April 24-26, 2019 | Bern, Switzerland
Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
105R: Designing sustainable urban land systems + Mapping, analysing and modeling human settlements at global scale
What are the visions for the planetary land system?
Land systems scientists have demonstrated that many current patterns of urbanization are degrading the environment, driving social inequity, and suggest that alternative approaches are imperative to improve the long-term sustainability of cities. The need to integrate science and design has been formally recognized in what some have labelled “urbanization science.” Planners and architects have proposed alternative patterns and processes of development that are intended to enhance the sustainability of cities and have begun to put these ideas into action in built sites that constitute experiments in sustainable design. Research has built a greater understanding of the functioning of urban social-ecological systems (SES) and is feeding into the production of tools for supporting the needs of urban land-use planners and decision makers. Despite these advances and mutual interest in designing more sustainable urban land systems, the engagement between SES and design-oriented communities has been slow to emerge, inhibiting potentially transformative knowledge exchange. A key goal of this session is to bring scholars and practitioners from planning, architecture, design, and environmental engineering together with ecologists, economists, and sociologists to identify cutting-edge approaches and potential new points of synthesis between land systems science and design. A second goal of this session is to begin to forge overlap between the institutions that support land systems science and the design fields. Specific questions this session will address include: What are the challenges to bridging empirical and design-oriented approaches? What theoretical frameworks, data, tools, and methods have enabled synthesis between SES and design-fields? Are there approaches to synthesis that could be potentially helpful but have not yet been fully explored? How do the challenges and solutions related to synthesizing design and science vary based on regional context internationally? This session will include a series of short "flash" talks and leave ample time for discussion related to geospatial and field-based data and analysis for evaluating outcomes of design interventions; enhancing institutional fit to enable cities to plan, build, and manage for long-term sustainability; expanding stakeholder engagement to enhance outcomes, especially marginalized communities and private sector actors; and means for forging pathways for information sharing and collaboration between between science and design communities and on-the-ground urban decision makers. ******* Recent studies have estimated that in the upcoming 30 years the land surface occupied by human settlements (HS) on Earth is going to double. This urbanization tsunami is expected to hit current megalopolis, but also to affect medium- and small-size settlements (including rural ones) thus playing a major role in driving global land-systems. In such framework, a precise and quantitative understanding of the entire spectrum of HS sizes is a fundamental requirement to implement any regulatory strategy for managing and containing future urbanization. Nevertheless, so far this has been precluded by the lack of reliable and detailed global inventories of HS, hence forcing the scientific community to mostly focus on large urban agglomerations with a consequent bias on the resulting analyses and modeling studies. To overcome such drawback, in this session we present a novel pipeline for a global and comprehensive analysis of HS based on the use of big Earth Observation satellite data and advanced modelling techniques. Specifically, first we introduce Google Earth Engine (GEE), the most advanced cloud-based geospatial processing platform which combines a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities. Next, we present the Word Settlements Footprint (WSF), a novel dataset generated by means of GEE which outlines HS globally over time with unprecedented accuracy and reliability. Finally, we discuss the output of a global spatial analysis on HS carried out by exploiting the WSF with focus on quantitative comparisons and explicit spatial urban modeling.
|External Resource: - SESSION RECORDING - https://youtu.be/lX4MdchzHB8|
ID: 596 / 105R: 1
105R Designing sustainable urban land systems
Keywords: Parcel Landscaping, Pattern Analysis, Google Earth Engine, Sustainable Urban Design
Parcel-level landscaping classification and spatial pattern analysis using Google Earth Engine
School of Geographical Sciences and Urban Planning, Arizona State University, United States of America
Residential landscaping choices heavily affect municipal water use and urban sustainability in the U.S. southwestern cities. Specifically, water use for irrigation accounts for more than half of the residential demand in the semi-arid Phoenix metropolitan region. Typical residential landscaping include mesic, oasis, and xeric types, ranging from a mixture of trees with abundant turf grass to sparse tree and bush cover. The distinctions lead to dramatic variations in water use and micro-climate conditions during summer, as well as potentially affecting other urban environmental services.
Spatial explicit analysis of the heterogeneous landscaping distribution across the metropolitan region is a prerequisite for outdoor water use assessment. In this study, we integrated the 1-m NAIP imagery and Lidar DSM data to classify the parcel landscaping types using the Google Earth Engine platform (GEE) and machining learning techniques. The results demonstrate the effect of social norms and homeowner associations on neighborhood landscaping patterns and the variations within neighborhoods. Facilitated by the massive computational capacities of GEE, our fine-scale metropolitan-extent analysis connects urban land system with household level land management decisions, strategically supports outdoor water conservation planning for drought response programs, and can be easily applied to other cities in the American Southwest.
ID: 307 / 105R: 2
105R Designing sustainable urban land systems
Keywords: design, urbanization, tropical cities, point clouds, ecosystem services
Integrated science-design loop for fostering urban ecosystem services in a tropical city
1ETH Zürich, Switzerland; 2ETH-SEC, Future Cities Lab, Singapore
Current urbanization processes profoundly disrupt the environment but generate many opportunities to create more liveable, healthy and resilience environments. With many of the world’s fastest growing cities located within the tropics and their high vulnerability to climate change, they call for integrated solutions linking urban and ecological functionalities.
While design is increasingly being recognized as common ground to bring scientific knowledge into decision making, it has mostly been used to create place specific responses expressing particular values until now, rather than framing the natural and physical sciences to become more salient and find legitimation. New technological achievements in data acquisition and processing, as well as simulating and visualizing landscapes are however opening new ways to foster iterative feedback loops between data obtained from the environment and the process of designing and planning. For example, new sensing technologies, such as LIDAR technologies, can on one hand be used by architects to design urban landscapes. On the other hand, engineers and scientists can process the same data to quantify ecological processes.
This contribution illustrates an integrated science-design approach in an urban experiment in the frame of the Natural Capital Singapore. We demonstrate how an iterative loop between landscape design and various eco-hydrological and radiative transfer models used to calculate ecosystem services provision can help incorporate scientific knowledge into the design of a neighborhood in Singapore. We then discuss to what extent this approach can create a knowledge-driven rather than a data-driven design and planning process and conclude on the use of point cloud data as a new language to integrate land system science and design.
ID: 334 / 105R: 3
109R Mapping, analysing and modeling human settlements at global scale
Keywords: Google Earth Engine
Exploiting global remote sensing data with Google Earth Engine
Google, United States of America
This contribution will introduce Google Earth Engine, a cloud-based geospatial processing platform for analyzing satellite imagery and other geospatial datasets at global scale. The background, data catalog, processing capabilitites and API of Earth Engine will be described. Relevant publications and web applications powered by Earth Engine will be reviewed. Case studies will be presented that demonstrate the use of Earth Engine and convolutional neural networks for global mapping.
ID: 605 / 105R: 4
109R Mapping, analysing and modeling human settlements at global scale
Keywords: Global Urbanization, Remote Sensing, Artificial Intelligence, Settlemnt Growth
Monitoring global urbanization using big earth data - the world settlement footprint
1German Aerospace Center - DLR, Germany; 2Google, Inc., Zürich, Switzerland; 3MindEarth, Biel / Bienne, Switzerland
Reliably monitoring global urbanization is of key importance to accurately estimate the distribution of the continually expanding population, along with its effects on the use of resources (e.g., soil, energy, water), infrastructure and transport needs, socioeconomic development, human health, food security, etc. To this purpose, since the last decade several global maps outlining settlements have started being produced by means of satellite imagery. In this framework, the two currently most largely employed are JRC’s Global Human Settlement Layer (GHSL) derived at 30m spatial resolution from Landsat optical data and, especially, DLR’s Global Urban footprint (GUF) derived at 12m spatial resolution from TanDEM-X/TerraSAR-X radar data. However, it is worth noting that, despite generally accurate, these layers still exhibit both some over- and underestimation issues. Specifically, this is mostly due to the fact that they have been generated by means of: i) single-date scenes (which can be strongly affected by the specific acquisition conditions) and ii) solely using either optical or radar data, which are sensible to different structures on the ground (e.g., with optical imagery bare soil and sand generally tend to be misclassified as settlements, while this does not occur with radar data; on the contrary, with radar imagery complex topography areas or forested regions can be wrongly categorized as settlements, whereas this does generally not happen if optical data are employed).
In order to overcome these limitations, we have developed a novel methodology that jointly exploits for the first time mass multitemporal optical and radar data. In particular, the rationale is that the temporal dynamics of human settlements over time are sensibly different than those of all other information classes. Hence, given all the multitemporal images available over a region of interest in the selected time interval, we first extract key temporal statistics (i.e., temporal mean, minimum, maximum, etc.) of: i) the original backscattering value in the case of radar data; and ii) different spectral indices (e.g., vegetation index, built-up index, etc.) derived after performing cloud masking in the case of optical imagery. Then, different classification schemes based on advanced machine learning are separately applied to the optical and radar temporal features, respectively, and, finally, the two outputs are properly combined together.
In the light of its great robustness, the method has been recently used for generating the new World Settlement Footprint (WSF) 2015, i.e. a 10m resolution binary mask outlining the extent of human settlements globally derived by means of 2014-2015 multitemporal Landsat-8 and Sentinel-1 imagery (of which ~420,000 and ~250,000 scenes have been processed, respectively). Furthermore, to quantitatively assess the high accuracy and reliability of the layer we have carried out in collaboration with Google an unprecedented validation exercise based on a huge amount of ground-truth samples (i.e. 900,000) labelled by crow-sourcing photointerpretation. In particular, a statistically robust and transparent protocol has been defined following the state-of-the-art practices currently recommended in the literature. Overall, results assess the great effectiveness of the WSF2015, which also outperforms all other currently existing similar global layers including the GUF and the GHSL.
Nevertheless, to properly analyze and understand the complexity of human settlements and ensure their sustainable development, not only information about the current extent is sufficient. Rather, outlining their growth over time is also fundamental for modelling ongoing trends and implementing dedicated suitable planning strategies. However, so far the existing products are mostly available for few time steps in the past and their quality– yet by simple qualitative assessment against e.g. Google Earth historical imagery – appears rather poor.
To address this issue, we designed and implemented a novel iterative technique for outlining the past settlement extent from Landsat multitemporal imagery available from late 1984 (indeed no comparable radar dataset is continuously and globally consistently available over the last 30 years). First, under the assumption that pixels categorized as non-settlement at a later time cannot be marked as settlement at an earlier time, all areas excluded from the WSF2015 are discarded a priori from the analysis. Next, for each target year in the past all available Landsat scenes acquired over the investigated area of interest are gathered and cloud masking is performed. Key temporal statistics (i.e., temporal mean, minimum, maximum, etc.) are then extracted for different spectral indices including the normalized difference built-up index (NDBI), the normalized difference vegetation index (NDVI) and the modified normalized difference water index (MNDWI). Going backwards in time, training samples for the given target year are iteratively extracted by applying morphological filtering to the settlement mask derived for the previous time step, as well as excluding potentially mislabeled samples by adaptive thresholding on the temporal mean NDBI, MNDWI and NDVI. Finally, random forest classification in performed.
Extensive experimental analyses over several test sites assessed the great effectiveness and robustness of the methodology. Accordingly, it is currently being employed within the Google Earth Engine environment for generating the WSF Evolution, i.e. a novel dataset aimed at outlining the growth of settlement extent globally at 30m spatial resolution and high temporal resolution (i.e., 5-year or even finer) from 1985 to 2015. The WSF Evolution is envisaged to be completed in the next few months and it is expected to become a revolutionary product in support to a variety of end users in the framework of several thematic applications, helping to understand, as never before, how settlements evolved over three decades while capturing specific temporal trends.
Contact and Legal Notice · Contact Address: email@example.com
Privacy Statement · Conference: OSM 2019
|Conference Software - ConfTool Pro 2.6.131+TC+CC
© 2001 - 2020 by Dr. H. Weinreich, Hamburg, Germany