From these input areal units, we simulated 10 levels of spatial resolutions, ranging from levels with 5,515-52,388 units and 100 simulated zonal configurations for each level - totalling 1,000 simulated sets of areal units. These data facilitate interrogating various realizations of the data and the effects of the spatial coarseness and zonal configurations, the Modifiable Areal Unit Problem (MAUP), on applications such as model training, model prediction, disaggregation, and aggregation processes. Further, these data can facilitate the production of spatially explicit, non-parametric estimates of confidence intervals via bootstrapping. We provide a pre-processed version of these 1,000 simulated sets of areal units, meta- and summary data to assist in their use, and a code notebook with the means to alter and/or reproduce these data.

Wildlife impacts and changing climate pose compounding threats to human food security.

High-level policy debates surrounding elephant management often dominate global conservation headlines, yet realities for people living with wildlife are not adequately incorporated into policymaking or evident in related discourse.1,2 Human health and livelihoods can be severely impacted by wildlife and indirectly by policy outcomes.3 In landscapes where growing human and elephant (Loxodonta spp. and Elephas maximus) populations compete over limited resources, human-elephant conflict causes crop loss, human injury and death, and retaliatory killing of wildlife.4-6 Across Africa, these problems may be increasingly compounded by climate change, which intensifies resource competition and food insecurity.6-9 Here, we examine how human-wildlife impacts interact with climate change and household food insecurity across the Kavango-Zambezi Transfrontier Conservation Area, the world's largest terrestrial transboundary conservation area, spanning five African nations. We use hierarchical Bayesian statistical models to analyze multi-country household data together with longitudinal satellite-based climate measures relevant to rainfed agriculture. We find that crop depredation by wildlife, primarily elephants, impacts 58% of sampled households annually and is associated with significant increases in food insecurity. These wildlife impacts compound effects of changing climate on food insecurity, most notably observed as a 5-day shortening of the rainy season per 10 years across the data record (1981-2018). To advance sustainability goals, global conservation policy must better integrate empirical evidence on the challenges of human-wildlife coexistence into longer term strategies at transboundary scales, specifically in the context of climate change.3,9-11.

Measuring the contribution of built-settlement data to global population mapping.

Top-down population modelling has gained applied prominence in public health, planning, and sustainability applications at the global scale. These top-down population modelling methods often rely on remote-sensing (RS) derived representation of the built-environment and settlements as key predictive covariates. While these RS-derived data, which are global in extent, have become more advanced and more available, gaps in spatial and temporal coverage remain. These gaps have prompted the interpolation of the built-environment and settlements, but the utility of such interpolated data in further population modelling applications has garnered little research. Thus, our objective was to determine the utility of modelled built-settlement extents in a top-down population modelling application. Here we take modelled global built-settlement extents between 2000 and 2012, created using a spatio-temporal disaggregation of observed settlement growth. We then demonstrate the applied utility of such annually modelled settlement data within the application of annually modelling population, using random forest informed dasymetric disaggregations, across 172 countries and a 13-year period. We demonstrate that the modelled built-settlement data are consistently the 2nd most important covariate in predicting population density, behind annual lights at night, across the globe and across the study period. Further, we demonstrate that this modelled built-settlement data often provides more information than current annually available RS-derived data and last observed built-settlement extents.

Wildlife impacts and vulnerable livelihoods in a transfrontier conservation landscape.

Interactions between humans and wildlife resulting in negative impacts are among the most pressing conservation challenges globally. In regions of smallholder livestock and crop production, interactions with wildlife can compromise human well-being and motivate negative sentiment and retaliation toward wildlife, undermining conservation goals. Although impacts may be unavoidable when human and wildlife land use overlap, scant large-scale human data exist quantifying the direct costs of wildlife to livelihoods. In a landscape of global importance for wildlife conservation in southern Africa, we quantified costs for people living with wildlife through a fundamental measure of human well-being, food security, and we tested whether existing livelihood strategies buffer certain households against crop depredation by wildlife, predominantly elephants. To do this, we estimated Bayesian multilevel statistical models based on multicounty household data (n = 711) and interpreted model results in the context of spatial data from participatory land-use mapping. Reported crop depredation by wildlife was widespread. Over half of the sample households were affected and household food security was reduced significantly (odds ratio 0.37 [0.22, 0.63]). The most food insecure households relied on gathered food sources and welfare programs. In the event of crop depredation by wildlife, these 2 livelihood sources buffered or reduced harmful effects of depredation. The presence of buffering strategies suggests a targeted compensation strategy could benefit the region's most vulnerable people. Such strategies should be combined with dynamic and spatially explicit land-use planning that may reduce the frequency of negative human-wildlife impacts. Quantifying and mitigating the human costs from wildlife are necessary steps in working toward human-wildlife coexistence.

Impactos de la Fauna y Medios de Subsistencia Vulnerables en unkl Paisaje de Conservación Transfronteriza Resumen Las interacciones entre los humanos y la fauna que resultan en impactos negativos se encuentran entre los desafíos más apremiantes para la conservación a nivel mundial. En las regiones de ganaderos y agricultores minifundistas, las interacciones con la fauna pueden poner en peligro el bienestar humano y motivar sentimientos negativos y represalias hacia la fauna, lo que debilita los objetivos de conservación. Aunque los impactos pueden evitarse cuando el uso de suelo por humanos y fauna se traslapa, existen pocos datos humanos a gran escala que cuantifiquen el costo directo de la fauna para los medios de subsistencia. Cuantificamos el costo para las personas que conviven con animales silvestres en un paisaje de importancia global para la conservación de fauna en el sur de África. La cuantificación fue realizada por medio de una medida fundamental de bienestar humano y seguridad alimentaria, y probamos si las estrategias existentes de subsistencia amortiguan a ciertos hogares ante la depredación de cultivos realizada por animales silvestres, predominantemente los elefantes. Para realizar esto, estimamos algunos modelos estadísticos bayesianos de niveles múltiples basados en los datos de hogares ubicados en múltiples condados (n = 711) e interpretamos los resultados de los modelos en el contexto de los datos espaciales a partir de un mapeo participativo de uso de suelo. La depredación de cultivos por animales silvestres fue reportada de manera generalizada. Más de la mitad de los hogares en la muestra estuvieron afectados y la seguridad alimenticia de los hogares se redujo significativamente (proporción de probabilidades 0.37 [0.22, 0.63]). Los hogares con la menor seguridad alimentaria dependían de fuentes de recolección de alimentos y programas de bienestar. En el evento de la depredación por fauna de los cultivos, estas dos fuentes de subsistencia amortiguaron o redujeron los efectos dañinos de la depredación. La presencia de las estrategias de amortiguamiento sugiere que una estrategia de compensación enfocada podría beneficiar a las personas más vulnerables de la región. Dichas estrategias deberían estar combinadas con la planeación del uso de suelo dinámica y espacialmente explícita, la cual podría reducir la frecuencia de los impactos negativos entre los humanos y la fauna. La cuantificación y mitificación del costo humano a partir de la fauna son pasos necesarios en el camino hacia la coexistencia entre los humanos y la fauna.

Annually modelling built-settlements between remotely-sensed observations using relative changes in subnational populations and lights at night.

Mapping urban features/human built-settlement extents at the annual time step has a wide variety of applications in demography, public health, sustainable development, and many other fields. Recently, while more multitemporal urban features/human built-settlement datasets have become available, issues still exist in remotely-sensed imagery due to spatial and temporal coverage, adverse atmospheric conditions, and expenses involved in producing such datasets. Remotely-sensed annual time-series of urban/built-settlement extents therefore do not yet exist and cover more than specific local areas or city-based regions. Moreover, while a few high-resolution global datasets of urban/built-settlement extents exist for key years, the observed date often deviates many years from the assigned one. These challenges make it difficult to increase temporal coverage while maintaining high fidelity in the spatial resolution. Here we describe an interpolative and flexible modelling framework for producing annual built-settlement extents. We use a combined technique of random forest and spatio-temporal dasymetric modelling with open source subnational data to produce annual 100 m × 100 m resolution binary built-settlement datasets in four test countries located in varying environmental and developmental contexts for test periods of five-year gaps. We find that in the majority of years, across all study areas, the model correctly identified between 85 and 99% of pixels that transition to built-settlement. Additionally, with few exceptions, the model substantially out performed a model that gave every pixel equal chance of transitioning to built-settlement in each year. This modelling framework shows strong promise for filling gaps in cross-sectional urban features/built-settlement datasets derived from remotely-sensed imagery, provides a base upon which to create urban future/built-settlement extent projections, and enables further exploration of the relationships between urban/built-settlement area and population dynamics.

Global spatio-temporally harmonised datasets for producing high-resolution gridded population distribution datasets.

Multi-temporal, globally consistent, high-resolution human population datasets provide consistent and comparable population distributions in support of mapping sub-national heterogeneities in health, wealth, and resource access, and monitoring change in these over time. The production of more reliable and spatially detailed population datasets is increasingly necessary due to the importance of improving metrics at sub-national and multi-temporal scales. This is in support of measurement and monitoring of UN Sustainable Development Goals and related agendas. In response to these agendas, a method has been developed to assemble and harmonise a unique, open access, archive of geospatial datasets. Datasets are provided as global, annual time series, where pertinent at the timescale of population analyses and where data is available, for use in the construction of population distribution layers. The archive includes sub-national census-based population estimates, matched to a geospatial layer denoting administrative unit boundaries, and a number of co-registered gridded geospatial factors that correlate strongly with population presence and density. Here, we describe these harmonised datasets and their limitations, along with the production workflow. Further, we demonstrate applications of the archive by producing multi-temporal gridded population outputs for Africa and using these to derive health and development metrics. The geospatial archive is available at https://doi.org/10.5258/SOTON/WP00650.

Missing millions: undercounting urbanization in India.

The measurement and characterization of urbanization crucially depends upon defining what counts as urban. The government of India estimates that only 31% of the population is urban. We show that this is an artifact of the definition of urbanity and an underestimate of the level of urbanization in India. We use a random forest-based model to create a high-resolution (~ 100 m) population grid from district-level data available from the Indian Census for 2001 and 2011, a novel application of such methods to create temporally consistent population grids. We then apply a community-detection clustering algorithm to construct urban agglomerations for the entire country. Compared with the 2011 official statistics, we estimate 12% more of urban population, but find fewer mid-size cities. We also identify urban agglomerations that span jurisdictional boundaries across large portions of Kerala and the Gangetic Plain.

Evaluating nighttime lights and population distribution as proxies for mappinganthropogenic CO2 emission in Vietnam, Cambodia and Laos.

Tracking spatiotemporal changes in GHG emissions is key to successful implementation of the United Nations Framework Convention on Climate Change (UNFCCC). And while emission inventories often provide a robust tool to track emission trends at the country level, subnational emission estimates are often not reported or reports vary in robustness as the estimates are often dependent on the spatial modeling approach and ancillary data used to disaggregate the emission inventories. Assessing the errors and uncertainties of the subnational emission estimates is fundamentally challenging due to the lack of physical measurements at the subnational level. To begin addressing the current performance of modeled gridded CO2 emissions, this study compares two common proxies used to disaggregate CO2 emission estimates. We use a known gridded CO2 model based on satellite-observed nighttime light (NTL) data (Open Source Data Inventory for Anthropogenic CO2, ODIAC) and a gridded population dataset driven by a set of ancillary geospatial data. We examine the association at multiple spatial scales of these two datasets for three countries in Southeast Asia: Vietnam, Cambodia and Laos and characterize the spatiotemporal similarities and differences for 2000, 2005, and 2010. We specifically highlight areas of potential uncertainty in the ODIAC model, which relies on the single use of NTL data for disaggregation of the non-point emissions estimates. Results show, over time, how a NTL-based emissions disaggregation tends to concentrate CO2 estimates in different ways than population-based estimates at the subnational level. We discuss important considerations in the disconnect between the two modeled datasets and argue that the spatial differences between data products can be useful to identify areas affected by the errors and uncertainties associated with the NTL-based downscaling in a region with uneven urbanization rates.

Regional-scale management maps for forested areas of the Southeastern United States and the US Pacific Northwest.

Forests in the United States are managed by multiple public and private entities making harmonization of available data and subsequent mapping of management challenging. We mapped four important types of forest management, production, ecological, passive, and preservation, at 250-meter spatial resolution in the Southeastern (SEUS) and Pacific Northwest (PNW) USA. Both ecologically and socio-economically dynamic regions, the SEUS and PNW forests represent, respectively, 22.0% and 10.4% of forests in the coterminous US. We built a random forest classifier using seasonal time-series analysis of 16 years of MODIS 16-day composite Enhanced Vegetation Index, and ancillary data containing forest ownership, roads, US Forest Service wilderness and forestry areas, proportion conifer and proportion riparian. The map accuracies for SEUS are 89% (10-fold cross-validation) and 67% (external validation) and PNW are 91% and 70% respectively with the same validation. The now publicly available forest management maps, probability surfaces for each management class and uncertainty layer for each region can be viewed and analysed in commercial and open-source GIS and remote sensing software.

Gridded Population Maps Informed by Different Built Settlement Products.

The spatial distribution of humans on the earth is critical knowledge that informs many disciplines and is available in a spatially explicit manner through gridded population techniques. While many approaches exist to produce specialized gridded population maps, little has been done to explore how remotely sensed, built-area datasets might be used to dasymetrically constrain these estimates. This study presents the effectiveness of three different high-resolution built area datasets for producing gridded population estimates through the dasymetric disaggregation of census counts in Haiti, Malawi, Madagascar, Nepal, Rwanda, and Thailand. Modeling techniques include a binary dasymetric redistribution, a random forest with a dasymetric component, and a hybrid of the previous two. The relative merits of these approaches and the data are discussed with regards to studying human populations and related spatially explicit phenomena. Results showed that the accuracy of random forest and hybrid models was comparable in five of six countries.

Examining the correlates and drivers of human population distributions across low- and middle-income countries.

Geographical factors have influenced the distributions and densities of global human population distributions for centuries. Climatic regimes have made some regions more habitable than others, harsh topography has discouraged human settlement, and transport links have encouraged population growth. A better understanding of these types of relationships enables both improved mapping of population distributions today and modelling of future scenarios. However, few comprehensive studies of the relationships between population spatial distributions and the range of drivers and correlates that exist have been undertaken at all, much less at high spatial resolutions, and particularly across the low- and middle-income countries. Here, we quantify the relative importance of multiple types of drivers and covariates in explaining observed population densities across 32 low- and middle-income countries over four continents using machine-learning approaches. We find that, while relationships between population densities and geographical factors show some variation between regions, they are generally remarkably consistent, pointing to universal drivers of human population distribution. Here, we find that a set of geographical features relating to the built environment, ecology and topography consistently explain the majority of variability in population distributions at fine spatial scales across the low- and middle-income regions of the world.

Modelling changing population distributions: an example of the Kenyan Coast, 1979-2009.

Large-scale gridded population datasets are usually produced for the year of input census data using a top-down approach and projected backward and forward in time using national growth rates. Such temporal projections do not include any subnational variation in population distribution trends and ignore changes in geographical covariates such as urban land cover changes. Improved predictions of population distribution changes over time require the use of a limited number of covariates that are time-invariant or temporally explicit. Here we make use of recently released multi-temporal high-resolution global settlement layers, historical census data and latest developments in population distribution modelling methods to reconstruct population distribution changes over 30 years across the Kenyan Coast. We explore the methodological challenges associated with the production of gridded population distribution time-series in data-scarce countries and show that trade-offs have to be found between spatial and temporal resolutions when selecting the best modelling approach. Strategies used to fill data gaps may vary according to the local context and the objective of the study. This work will hopefully serve as a benchmark for future developments of population distribution time-series that are increasingly required for population-at-risk estimations and spatial modelling in various fields.

GridSample: an R package to generate household survey primary sampling units (PSUs) from gridded population data.

BACKGROUND: Household survey data are collected by governments, international organizations, and companies to prioritize policies and allocate billions of dollars. Surveys are typically selected from recent census data; however, census data are often outdated or inaccurate. This paper describes how gridded population data might instead be used as a sample frame, and introduces the R GridSample algorithm for selecting primary sampling units (PSU) for complex household surveys with gridded population data. With a gridded population dataset and geographic boundary of the study area, GridSample allows a two-step process to sample "seed" cells with probability proportionate to estimated population size, then "grows" PSUs until a minimum population is achieved in each PSU. The algorithm permits stratification and oversampling of urban or rural areas. The approximately uniform size and shape of grid cells allows for spatial oversampling, not possible in typical surveys, possibly improving small area estimates with survey results. RESULTS: We replicated the 2010 Rwanda Demographic and Health Survey (DHS) in GridSample by sampling the WorldPop 2010 UN-adjusted 100 m × 100 m gridded population dataset, stratifying by Rwanda's 30 districts, and oversampling in urban areas. The 2010 Rwanda DHS had 79 urban PSUs, 413 rural PSUs, with an average PSU population of 610 people. An equivalent sample in GridSample had 75 urban PSUs, 405 rural PSUs, and a median PSU population of 612 people. The number of PSUs differed because DHS added urban PSUs from specific districts while GridSample reallocated rural-to-urban PSUs across all districts. CONCLUSIONS: Gridded population sampling is a promising alternative to typical census-based sampling when census data are moderately outdated or inaccurate. Four approaches to implementation have been tried: (1) using gridded PSU boundaries produced by GridSample, (2) manually segmenting gridded PSU using satellite imagery, (3) non-probability sampling (e.g. random-walk, "spin-the-pen"), and random sampling of households. Gridded population sampling is in its infancy, and further research is needed to assess the accuracy and feasibility of gridded population sampling. The GridSample R algorithm can be used to forward this research agenda.

Improving Large Area Population Mapping Using Geotweet Densities.

Many different methods are used to disaggregate census data and predict population densities to construct finer scale, gridded population data sets. These methods often involve a range of high resolution geospatial covariate datasets on aspects such as urban areas, infrastructure, land cover and topography; such covariates, however, are not directly indicative of the presence of people. Here we tested the potential of geo-located tweets from the social media application, Twitter, as a covariate in the production of population maps. The density of geo-located tweets in 1x1 km grid cells over a 2-month period across Indonesia, a country with one of the highest Twitter usage rates in the world, was input as a covariate into a previously published random forests-based census disaggregation method. Comparison of internal measures of accuracy and external assessments between models built with and without the geotweets showed that increases in population mapping accuracy could be obtained using the geotweet densities as a covariate layer. The work highlights the potential for such social media-derived data in improving our understanding of population distributions and offers promise for more dynamic mapping with such data being continually produced and freely available.

Spatiotemporal patterns of population in mainland China, 1990 to 2010.

According to UN forecasts, global population will increase to over 8 billion by 2025, with much of this anticipated population growth expected in urban areas. In China, the scale of urbanization has, and continues to be, unprecedented in terms of magnitude and rate of change. Since the late 1970s, the percentage of Chinese living in urban areas increased from ~18% to over 50%. To quantify these patterns spatially we use time-invariant or temporally-explicit data, including census data for 1990, 2000, and 2010 in an ensemble prediction model. Resulting multi-temporal, gridded population datasets are unique in terms of granularity and extent, providing fine-scale (~100 m) patterns of population distribution for mainland China. For consistency purposes, the Tibet Autonomous Region, Taiwan, and the islands in the South China Sea were excluded. The statistical model and considerations for temporally comparable maps are described, along with the resulting datasets. Final, mainland China population maps for 1990, 2000, and 2010 are freely available as products from the WorldPop Project website and the WorldPop Dataverse Repository.

High-resolution gridded population datasets for Latin America and the Caribbean in 2010, 2015, and 2020.

The Latin America and the Caribbean region is one of the most urbanized regions in the world, with a total population of around 630 million that is expected to increase by 25% by 2050. In this context, detailed and contemporary datasets accurately describing the distribution of residential population in the region are required for measuring the impacts of population growth, monitoring changes, supporting environmental and health applications, and planning interventions. To support these needs, an open access archive of high-resolution gridded population datasets was created through disaggregation of the most recent official population count data available for 28 countries located in the region. These datasets are described here along with the approach and methods used to create and validate them. For each country, population distribution datasets, having a resolution of 3 arc seconds (approximately 100 m at the equator), were produced for the population count year, as well as for 2010, 2015, and 2020. All these products are available both through the WorldPop Project website and the WorldPop Dataverse Repository.

Modelling associations between public understanding, engagement and forest conditions in the Inland Northwest, USA.

Opinions about public lands and the actions of private non-industrial forest owners in the western United States play important roles in forested landscape management as both public and private forests face increasing risks from large wildfires, pests and disease. This work presents the responses from two surveys, a random-sample telephone survey of more than 1500 residents and a mail survey targeting owners of parcels with 10 or more acres of forest. These surveys were conducted in three counties (Wallowa, Union, and Baker) in northeast Oregon, USA. We analyze these survey data using structural equation models in order to assess how individual characteristics and understanding of forest management issues affect perceptions about forest conditions and risks associated with declining forest health on public lands. We test whether forest understanding is informed by background, beliefs, and experiences, and whether as an intervening variable it is associated with views about forest conditions on publicly managed forests. Individual background characteristics such as age, gender and county of residence have significant direct or indirect effects on our measurement of understanding. Controlling for background factors, we found that forest owners with higher self-assessed understanding, and more education about forest management, tend to hold more pessimistic views about forest conditions. Based on our results we argue that self-assessed understanding, interest in learning, and willingness to engage in extension activities together have leverage to affect perceptions about the risks posed by declining forest conditions on public lands, influence land owner actions, and affect support for public policies. These results also have broader implications for management of forested landscapes on public and private lands amidst changing demographics in rural communities across the Inland Northwest where migration may significantly alter the composition of forest owner goals, understanding, and support for various management actions.

Disaggregating census data for population mapping using random forests with remotely-sensed and ancillary data.

High resolution, contemporary data on human population distributions are vital for measuring impacts of population growth, monitoring human-environment interactions and for planning and policy development. Many methods are used to disaggregate census data and predict population densities for finer scale, gridded population data sets. We present a new semi-automated dasymetric modeling approach that incorporates detailed census and ancillary data in a flexible, "Random Forest" estimation technique. We outline the combination of widely available, remotely-sensed and geospatial data that contribute to the modeled dasymetric weights and then use the Random Forest model to generate a gridded prediction of population density at ~100 m spatial resolution. This prediction layer is then used as the weighting surface to perform dasymetric redistribution of the census counts at a country level. As a case study we compare the new algorithm and its products for three countries (Vietnam, Cambodia, and Kenya) with other common gridded population data production methodologies. We discuss the advantages of the new method and increases over the accuracy and flexibility of those previous approaches. Finally, we outline how this algorithm will be extended to provide freely-available gridded population data sets for Africa, Asia and Latin America.

Dynamic population mapping using mobile phone data.

During the past few decades, technologies such as remote sensing, geographical information systems, and global positioning systems have transformed the way the distribution of human population is studied and modeled in space and time. However, the mapping of populations remains constrained by the logistics of censuses and surveys. Consequently, spatially detailed changes across scales of days, weeks, or months, or even year to year, are difficult to assess and limit the application of human population maps in situations in which timely information is required, such as disasters, conflicts, or epidemics. Mobile phones (MPs) now have an extremely high penetration rate across the globe, and analyzing the spatiotemporal distribution of MP calls geolocated to the tower level may overcome many limitations of census-based approaches, provided that the use of MP data is properly assessed and calibrated. Using datasets of more than 1 billion MP call records from Portugal and France, we show how spatially and temporarily explicit estimations of population densities can be produced at national scales, and how these estimates compare with outputs produced using alternative human population mapping methods. We also demonstrate how maps of human population changes can be produced over multiple timescales while preserving the anonymity of MP users. With similar data being collected every day by MP network providers across the world, the prospect of being able to map contemporary and changing human population distributions over relatively short intervals exists, paving the way for new applications and a near real-time understanding of patterns and processes in human geography.