Spatially explicit accuracy assessment of deep learning-based, fine-resolution built-up land data in the United States.

Geospatial datasets derived from remote sensing data by means of machine learning methods are often based on probabilistic outputs of abstract nature, which are difficult to translate into interpretable measures. For example, the Global Human Settlement Layer GHS-BUILT-S2 product reports the probability of the presence of built-up areas in 2018 in a global 10 m × 10 m grid. However, practitioners typically require interpretable measures such as binary surfaces indicating the presence or absence of built-up areas or estimates of sub-pixel built-up surface fractions. Herein, we assess the relationship between the built-up probability in GHS-BUILT-S2 and reference built-up surface fractions derived from a highly reliable reference database for several regions in the United States. Furthermore, we identify a binarization threshold using an agreement maximization method that creates binary built-up land data from these built-up probabilities. These binary surfaces are input to a spatially explicit, scale-sensitive accuracy assessment which includes the use of a novel, visual-analytical tool which we call focal precision-recall signature plots. Our analysis reveals that a threshold of 0.5 applied to GHS-BUILT-S2 maximizes the agreement with binarized built-up land data derived from the reference built-up area fraction. We find high levels of accuracy (i.e., county-level F-1 scores of almost 0.8 on average) in the derived built-up areas, and consistently high accuracy along the rural-urban gradient in our study area. These results reveal considerable accuracy improvements in human settlement models based on Sentinel-2 data and deep learning, as compared to earlier, Landsat-based versions of the Global Human Settlement Layer.

Assessing the relationship between morphology and mapping accuracy of built-up areas derived from global human settlement data.

It is common knowledge that the level of landscape heterogeneity may affect the performance of remote sensing based land use / land cover classification. While this issue has been studied in depth for land cover data in general, the specific relationship between the mapping accuracy and morphological characteristics of built-up surfaces has not been analyzed in detail, an urgent need given the recent emergence of a variety of global, fine-resolution settlement datasets. Moreover, previous studies typically rely on aggregated, broad-scale landscape metrics to quantify the morphology of built-up areas, neglecting the fine-grained spatial variation and scale dependency of such metrics. Herein, we aim to fill this knowledge gap by assessing the associations between localized (focal) landscape metrics, derived from binary built-up surfaces and localized data accuracy estimates. We tested our approach for built-up surfaces from the Global Human Settlement Layer (GHSL) for Massachusetts (USA). Specifically, we examined the explanatory power of landscape metrics with respect to both commission and omission errors in the multi-temporal GHS-BUILT R2018A data product. We found that the Landscape Shape Index (LSI) calculated in focal windows exhibits, on average, the highest levels of correlation to focal accuracy measures. These relationships are scale-dependent, and become stronger with increasing level of spatial support. We found that thematic omission error, as measured by Recall, has the strongest relationship to measures of built-up surface morphology across different temporal epochs and spatial resolutions. The results of our regression analysis (R2>0.9), estimating accuracy based on landscape metrics, confirmed these findings. Lastly, we tested the generalizability of our findings by regionally stratifying our regression models and applying them to a different version of the GHSL (i.e., the GHS-BUILT-S2) and a different study area. We observed varying levels of model transferability, indicating that the relationship between accuracy and landscape metrics may be sensor-specific, and is heavily localized for most accuracy metrics, but quite generalizable for the Recall measure. This indicates that there is a strong and generalizable association between morphological properties of built-up land and the degree to which it is "undermapped".

Data-enriched Interpolation for Temporally Consistent Population Compositions.

This research evaluates the performance of areal interpolation coupled with dasymetric refinement to estimate different demographic attributes, namely population sub-groups based on race, age structure and urban residence, within consistent census tract boundaries from 1990 to 2010 in Massachusetts. The creation of such consistent estimates facilitates the study of the nuanced micro-scale evolution of different aspects of population, which is impossible using temporally incompatible small-area census geographies from different points in time. Various unexplored ancillary variables, including the Global Human Settlement Layer (GHSL), the National Land-Cover Database (NLCD), parcels, building footprints and the proprietary ZTRAX® dataset are utilized for dasymetric refinement prior to areal interpolation to examine their effectiveness in improving the accuracy of multi-temporal population estimates. Different areal interpolation methods including Areal Weighting (AW), Target Density Weighting (TDW), Expectation Maximization (EM) and its data-extended approach are coupled with different dasymetric refinement scenarios based on these ancillary variables. The resulting consistent small area estimates of white and black subpopulations, people of age 18-65 and urban population show that dasymetrically refined areal interpolation is particularly effective when the analysis spans a longer time period (1990-2010 instead of 2000-2010) and the enumerated population is sufficiently large (e.g., counts of white vs. black). The results also demonstrate that current census-defined urban areas overestimate the spatial distribution of urban population and dasymetrically refined areal interpolation improves estimates of urban population. Refined TDW using building footprints or the ZTRAX® dataset outperforms all other methods. The implementation of areal interpolation enriched by dasymetric refinement represents a promising strategy to create more reliable multi-temporal and consistent estimates of different population subgroups and thus demographic compositions. This methodological foundation has the potential to advance micro-scale modeling of various subpopulations, particularly urban population to inform studies of urbanization and population change over time as well as future population projections.

Assessing the Accuracy of Multi-Temporal Built-Up Land Layers across Rural-Urban Trajectories in the United States.

Global data on settlements, built-up land and population distributions are becoming increasingly available and represent important inputs to a better understanding of key demographic processes such as urbanization and interactions between human and natural systems over time. One persistent drawback that prevents user communities from effectively and objectively using these data products more broadly, is the absence of thorough and transparent validation studies. This study develops a validation framework for accuracy assessment of multi-temporal built-up land layers using integrated public parcel and building records as validation data. The framework is based on measures derived from confusion matrices and incorporates a sensitivity analysis for potential spatial offsets between validation and test data as well as tests for the effects of varying criteria of the abstract term built-up land on accuracy measures. Furthermore, the framework allows for accuracy assessments by strata of built-up density, which provides important insights on the relationship between classification accuracy and development intensity to better instruct and educate user communities on quality aspects that might be relevant to different purposes. We use data from the newly-released Global Human Settlement Layer (GHSL), for four epochs since 1975 and at fine spatial resolution (38m), in the United States for a demonstration of the framework. The results show very encouraging accuracy measures that vary across study areas, generally improve over time but show very distinct patterns across the rural-urban trajectories. Areas of higher development intensity are very accurately classified and highly reliable. Rural areas show low degrees of accuracy, which could be affected by misalignment between the reference data and the data under test in areas where built-up land is scattered and rare. However, a regression analysis, which examines how well GHSL can estimate built-up land using spatially aggregated analytical units, indicates that classification error is mainly of thematic nature. Thus, caution should be taken in using the data product in rural regions. The results can be useful in further improving classification procedures to create measures of the built environment. The validation framework can be extended to data-poor regions of the world using map data and Volunteered Geographic Information.

Internal and International Mobility as Adaptation to Climatic Variability in Contemporary Mexico: Evidence from the Integration of Census and Satellite Data.

Migration provides a strategy for rural Mexican households to cope with, or adapt to, weather events and climatic variability. Yet prior studies on "environmental migration" in this context have not examined the differences between choices of internal (domestic) or international movement. In addition, much of the prior work relied on very coarse spatial scales to operationalize the environmental variables such as rainfall patterns. To overcome these limitations, we use fine-grain rainfall estimates derived from NASA's Tropical Rainfall Measuring Mission (TRMM) satellite. The rainfall estimates are combined with Population and Agricultural Census information to examine associations between environmental changes and municipal rates of internal and international migration 2005-2010. Our findings suggest that municipal-level rainfall deficits relative to historical levels are an important predictor of both international and internal migration, especially in areas dependent on seasonal rainfall for crop productivity. Although our findings do not contradict results of prior studies using coarse spatial resolution, they offer clearer results and a more spatially nuanced examination of migration as related to social and environmental vulnerability and thus higher degrees of confidence.

An Integrated Multi-Source Dataset for Measuring Settlement Evolution in the United States from 1810 to 2020.

Understanding changes in the built environment is vital for sustainable urban development and disaster preparedness. Recent years have seen the emergence of a variety of global, continent-level, and nation-wide datasets related to the current state and the evolution of the built environment, human settlements or building stocks. However, such datasets may face limitations like incomplete coverage, sparse building information, coarse resolution, and limited timeframes. This study addresses these challenges by integrating three spatial datasets to create an extensive, attribute-rich sequence of settlement layers spanning 200 years for the contiguous U.S. This integration process involves complex data processing, merging property-level real estate, parcel, and remote sensing-based building footprint data, and creating gridded multi-temporal settlement layers. This effort unveils the latest edition (Version 2) of the Historical Settlement Data Compilation for the U.S. (HISDAC-US), which includes the latest land use and structural information as of the year 2021. It enables detailed research on urban form and structure, helps assess and map the built environment's risk to natural hazards, assists in population modeling, supports land use analysis, and aids health studies.

The fastest-growing and most destructive fires in the US (2001 to 2020).

The most destructive and deadly wildfires in US history were also fast. Using satellite data, we analyzed the daily growth rates of more than 60,000 fires from 2001 to 2020 across the contiguous US. Nearly half of the ecoregions experienced destructive fast fires that grew more than 1620 hectares in 1 day. These fires accounted for 78% of structures destroyed and 61% of suppression costs ($18.9 billion). From 2001 to 2020, the average peak daily growth rate for these fires more than doubled (+249% relative to 2001) in the Western US. Nearly 3 million structures were within 4 kilometers of a fast fire during this period across the US. Given recent devastating wildfires, understanding fast fires is crucial for improving firefighting strategies and community preparedness.

Spatially and Temporally Varying Associations between Temporary Outmigration and Natural Resource Availability in Resource-Dependent Rural Communities in South Africa: A Modeling Framework.

Migration-environment models tend to be aspatial within chosen study regions, although associations between temporary outmigration and environmental explanatory variables likely vary across the study space. This research extends current approaches by developing migration models considering spatial non-stationarity and temporal variation - through examination of the migration-environment association at nested geographic scales (i.e. whole-population, village, and subvillage) within a specific study site. Demographic survey data from rural South Africa, combined with indicators of natural resource availability from satellite imagery, are employed in a nested modeling approach that brings out distinct patterns of spatial variation in model associations derived at finer geographic scales. Given recent heightened public and policy concern with the human migratory implications of climate change, we argue that consideration of spatial variability adds important nuance to scientific understanding of the migration-environment association.

MTBF-33: A multi-temporal building footprint dataset for 33 counties in the United States (1900 - 2015).

Despite abundant data on the spatial distribution of contemporary human settlements, historical datasets on the long-term evolution of human settlements at fine spatial and temporal granularity are scarce, limiting our quantitative understanding of long-term changes of built-up areas. This is because commonly used large-scale mapping methods (e.g., computer vision) and suitable data sources (i.e., aerial imagery, remote sensing data, LiDAR data) have only been available in recent decades. However, there are alternative data sources such as cadastral records that are digitally available, containing relevant information such as building construction dates, allowing for an approximate, digital reconstruction of past building distributions. We conducted a non-exhaustive search of open and publicly available data resources from administrative institutions in the United States and gathered, integrated, and harmonized cadastral parcel data, tax assessment data, and building footprint data for 33 counties, wherever building footprint geometries and building construction year information was available. The result of this effort is a unique dataset that we call the Multi-Temporal Building Footprint Dataset for 33 U.S. Counties (MTBF-33). MTBF-33 contains over 6.2 million building footprints including their construction year, and can be used to derive retrospective depictions of built-up areas from 1900 to 2015, at fine spatial and temporal grain. Moreover, MTBF-33 can be employed for data validation purposes, or to train statistical learning models aiming to extract historical information on human settlements from remote sensing data, historical maps, or similar data sources.

Gridded land use data for the conterminous United States 1940-2015.

Multiple aspects of our society are reflected in how we have transformed land through time. However, limited availability of historical-spatial data at fine granularity have hindered our ability to advance our understanding of the ways in which land was developed over the long-term. Using a proprietary, national housing and property database, which is a result of large-scale, industry-fuelled data harmonization efforts, we created publicly available sequences of gridded surfaces that describe built land use progression in the conterminous United States at fine spatial (i.e., 250 m × 250 m) and temporal resolution (i.e., 1 year - 5 years) between the years 1940 and 2015. There are six land use classes represented in the data product: agricultural, commercial, industrial, residential-owned, residential-income, and recreational facilities, as well as complimentary uncertainty layers informing the users about quantifiable components of data uncertainty. The datasets are part of the Historical Settlement Data Compilation for the U.S. (HISDAC-US) and enable the creation of new knowledge of long-term land use dynamics, opening novel avenues of inquiry across multiple fields of study.

Creeping disaster along the U.S. coastline: Understanding exposure to sea level rise and hurricanes through historical development.

Current estimates of U.S. property at risk of coastal hazards and sea level rise (SLR) are staggering-evaluated at over a trillion U.S. dollars. Despite being enormous in the aggregate, potential losses due to SLR depend on mitigation, adaptation, and exposure and are highly uneven in their distribution across coastal cities. We provide the first analysis of how changes in exposure (how and when) have unfolded over more than a century of coastal urban development in the United States. We do so by leveraging new historical settlement layers from the Historical Settlement Data Compilation for the U.S. (HISDAC-US) to examine building patterns within and between the SLR zones of the conterminous United States since the early twentieth century. Our analysis reveals that SLR zones developed faster and continue to have higher structure density than non-coastal, urban, and inland areas. These patterns are particularly prominent in locations affected by hurricanes. However, density levels in historically less-developed coastal areas are now quickly converging on early settled SLR zones, many of which have reached building saturation. These "saturation effects" suggest that adaptation polices targeting existing buildings and developed areas are likely to grow in importance relative to the protection of previously undeveloped land.

Towards the automated large-scale reconstruction of past road networks from historical maps.

Transportation infrastructure, such as road or railroad networks, represent a fundamental component of our civilization. For sustainable planning and informed decision making, a thorough understanding of the long-term evolution of transportation infrastructure such as road networks is crucial. However, spatially explicit, multi-temporal road network data covering large spatial extents are scarce and rarely available prior to the 2000s. Herein, we propose a framework that employs increasingly available scanned and georeferenced historical map series to reconstruct past road networks, by integrating abundant, contemporary road network data and color information extracted from historical maps. Specifically, our method uses contemporary road segments as analytical units and extracts historical roads by inferring their existence in historical map series based on image processing and clustering techniques. We tested our method on over 300,000 road segments representing more than 50,000 km of the road network in the United States, extending across three study areas that cover 42 historical topographic map sheets dated between 1890 and 1950. We evaluated our approach by comparison to other historical datasets and against manually created reference data, achieving F-1 scores of up to 0.95, and showed that the extracted road network statistics are highly plausible over time, i.e., following general growth patterns. We demonstrated that contemporary geospatial data integrated with information extracted from historical map series open up new avenues for the quantitative analysis of long-term urbanization processes and landscape changes far beyond the era of operational remote sensing and digital cartography.

Differences between gridded population data impact measures of geographic access to healthcare in sub-Saharan Africa.

Background: Access to healthcare is imperative to health equity and well-being. Geographic access to healthcare can be modeled using spatial datasets on local context, together with the distribution of existing health facilities and populations. Several population datasets are currently available, but their impact on accessibility analyses is unknown. In this study, we model the geographic accessibility of public health facilities at 100-meter resolution in sub-Saharan Africa and evaluate six of the most popular gridded population datasets for their impact on coverage statistics at different administrative levels. Methods: Travel time to nearest health facilities was calculated by overlaying health facility coordinates on top of a friction raster accounting for roads, landcover, and physical barriers. We then intersected six different gridded population datasets with our travel time estimates to determine accessibility coverages within various travel time thresholds (i.e., 30, 60, 90, 120, 150, and 180-min). Results: Here we show that differences in accessibility coverage can exceed 70% at the sub-national level, based on a one-hour travel time threshold. The differences are most notable in large and sparsely populated administrative units and dramatically shape patterns of healthcare accessibility at national and sub-national levels. Conclusions: The results of this study show how valuable and critical a comparative analysis between population datasets is for the derivation of coverage statistics that inform local policies and monitor global targets. Large differences exist between the datasets and the results underscore an essential source of uncertainty in accessibility analyses that should be systematically assessed.

Global Harmonization of Urbanization Measures: Proceed with Care.

By 2050, two-thirds of the world's population is expected to be living in cities and towns, a marked increase from today's level of 55 percent. If the general trend is unmistakable, efforts to measure it precisely have been beset with difficulties: the criteria defining urban areas, cities and towns differ from one country to the next and can also change over time for any given country. The past decade has seen great progress toward the long-awaited goal of scientifically comparable urbanization measures, thanks to the combined efforts of multiple disciplines. These efforts have been organized around what is termed the "statistical urbanization" concept, whereby urban areas are defined by population density, contiguity and total population size. Data derived from remote-sensing methods can now supply a variety of spatial proxies for urban areas defined in this way. However, it remains to be understood how such proxies complement, or depart from, meaningful country-specific alternatives. In this paper, we investigate finely resolved population census and satellite-derived data for the United States, Mexico and India, three countries with widely varying conceptions of urban places and long histories of debate and refinement of their national criteria. At the extremes of the urban-rural continuum, we find evidence of generally good agreement between the national and remote sensing-derived measures (albeit with variation by country), but identify significant disagreements in the middle ranges where today's urban policies are often focused.

A century of decoupling size and structure of urban spaces in the United States.

Most cities in the United States of America are thought to have followed similar development trajectories to evolve into their present form. However, data on spatial development of cities are limited prior to 1970. Here we leverage a compilation of high-resolution spatial land use and building data to examine the evolving size and form (shape and structure) of US metropolitan areas since the early twentieth century. Our analysis of building patterns over 100 years reveals strong regularities in the development of the size and density of cities and their surroundings, regardless of timing or location of development. At the same time, we find that trajectories regarding shape and structure are harder to codify and more complex. We conclude that these discrepant developments of urban size- and form-related characteristics are driven, in part, by the long-term decoupling of these two sets of attributes over time.

Combining Remote-Sensing-Derived Data and Historical Maps for Long-Term Back-Casting of Urban Extents.

Spatially explicit, fine-grained datasets describing historical urban extents are rarely available prior to the era of operational remote sensing. However, such data are necessary to better understand long-term urbanization and land development processes and for the assessment of coupled nature-human systems (e.g., the dynamics of the wildland-urban interface). Herein, we propose a framework that jointly uses remote-sensing-derived human settlement data (i.e., the Global Human Settlement Layer, GHSL) and scanned, georeferenced historical maps to automatically generate historical urban extents for the early 20th century. By applying unsupervised color space segmentation to the historical maps, spatially constrained to the urban extents derived from the GHSL, our approach generates historical settlement extents for seamless integration with the multitemporal GHSL. We apply our method to study areas in countries across four continents, and evaluate our approach against historical building density estimates from the Historical Settlement Data Compilation for the US (HISDAC-US), and against urban area estimates from the History Database of the Global Environment (HYDE). Our results achieve Area-under-the-Curve values > 0.9 when comparing to HISDAC-US and are largely in agreement with model-based urban areas from the HYDE database, demonstrating that the integration of remote-sensing-derived observations and historical cartographic data sources opens up new, promising avenues for assessing urbanization and long-term land cover change in countries where historical maps are available.

Fine-grained, spatiotemporal datasets measuring 200 years of land development in the United States.

The collection, processing, and analysis of remote sensing data since the early 1970s has rapidly improved our understanding of change on the Earth's surface. While satellite-based Earth observation has proven to be of vast scientific value, these data are typically confined to recent decades of observation and often lack important thematic detail. Here, we advance in this arena by constructing new spatially explicit settlement data for the United States that extend back to the early 19th century and are consistently enumerated at fine spatial and temporal granularity (i.e. 250m spatial and 5-year temporal resolution). We create these time series using a large, novel building-stock database to extract and map retrospective, fine-grained spatial distributions of built-up properties in the conterminous United States from 1810 to 2015. From our data extraction, we analyse and publish a series of gridded geospatial datasets that enable novel retrospective historical analysis of the built environment at an unprecedented spatial and temporal resolution. The datasets are part of the Historical Settlement Data Compilation for the United States (https://dataverse.harvard.edu/dataverse/hisdacus, last access: 25 January 2021) and are available at https://doi.org/10.7910/DVN/YSWMDR (Uhl and Leyk, 2020a), https://doi.org/10.7910/DVN/SJ213V (Uhl and Leyk, 2020b), and https://doi.org/10.7910/DVN/J6CYUJ (Uhl and Leyk, 2020c).

The tree cover and temperature disparity in US urbanized areas: Quantifying the association with income across 5,723 communities.

Urban tree cover provides benefits to human health and well-being, but previous studies suggest that tree cover is often inequitably distributed. Here, we use National Agriculture Imagery Program digital ortho photographs to survey the tree cover inequality for Census blocks in US large urbanized areas, home to 167 million people across 5,723 municipalities and other Census-designated places. We compared tree cover to summer land surface temperature, as measured using Landsat imagery. In 92% of the urbanized areas surveyed, low-income blocks have less tree cover than high-income blocks. On average, low-income blocks have 15.2% less tree cover and are 1.5°C hotter than high-income blocks. The greatest difference between low- and high-income blocks was found in urbanized areas in the Northeast of the United States, where low-income blocks in some urbanized areas have 30% less tree cover and are 4.0°C hotter. Even after controlling for population density and built-up intensity, the positive association between income and tree cover is significant, as is the positive association between proportion non-Hispanic white and tree cover. We estimate, after controlling for population density, that low-income blocks have 62 million fewer trees than high-income blocks, equal to a compensatory value of $56 billion ($1,349/person). An investment in tree planting and natural regeneration of $17.6 billion would be needed to close the tree cover disparity, benefitting 42 million people in low-income blocks.

Risky Development: Increasing Exposure to Natural Hazards in the United States.

Losses from natural hazards are escalating dramatically, with more properties and critical infrastructure affected each year. Although the magnitude, intensity, and/or frequency of certain hazards has increased, development contributes to this unsustainable trend, as disasters emerge when natural disturbances meet vulnerable assets and populations. To diagnose development patterns leading to increased exposure in the conterminous United States (CONUS), we identified earthquake, flood, hurricane, tornado, and wildfire hazard hotspots, and overlaid them with land use information from the Historical Settlement Data Compilation data set. Our results show that 57% of structures (homes, schools, hospitals, office buildings, etc.) are located in hazard hotspots, which represent only a third of CONUS area, and ∼1.5 million buildings lie in hotspots for two or more hazards. These critical levels of exposure are the legacy of decades of sustained growth and point to our inability, lack of knowledge, or unwillingness to limit development in hazardous zones. Development in these areas is still growing more rapidly than the baseline rates for the nation, portending larger future losses even if the effects of climate change are not considered.