A 1.375Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
We have recently successfully used LiDAR data to assess the green roof potential during our work in the project of TURaS. TURaS (Transitioning towards Urban Resilience and Sustainability) is funded by the Seventh Framework Programme of European Union. Our work in this project focuses on green roof development in London. This application has been evolved to a straightforward and effective approach which can be used in green infrastructure development around urban area.
Green roofs are important for resilient urban communities as they assist runoff attenuation, promote evapotranspiration, help improve air quality, result in energy savings, and provide recreational spaces. There has been very little work on rapid appraisal of green roof potential which this study addresses using LiDAR data. The London case study shows that LiDAR can cost-effectively classify roof geometry for large areas according to green roof design criteria as an input to the planning process.
Methodology
Building heights are extracted from DSM (Digital Surface Model) of LiDAR data. The LiDAR data used here is provided by UK Environment Agency (EA) Geomatics Group. EA LIDAR data are all with a vertical accuracy in the range of 5cm to 15cm. Spatial resolutions range from 50cm to 2 meters. Building features are adopted from the topography layer of MasterMap which is a digital map product of UK Ordnance Survey (OS). Overall positional accuracy of the topography layer is 1.0 meter.
Extraction of Building Height
The procedure for building height extraction is introduced here with the example at Ravenscourt Park of Hammersmith, West London. In EA LiDAR data, DSM includes height information of all features while DTM only represents the bare ground. Subtracting DSM by DTM will extract height information of buildings. However it will also extract heights of bridge, flyover, rail way and even water depth of river at the same time. It is difficult to separate height information of building from other features.
Building polygon from the topography layer of OS MasterMap is defined as outline of permanent roofedconstruction. Although there is an ESRI GIS command of cutting DEM by polygon, it might introduce errors when DSM is cut by a number of separate polygons with complex shapes.
This study proposes to extract building height by a series of spatial operations in GIS. Firstly, building polygons will be rasterised with the same resolution as the targeted DSM tiles. Then, rasterised building polygons and the targeted DSM tile will be added up. Consequently, DSM will be cut by building polygon and height information of building will be extracted. When the study area covers a wider region, large data volume may lead to heavy computation load. In such cases, sampling technique will be used to select representative areas based on land-use variables.
Computation of Roof Slope
With reference to structure requirements of green roofing, roof slope must be taken into account. Roofs with slope of 20 degree or less will not require special structure measures. With a pitch in excess of 20 degree, structural anti-shear protection is needed. Once the angle exceeds 30 degree, a separate set of construction calculations will be needed. German standard allows greening on the pitch roofs up to 45 degrees.
Roof slope is computed from the preprocessed DSM with GIS functionality. By querying the DSM cut by building polygons, area of roofs with different slopes can be obtained. In this example of Ravenscourt Park, area of roofs with slope no more than 20 degree is 115492 square meters, area of roofs in excess of 20 degree and up to 30 degree is 42304 square meters, area of roofs in excess of 30 degree and up to 45 degree is 38460 square meters.
Sensitivity Test of Resolution
The sensitivity of roof area estimate to DSM resolution has been initially tested on a green roof at bus depot in East London. From test results, roof area estimate with slope no more than 20 degree is not sensitive to DSM resolution. Roof area estimate with slope between 20 and 30 degree is sensitive to DSM where DSM resolution of 1 meter is still not sufficient to separate roofs with slope of 2030 degree and slope of 3045 degree. Roof area estimate with slope between 30 and 45 degree is also sensitive to DSM resolution.
Case Study in Central London and South Bank
This case study area covers Central London and South Bank, where there are a number of government departments, commercial streets and business districts. In contrast, there are less residential buildings compared with the rest of London. The focus of this case study is therefore on roofs of non-domestic building, which are not likely to exceed 30 degree of slope. DSM with 2 meter resolution is then adopted according to the early sensitivity test. Overall, there is a large proportion (38%) of roofs with gentle slope (<= 20 degree) in this case study area, particularly at south of the river (43%).
Case Study in London Borough Newham
London Borough Newham is a typical urban district in UK. It has town centres, business areas, retail parks, industry areas, large residential areas as well as City Airport and Olympic Park / Village. Because there are both large number of residential buildings and commercial / public buildings in Newham, areas are estimated for all potential green roofs with slope from 0 degree to 45 degree. Based on the early sensitivity test, DSM with 0.5 meter resolution is chosen.
Admin geography of Newham is about 39 square km. It needs more than 100k building polygons and nearly 180 DSM tiles (0.5m resolution) to cover the whole borough. In order to control the computation load, the method of clustering and sampling is designed here. Firstly, hierarchical clustering technique is deployed to classify Newham into six classes of area. The clustering analysis takes into account land-use variables of domestic building, non-domestic building, domestic garden, green space, road, rail and water. Secondly, samples of DSM tiles are selected to represent each class with local knowledge and map study.
Class 1: Residential area with park / green space
Class 2: Town centre, Olympic Park, industrial areas, business areas, City Airport and retail parks.
Class 3: Residential area mixed with commercial buildings
Class 4: Residential area of estate
Class 5: Residential area with mix of estate and terrace houses
Class 6: Residential area, Victoria streets with mainly terrace houses
With our approach, potential green roofs are identified and roof areas / percentages are computed by slope. It is interesting to see that there is high percentage (66%) of roof with gentle slope (<= 20 degree) in areas of class 2. In contrast, there is high percentage (38%) of chalet roof (3045 degree) in areas of class 6. At one Victoria style residential area, there are up to 41% of chalet roofs.
Conclusion
An innovative approach was developed to rapidly estimate areas of potential green roofing. It is based on the computation of roof slope by LIDAR data and topographic data. It is a straightforward and effective approach to support initial planning, strategic assessment and impact study of green roof development. Most importantly, this new approach is generally applicable in UK with easy data access while it is also likely to be feasible in other EU countries as similar data are available.
Dr. Yang Li is the senior research fellow in Centre for Geo-Information Studies, University of East London. He is also a fellow of UK Royal Geographical Society.
Professor Allan J. Brimicombe is the head of Centre for Geo-Information Studies, University of East London. He is also an academician in UK Academy of Social Sciences, fellow of UK Royal Geographical Society and fellow of UK Royal Statistical Society.
A 1.375Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE