Emerging Methods

Rapid urban growth is one of the major challenges of the 21^st century. Earth is increasingly urbanized, imposing steady pressure upon human-natural urban systems. The forthcoming hyperspectral satellite mission Environment Mapping and Analysis Program (EnMAP) will frequently and globally sample high quality imaging spectrometer data. Novel opportunities for deriving improved descriptions of urban environments will be created. These are relevant for disciplines involved in urban environmental research or urban planning, and include details on the amount and spatial distribution of different impervious, pervious, and vegetation types but also information on the functional land use characteristics. This study demonstrates the potential of the EnMAP mission for providing profound information on urban land use from detailed urban land cover fraction maps. Different hierarchical levels of urban functional areas as depicted by the European Urban Atlas (UA) were used as baseline. This Atlas is part of the local component of the GMES/Copernicus land monitoring services and intends to fill the gap in the knowledge about land use in European cities. In the first step, a simulated EnMAP scene and the combination of support vector regression with synthetically mixed training data were used for quantifying land cover along Berlin’s urban-rural gradient. Results demonstrate the value of EnMAP data for accurately mapping vegetation, impervious and soil surface types (MAE < 12%, R² > 0.8) according to the well-recognized VIS framework. Moreover, EnMAP data allows for extending VIS by the more detailed sub-components roof, pavement, low vegetation and tree (MAE between 7.3% and 21.0%; R² between 0.52 and 0.84). In the second step, urban functional areas were characterized using statistical metrics derived from VIS and extended VIS fraction maps at the urban block scale. Results from a random forest classification clearly reveal the possibility for mapping three different levels of urban functional areas with overall accuracies of 96% (3 UA classes), 79% (5 UA classes) and 60% (9 UA classes). Observed relationships between land cover maps and urban land uses demonstrate the value of EnMAP-based mapping for describing functional characteristics of urban areas. Thus, a significant contribution to the automated derivation of reliable and regularly updated urban surface inventories such as the UA can be made.

The City Biodiversity Index (CBI), also known as the Singapore Index on Cities' Biodiversity, has been specifically developed for monitoring and evaluating biodiversity in cities. In times of an ever increasing population in urban areas - today more than half of the world's population live in cities - this trend poses a range of challenges for cities and their surroundings, not only in terms of resource availability and the quality of urban environments, but also in terms of impacts on biodiversity in cities.

Since the 2010 target to halt biodiversity loss has been collectively missed, the new Aichi biodiversity targets aim to improve the status of biodiversity and to reduce the pressures on biodiversity by 2020. Capturing the status and trends of biodiversity, ecosystems and their services has therefore become a central aspect at local, regional and global scales.

The CBI has been adopted during COP-10 of the CBD in 2008 and is conceived as a self-assessment tool to evaluate the state of biodiversity in cities and to provide insights for improving conservation efforts. This includes an initial baseline measurement, the identification of policy priorities based on their measurements and then a monitoring at periodic intervals.

The CBI includes 23 indicators from three different categories:

Native biodiversity in the city

Ecosystem services provided by biodiversity

Governance and management of biodiversity

During the early implementation of the CBI, it turned out that many cities miss relevant data, personnel and (GIS) skills to assess some of the 23 indicators. The current project addresses this gap and will develop an approach to provide the required information for four indicators which are often difficult for cities to implement:

Indicator 1: Proportion of natural areas in the city

Indicator 2: Connectivity measures or ecological networks to counter fragmentation

Indicator 11: Regulation of quantity of water

Indicator 12: Climate regulation: carbon storage and cooling effect of water

The project partners will use satellite-based data and combine them with appropriate in-situ and ancillary data to produce those indicators. They will be designed in a way to be directly usable by cities to assess their performance regarding the biodiversity targets.

During the demonstration phase the indicator will be developed for the cities of Barcelona, Edmonton and Tallinn. In phase 2, the approach will be rolled out to additional non-European cities, such as Addis Ababa and Durban.

Some of the world’s fastest growing cities are located in the semi-arid West Africa. Driven by rural-to-urban migration, natural growth rates and by restructuring of land zones from rural to urban, urbanization follows irregular patterns including formal and informal processes often causing urban sprawl with associated infrastructural problems. Earth observation (EO) offers the means to retrospectively monitor patterns of urban expansion and has frequently been applied to map urban extent particularly in the Global South. However, for a better understanding of social, demographic and policy drivers of urban growth, maps are needed that go beyond bio-physical properties directly extracted from EO data. Using Landsat TM/ETM+/OLI data we have monitored the urbanization of Burkina Faso’s capital Ouagadougou between 1985 and 2014. In a two-step process we first mapped land cover fractions for six points in time. Subsequently, the temporal pattern of fractions was interpreted together with information from a socio-demographic survey in order to create maps showing different stages of urbanization. The land cover fractions for each time step were derived from multi-seasonal image stacks. Here the often strong differences between dry and wet season often preventing sound analysis were actually utilized to differentiate between, e.g., permanent soil and soil on rain-fed agricultural areas or areas with natural seasonal vegetation. This way, early stages of informal settlement, characterized by often roof-less mud buildings could be identified. Based on multi-temporal information on land cover fractions, the transition of soil to human made materials such as burned bricks could be observed, which typically comes along formalized land rights. We present maps on the spatial-temporal distribution of informal and formal urban areas and show temporal profiles of cover fractions for selected neighborhoods. The applied methodology shows the high potential of high spatial resolution multi-spectral data with high frequent coverage such as Landsat-8 or Sentinel-2. The better 20 m spatial resolution of Sentinel-2 data will allow more reliable mapping of cover fractions. Therefore, an even increased reliability and accuracy of maps on urban coverage can be expected from 2015 onwards.

Stable night lights provide globally consistent proxies for lighted development associated with a variety of human settlement types.  Gridded census enumerations provide geospatial depictions of human population density associated with residential population distributions.  Early work, most notably by Sutton, Elvidge and colleagues, has showed broad spatial correspondence of population density with night light intensity over a wide range of each.  However, the large range of sizes of administrative units used in these early studies leads to widely varying spatial resolution of census enumerations, generally resulting in considerable spatial uncertainty of population and large disparities between night light intensity and interpolated population density.  Recent improvements in both night light imaging and census data resolution have largely resolved this issue and now allow us to compare stable night light and residential population density with much greater spatial correspondence than previously possible.  Quantitative comparison of population density and night light brightness provides a basis for estimation of density/intensity transfer functions in a diverse set of countries with spatially detailed census enumerations.  We use night light brightness from the Visible/Infrared Imaging Radiometer Suite (VIIRS) on the NASA/NOAA Suomi satellite and population density grids from the NASA SEDAC Gridded Population of the World v.4 (GPWv4) product to derive density/intensity transfer functions for a wide range of urban/rural gradients in the USA, Brazil, Sri Lanka, Malawi, South Africa and Portugal.  We find multiple forms of bivariate distribution and transfer functions with broad consistency both within and across countries over a range of intensities of development.  Comparison with Landsat imagery suggests general consistency with land use inferred from land cover.  These results suggest that VIIRS night lights may be used with GPWv4 population density grids in countries with detailed census enumerations to derive more general transfer functions that can be used to produce simple, transparent ambient population products with spatial uncertainty estimates.  Such products could be used for testing a wide range of environmental and socioeconomic scenarios in which a transparent depiction of ambient population density is required.