Chapter V 
Mapping urban areas using airborne hyperspectral remote sensing data

↓84

submitted manuscript

Abstract

↓85

Urbanization significantly influences ecosystem goods and services in various ways. Accordingly, reliable and spatially explicit information on urban land cover is required for analysis of ecological conditions and urban environmental modeling, e.g. in order to correlate patterns of impervious surfaces with information on urban climate or habitats. The quality of maps from recently available, high spatial resolution remote sensing data, however, often suffers from spectral ambiguity and inaccurate representation of the complex geometric composition of urban surfaces. This study uses hyperspectral airborne line scanner data from the city of Berlin, Germany, to map a heterogeneous urban environment. The data is characterized by high spatial and spectral resolution. Diverse accuracy assessments are performed to identify sources of inaccuracy, to quantify them and to investigate to what extent remote sensing data can function as a basis for detailed urban environmental analyses. Results show that inaccuracies of the final land cover and impervious surface maps can mainly be attributed to the influence of displaced buildings occluding surfaces as a function of view-angle and to tree crowns obscuring impervious areas underneath. Despite a possible improvement of results by precise digital surface models or cadastral information, some inaccuracies will remain. The described problems occur in data from all high resolution sensors, especially at large view-angles. Possible consequences depend on the scale of subsequent analysis or on a possible combination with additional data sources.

Chapter V:1 Introduction

Local, regional, and global ecosystems and the goods and services they provide are significantly influenced by urbanization (Alberti et al., 2003). Urban development and the inherent change in land use negatively impact natural habitats and taxonomic richness (e.g.Blair, 1996;Morse et al., 2003). They influence the microclimate as well as energy fluxes and air-flow, which again lead to phenomena like the urban heat island (UHI) and increased convective rainfall (e.g.Carlson and Arthur, 2000;Collier, 2006). Their positive correlation to changes in the hydrological system and pollution load in run-off or in rivers has been reported (Booth et al., 2004;Hatt et al., 2004).

Impervious surface coverage was identified as a key indicator in this context (Schueler, 1994;Arnold and Gibbons, 1996). Thresholds for total impervious area (TIA) in a watershed can be related to different health states of the receiving stream (Arnold and Gibbons, 1996). However, such aggregated measures are not sufficient to describe biological conditions in a watershed or to function as input for environmental models; instead, detailed information on the type, density, configuration and connectivity of impervious surfaces is needed (Brabec et al., 2002;Booth et al., 2004;Carle et al., 2005;Alberti et al., 2007). For example, a strong relationship between UHI intensity and patterns of impervious areas (Bottyan et al., 2005) or between the benthic index of biological integrity and landscape variables such as mean patch size and number of road crossings (Alberti et al., 2007) have been documented. Spatially explicit land cover information has been used for describing urban environmental quality (Nichol and Wong, 2005) or modeling such environmental indicators as surface temperature, run-off or biodiversity measures (e.g.Pauleit et al., 2005).

↓86

Besides information on spatial patterns of impervious land cover, it is relevant for urban planning to delineate, for example, built-up areas and open spaces (Pauleit and Duhme, 2000;Nichol and Wong, 2005). The density of built-up areas influences people's health and comfort (Svensson and Eliasson, 2002); such information is usually not included in standard cadastre data. Detailed information on the geometry and distribution of buildings is needed in urban climatology since UHI and convective rainfall can be connected to urban morphology (Collier, 2006).

In general, information on land cover can serve as an independent variable for the derivation of land use or other functional variables for ecological applications (Cadenasso et al., 2007). Remote sensing has proved useful to provide spatially explicit information for urban land cover analyses. Aerial photography, both analog and digital, has been successfully used in this context (e.g.Pauleit et al., 2005;Cadenasso et al., 2007). At a spatial resolution of about 0.25 m and below, it allows for a detailed mapping of urban land cover and use. It is limited, however, by its little spectral information content and a high degree of user interaction required for surface delineation. Recently, multispectral airborne line scanners like HRSC-AX or ADS40 have been introduced (e.g.Ehlers et al., 2006). Images from such instruments yield very high spatial resolutions below 0.2 m and allow fully digital semi-automated processing. Spaceborne multispectral data at a spatial resolution of 4 m and below has become available with the launch of Ikonos and Quickbird-2 in 1999 and 2001 respectively. Ever since, the number of more detailed urban remote sensing analyses has increased (e.g.Small, 2003;Nichol and Wong, 2005;Thanapura et al., 2007) and detailed assessments like private garden mapping (Mathieu et al., 2007) are now feasible. Beforehand, multispectral moderate resolution satellite imagery was instead used to describe urbanization processes at larger scales and aggregated densities of impervious surface coverage (e.g.Ward et al., 2000;Lu and Weng, 2006). The spatial resolution of data from airborne hyperspectral scanners is similar to that of satellites like Quickbird. Its very high spectral resolution recommends hyperspectral imagery for urban analysis because the spectral similarity of anthropogenic surface materials in urban environments does not allow to directly map land cover from multispectral data (Herold et al., 2003).

Regardless of spectral characteristics, high and very high spatial resolution imagery reveals general drawbacks of remote sensing based urban analyses that are caused by the exhibited central perspective sensor view:

↓87

(1) 3-D-objects will be displaced at large sensor view-angles. This phenomenon can be observed in large field-of-view (FOV) airborne line scanner data and in aerial photographs, but similarly in data from high resolution spaceborne instruments like Quickbird or Ikonos that are often acquired with off-nadir view to allow more frequent acquisition. With increasing distance from nadir, roof-tops of buildings are displaced and façades will appear at the position of the buildings' ground plots. The area behind the 3-D-objects is occluded. This slant projection of buildings can lead to significant misestimates of impervious land cover during image analysis (Pauleit and Duhme, 2000).

(2) urban streets might be covered by the crowns of trees on sidewalks. This effect generally applies to remote sensing based approaches. In the case of impervious surface mapping, the sensor view will heavily overestimate vegetated surfaces.

Regardless of the final scale of analysis and focus of an application, a detailed description of three major elements of urban heterogeneity – vegetation, built structures and surface materials – is required to facilitate understanding relationships in the coupled human-natural urban system (Cadenasso et al., 2007). This study explores the potential of airborne hyperspectral remote sensing data to provide reliable information on urban land cover and impervious surface coverage for a spatially explicit analysis of the urban environment. Given the high spectral resolution of the data most inaccuracies during the mapping are expected to relate to problems that results from the complex geometrical composition and differences between sensor view and true ground cover of urban areas. This way, we focus on phenomena that might exist in any high spatial resolution remote sensing data used for detailed urban analysis. We quantify inaccuracies based on a variety of reference data to identify potential sources of error. Finally, we discuss to what extent maps based on hyperspectral imagery can provide reliable information for ecological analyses and environmental modeling of urban areas.

Chapter V:2 Conceptual framework

↓88

Urban land cover classifications that delineate built-up and non built-up impervious areas are challenging. The two spectrally heterogeneous classes include surfaces with diverse spectral properties, while spectrally similar materials exist in both classes (Herold et al., 2003). In addition, inorganic soils, i.e. pervious surfaces, can appear very similar to impervious materials like concrete.

When the spectral differentiation of land cover is successful during image classification, the complicated geometrical composition of urban areas needs to be coped with to achieve an accurate map that can be linked with other information for subsequent analyses. This is ideally achieved by incorporating accurate 3-D-models in the orthorectification process. Such digital surface models (DSM) are more frequent nowadays due to the increasing availability of laser scanning systems. It will take several more years, however, until detailed and up-to-date 3-D information exists for many urban areas world-wide. When no DSM is available, displaced buildings will - as a function of their height and the sensor's view-angle - impact the accuracy of results. In our case, no DSM was available. Even with a DSM, occluded surfaces due to an oblique view cannot be reconstructed in single orthophotos (Zhou et al., 2005). Also, the phenomenon of trees obscuring ground cover underneath remains and its impacts can hardly be predicted. Despite the importance of information on tree cover itself, several ecological assessments require knowledge on the underlying surface, e.g. for hydrological modeling.

This study is organized around three questions for the detailed assessment of possible inaccuracies of urban land cover maps generated from airborne hyperspectral data:

↓89

  1. Does airborne hyperspectral remote sensing data provide the spectral information needed to reliably delineate urban land cover?
  2. How spatially accurate are maps from airborne line scanner data in urban areas?
  3. How great is the impact of tree crowns obscuring impervious surface underneath?

To answer these questions the following steps are performed:

  1. a heterogeneous urban environment is classified with a state-of-the-art classifier. The land cover classes vegetation, buildings, non built-up impervious area (paved), soil, and water are delineated;
  2. a map on the impervious surface coverage is derived from the orthorectified land cover map;
  3. different accuracy assessments are performed that explicitly address the individual sources of inaccuracy. The quality of the image processing as a whole is assessed and drawbacks of individual processing steps are identified.

↓90

By performing the study in a well-known urban environment with diverse reference data sets available, a thorough assessment and quantification of inaccuracies is possible. This way, important insights for similar studies in less known areas of the world can be derived.

Chapter V:3 Airborne hyperspectral remote sensing of urban areas

Hyperspectral remote sensing data, also referred to as “imaging spectrometry data”, is characterized by its very high spectral information content (Goetz et al., 1985). For each pixel in the image a quasi-continuous spectrum exists which represents the measured reflected sunlight in the visible (VIS), near-infrared (NIR) and short wave-infrared wavelength regions (SWIR). Whereas the bands in multispectral imagery cover rather wide wavelength regions (e.g. 60-250 nm in the case of Landsat Thematic Mapper), the bands in hyperspectral data are narrow (e.g. 10-15 nm). The resulting spectra enable, for example, analyzing narrow absorption features (e.g.McMorrow et al., 2004) (Fig. V-1) and the quantification of ecological variables (Ustin et al., 2004).

Figure V-1: Reflectance spectra from the airborne Hyperspectral Mapper (HyMap) for different surfaces. Gaps are due to atmospheric absorption.

↓91

Airborne hyperspectral remote sensing was first introduced in the mid-1980s and emerged with the availability of data from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) (Vane et al., 1993). First used in geological applications (e.g.Kruse et al., 1993), hyperspectral data was introduced to other environmental applications such as green and non-photosynthetic vegetation mapping (Roberts et al., 1993), biophysical modeling (Jacquemoud et al., 1995) or wildfire mapping (Dennison, 2006). Nowadays, data from several airborne imaging spectrometers is used in various contexts, for example data from the Digital Airborne Imaging Spectrometer (DAIS7915) (e.g.Roessner et al., 2001), Hyperspectral Mapper (HyMap) (e.g.Schlerf et al., 2005) or the Compact Airborne Spectrographic Imager (CASI) (e.g.Wang et al., 2007). The quality of the data is determined by the sensor's spectral characteristics and its signal-to-noise ratio (SNR). The number of instruments is increasing and hence more hyperspectral data will be available for research and application.

Whereas the high spectral information content of airborne hyperspectral data is very beneficial for many applications, their pre-processing and analysis are more complicated than traditional spaceborne multispectral data. This is caused by data inherent phenomena like the presence of water vapor absorption features in the spectrum or sensor characteristics such as a wide field-of-view (FOV) (Fig. V-2). This FOV and hence large view-angles, θ v , towards the edges of the flight line are necessary to cover large areas at low operating altitudes but they enhance the differences between sensor view and true ground cover or between sun-facing and shaded façades. The problem of correcting atmospheric effects and reflectance anisotropy is solved for the urban environment to a satisfying extent (Richter and Schläpfer, 2002;Schiefer et al., 2006). The orthorectification of the data can achieve sub-pixel accuracy, but results depend heavily on the quality of the available digital elevation model (DEM) or ideally DSM (Schläpfer and Richter, 2002).

Figure V-2: Image acquisition by the large FOV airborne line scanner HyMap in urban areas. Façades appear in the image at large view-angles and are differently illuminated.

Chapter V:4 Material and methods 

Chapter V:4.1 Study area

↓92

We used data from the metropolitan area of Berlin, Germany, in this work. The history and structure of the city make it an ideal study area for the addressed research questions. The rise of the Prussian empire and the era of industrialization dominate the original structure of the city. After heavy destruction during World War II, a period of separation and parallel development under opposite political systems led to diverse new urban structures. Following the fall of the Berlin Wall and the closing down of industrial complexes from socialist times, many derelict sites exist and Berlin has recently been experiencing large scale development in very central areas (e.g. Postdamer Platz). The study area hence covers a great variety of urban structure types. This situation is unique for a metropolitan area in the western hemisphere and the three research questions can be thoroughly investigated: the spectral variety of roofing materials, for example, is high; the height and spatial composition of buildings is diverse; in the very most cases trees exist along streets. (Sukopp, 1990;Balder et al., 1997;SenStadt, 2007)

The 32.5 by 2.2 km study area is outlined by the extent of the available hyperspectral data set and covers a representative gradient from the western urban fringe through the city center to the eastern municipal boundary (Fig. V-3). It includes:

  1. the central business and governmental district with large administrative buildings, wide open spaces, historical boulevards or newly constructed shopping and transportation areas (Fig. V-4.a,b);
  2. residential areas of different densities and from different development periods
    (Fig. V-4.c-e);
  3. pre-cast apartment complexes and wide boulevards in former East Berlin (Fig. V-4.f);
  4. parks and recreational areas (Fig. V-4.g);
  5. private garden areas (Fig. V-4.h);
  6. industrial grounds (Fig. V-4.i);
  7. derelict land (Fig. V-4.j);
  8. suburban areas (Fig. V-4.k);
  9. inner-urban forests, water bodies and rivers, and small agricultural patches towards the end of the flight line (Fig. V-4.l).

↓93

Figure V-3: Study area and municipal boundary of Berlin. The outlines of the study area are determined by the extent of the airborne image data set. Image data is shown after preprocessing in false-color composite (R = 829 nm; G = 1648 nm; B = 662 nm).

Figure V-4: Examples of different urban structure types in the hyperspectral data set (R = 829 nm; G = 1648 nm; B = 662 nm). Details see text.

Chapter V:4.2 Image data

The airborne imaging spectrometer HyMap acquires data between 0.4 and 2.5 µm in 128 spectral bands (see eample spectra in Fig. V-1) with an average bandwidth of 10 to 15 nm. The sensor’s FOV is 61.3°. The 7277 by 512 pixel raw image was originally registered at a spatial resolution of 3.9 by 4.5 m at nadir.

↓94

The hyperspectral image used in this work was acquired on 20 June, 2005 around 10.46 AM local time. The data set was corrected for atmospheric effects and converted to surface reflectance (Richter and Schläpfer, 2002). View-angle dependent brightness gradients were eliminated (Schiefer et al., 2006, Chapter II of this work). The number of bands was reduced to 114 based on the signal-to-noise ratio. After image classification the resulting maps were orthorectified and resampled to a pixel size of 3.5 m (Schläpfer and Richter, 2002). A raster DEM derived from the contour lines of the official digital map was resampled from its original spatial and vertical resolution of 25 m and 0.1 m, respectively, to 3.5 m spatial resolution for the correction. A DSM was not available. The accuracy assessment of the orthorectification itself showed a root mean squared error (RMSE) of 2.9 and 3.1 m in easting and northing, respectively, that is explained by the missing accuracy of the DEM.

Chapter V:4.3 Land cover classification

The land cover classification was performed using support vector machines (SVM). SVM have recently been experiencing increased attention. They outperformed other classifiers in several studies, they require a relatively small number of training samples, and are insensitive to high dimensional data (e.g.Huang et al., 2002;Pal and Mather, 2006). SVM are a supervised classification algorithm that delineates classes by fitting a separating hyperplane in the spectral feature space (Burges, 1998). SVM can handle complex class distributions, including multi-modal classes, i.e. classes that contain a variety of materials with different spectral properties. Spectrally heterogeneous classes are difficult to assess with traditional parametric classifiers (Seto and Liu, 2003). For our study, the software imageSVM was used (Janz and van der Linden, 2007).

The training samples were acquired by a clustered sampling strategy: at first, 64 seed pixels were randomly drawn from the full image. 29 pixels around each of these seeds (a 5 x 5 window plus the four outer diagonal pixels) were then assigned to one of the five classes. A small number of additional seed pixels were interactively placed on rare but characteristic surfaces. All pixels were labeled based on very high resolution aerial photographs in Google Earth. Altogether, 2133 training samples were used for the SVM classification.

Chapter V:4.4 Reference data and accuracy assessment

↓95

A detailed accuracy analysis according to the research questions in this work requires reference data of different characteristics and from multiple sources. Digital aerial photographs and information from the Urban Environmental Information System (UEIS) and cadastre were available from the municipal administration. In addition, a detailed field survey was performed synchronously to the acquisition of the image data. Based on these different sources of reference data, reference products were generated for the assessment of the individual processing steps during image analysis (Fig. V-5). The different assessments are then discussed in the context of the three research questions.

Figure V-5: Hyperspectral image analysis steps, reference products and the data sets they are based on. The different reference products were derived from orthophotos, digital cadastral information and field surveys. They are used to assess different steps of image analysis (dotted lines). Italic numbers indicate, which research question is addressed by the corresponding assessment.

Orthophotos

↓96

A set 0.25 m resolution digital color aerial photographs was available through the Berlin Department for Urban Development. The data set covers the entire area of Berlin and was taken in August 2004. In addition, Google Earth provides aerial photographs of slightly higher spatial resolution for the city.

To assess how well urban land cover classes can be mapped based on hyperspectral information, a stratified set of independent reference pixels was selected from the image before orthorectification. These were used to derive a confusion matrix for the land cover map and to calculate the overall accuracy, kappa value (κ), and producer's and user's accuracy (Congalton and Green, 1999). At first, rectangular polygons of about 200 by 300 pixels were manually drawn on areas of six typical urban structure types to account for the heterogeneity of the urban environment. These included: the central business and governmental district (in the following named center); dense residential areas with attached buildings and narrow courtyards (dense); open residential areas with single houses and gardens (single); pre-cast apartment complexes surrounded by recreational areas (complexes); individual houses surrounded by agricultural patches and forest along the urban-suburban fringe (suburban); industrial and commercial grounds (industrial). About 150 reference pixels were randomly drawn for each urban structure type (Table V-1). To better investigate the classification quality in dark areas (dark), i.e. areas with low contrast like water or shaded surfaces, 158 extra points were randomly selected using a dark area mask (reflectance at 1.650 µm < 5%). 197 pixels were randomly selected from the rest of the image (rest), to represent remaining areas. Altogether, 1253 reference pixels in the image were assigned to one of the five classes based on their spectral properties and contextual information from the aerial photographs.

Table V-1: Reference pixels of five land cover classes as distributed in urban structure types.

Class

Reference pixels randomly selected from

Total

center

dense

single

complexes

suburban

industrial

dark

rest

vegetation

28

52

91

77

112

42

55

108

565

buildings

58

41

36

23

5

29

-

32

224

paved

57

54

18

40

2

62

42

34

309

soil

1

4

4

7

26

18

-

12

72

water

4

-

-

2

5

-

61

11

83

Total

148

151

149

149

150

151

158

197

1253

↓97

Digital environmental and cadastral information

The Berlin Department for Urban Development provides digital information based on a digital map at the scale of 1:5000 in the UEIS (SenStadt, 2007). The polygons in the UEIS represent structural units that refer to transportation areas, residential blocks or water bodies for the year 2005. In addition, the spatial extent of buildings was available through the city's digital cadastre. Altogether, 550,000 buildings exist in this second database. It is very accurate for residential buildings but misses some structures on industrial grounds and private gardening areas.

To investigate the influence of object displacement on map accuracy, all building outlines in the cadastral database that are located in the study area were extracted. Since the direction and degree of object displacement depends on the objects' position in relation to the nadir-line, i.e. increased displacement with increased oblique view, the extracted building outlines were stratified into three zones parallel to the flight direction: pixels north of the nadir line at large positive view-angles (θ v > 10°); pixels along the nadir line (10°≥  θ v ≥  -10°); and pixels south of the nadir line (θ v < -10°). In a similar way, the street network was extracted from the UEIS and stratified accordingly. Showing the true ground cover, this assessment was used to quantify the street area that is covered by trees. At large view-angles, high buildings might actually occlude parts of the street behind it and this assessment therefore relates indirectly to the issue of object displacement. The accuracy assessment based on building outlines and the street network also serves to indirectly investigate the quality of the spectral land cover classification.

↓98

In addition, 37 residential blocks of different size were randomly selected from the UEIS to assess the quality of the map on impervious surface coverage. For this purpose, the corresponding polygons were extracted, buildings were excluded from the polygon, and the outlines were then projected onto the aerial photographs. Based on an equidistant raster of 20 by 20 m (40 by 40 m for polygons greater 25,000 m²), a set of points within each polygon was labeled either pervious or impervious. Two different values of TIA were then assigned to the polygons: (1) a value based on the visible surface that relates to the sensor view; (2) a value corresponding to true ground cover, which was derived after identifying the ground cover underneath trees. The two values were used to assess the map on impervious surface coverage and the influence of tree cover.

Field survey

Parallel to the over-flight of the HyMap sensor, a detailed field survey was performed. Based on the 0.25 m aerial photographs, two different digital layers were mapped in a geographical information system (GIS). The first layer specifies 21 surface types associated to land use and more than 40 different types of dominant surface materials at ground level. In a second layer, the extent of individual trees or groups of trees is mapped on top of the ground level. Altogether, 17 survey areas of approximately 220 by 220 m were mapped this way; nine of them were located within the area covered by the hyperspectral image. The distribution of land cover in the survey areas is not representative for the whole city. To enable a continuous mapping, dense residential areas with closed courtyards or inaccessible industrial sites were avoided. Easy to map park areas with closed tree canopy were also excluded. The abundance of open space or derelict sites is above average. Displacements of buildings in the aerial photographs have been accounted for during mapping.

↓99

The information of the two GIS-layers was used in two ways. At first, it was intersected and generalized into 12 land cover related surface categories for a spatially contiguous assessment of the land cover map (Table V-2). Intersecting regions were then assigned the attributes from the tree cover layer. Thus, the resulting layer rather represents the sensor view situation (Fig. V-6, left). The 12 surface categories help to identify potentially critical surface types within the land cover classes.

At second, the ground layer was generalized to impervious and pervious surfaces, with values of 100% and 0%, respectively, and intersected with the tree canopy layer. The resulting four categories can be used to derive the relation between TIA of the survey areas in sensor view situation and true ground cover (Fig. V-6, right). This impervious surface survey was used to assess the impervious surface map from the HyMap data with regard to tree cover.

Table V-2: Surface categories for detailed assessment of the land cover classification with corresponding description and area for nine field survey areas. The class to which the surfaces were assigned in the training data for classification is indicated.

Category

Description

Label in training data

Area [m²]

individual trees

single coniferous and deciduous

vegetation

16,224

tree groups

closed canopies of more than 1 tree

vegetation

30,864

Shrubs

deciduous and coniferous at different heights; partly interrupted by bark mulch and organic soil

vegetation/soil

6737

Lawn

irrigated; non-irrigated; sparse

vegetation

90,650

Soil

soil, including pervious sports areas

soil

10,472

derelict sites

different stages of succession; might include rubble

soil/vegetation

9517

construction sites

open pits; might include heaps of rubble and sand

soil

4558

roof tops

all types of buildings and materials

buildings

46,040

Tracks

tram and railroad; tracks with gravel

paved

4386

Courtyard

industrial grounds and courtyards of different size

paved

15,581

Sealed

all other non built-up impervious surfaces

paved

147,867

Water

Water

water

7441

↓100

Figure V-6: Reference maps derived from ground mapping shown for one of nine subsets. The 12 land cover related surface categories were used for a spatially continuous assessment of the land cover map (left). The four categories of surface types related to imperviousness were used to assess the sensor view in comparison to the true ground cover (right).

Chapter V:5 Results

Chapter V:5.1 Land cover classification

The land cover classification without orthorectification yields an overall accuracy of 88.7% (κ = 0.84) based on the 1253 reference pixels. Confusion is low for all classes and the user's accuracy is relativels well balanced (Table V-3). The individual assessment within the urban structure types shows accuracies of 90% or better for green areas (single residential 89.3%, apartment complexes 94%, suburban 94.7%). The remaining classes exhibit accuracies above 80% (center 83.1%, dense residential 84.6%, industrial 80.8%). 52.7% of the study area are classified as vegetation, 16.2% as buildings. 22.3% of the area are identified as non built-up paved grounds. Soil and water constitute the smallest classes at 4.8% and 3.9% respectively. Since reference pixels were selected from the image itself and aerial photographs only used to aid during labeling, possible geometric inaccuracies between the classification output and reference products are not taken into account. Thus, overall accuracy relates to the spectral classification quality and not to final map accuracy.

Table V-3: Confusion matrix, producer's and user's accuracy for land cover classification results.

Classification

Reference pixels

Total

User’s accuracy

vegetation

built-up

impervious

pervious

water

vegetation

541

4

5

7

2

559

96.8

built-up

0

183

24

5

0

212

86.3

impervious

20

33

270

21

3

347

77.8

pervious

4

4

6

39

0

53

73.6

water

0

0

4

0

78

82

95.1

Total

565

224

309

72

83

1253

Producer's acc.

95.8

81.7

87.4

54.2

94.0

↓101

The assessment of land cover derived from hyperspectral imagery within the building outlines from the cadastral data leads to an accuracy for the class buildings that is below the user's accuracy of the classification without orthorectification. The accuracy differs clearly according to the distance to the nadir line (Table V-4). Whereas 65.5% of all pixels in the building polygons are classified correctly near nadir, values decrease to 52.3% in the southern and 62.4% in the northern parts of the study area. The small decrease in northern parts is compensated by vegetation, paved and soil pixels; the greater decrease in southern parts leads to an increase of paved and vegetated areas.

Table V-4: Distribution of land cover for stratified areas of building outlines and street network.

Class

Building outline

Street network

 

> 10°

nadir

< -10°

overall

> 10°

nadir

< -10°

overall

vegetation

13.0

11.5

17.8

14.2

36.5

31.5

31.9

33.3

buildings

62.4

65.5

52.3

60.0

10.7

8.1

13.2

10.6

paved

19.5

19.0

25.1

21.2

50.2

58.2

51.7

53.4

soil

4.8

3.5

3.9

4.1

2.4

2.1

2.7

2.4

water

0.3

0.5

0.7

0.5

0.2

0.2

0.2

0.2

The number of pixels correctly classified as paved surface on areas of the street network is generally lower than that of buildings within building outlines. Instead, a consistently high fraction of pixels is labeled as vegetation. This vegetation fraction is higher in the northern parts while a decrease of paved pixels combined with an increase in buildings can be observed in the southern parts (Table V-4).

↓102

The analysis based on surface categories derived from the field survey shows how the accuracy of the land cover map differs for different surface types within the land cover classes (Table V-5). For all surface categories, the correct land cover class constitutes the greatest fraction, but for some of them this fraction is only 40% to 50%. The low value for the small water area is negligible and caused by an inaccurate digitization along the shore.

Table V-5: Distribution of land cover as mapped from HyMap data for different surface categories. The correct assignment is indicated in bold.

Category

buildings

paved

vegetation

soil

water

individual trees

6.5

25.5

62.3

5.6

0.0

tree groups

1.0

5.5

91.6

1.9

0.0

Shrubs

13.6

17.9

60.2

8.3

0.0

Lawn

2.8

8.8

74.8

13.5

0.1

Soil

3.6

8.2

26.9

61.3

0.0

derelict sites

3.6

29.2

18.7

48.6

0.0

construction sites

19.8

29.9

9.5

40.8

0.0

roof tops

68.8

21.9

6.7

1.6

1.0

Tracks

7.9

72.8

17.2

2.2

0.0

Courtyard

26.3

44.5

23.3

5.3

0.6

Sealed

15.2

64.9

15.1

4.2

0.6

Water

2.1

3.4

0.2

8.0

86.3

Chapter V:5.2 Impervious surface coverage

↓103

The map on impervious surface coverage is derived from the land cover classification. All pixels were assigned a value of 0% - in the case of vegetation, soil, and water - or 100% for buildings and non built-up paved areas. Although more differentiated approaches to assign degrees of imperviousness to surface types exist in literature (e.g.Hodgson et al., 2003), it was decided to use these values that are also used by the city's planning department. Identical values were assigned to the land cover in the reference data sets and this simple approach is not expected to have a positive bias on presented accuracies.

Collapsing classes in the original 5 class confusion matrix by impervious and pervious surfaces leads to an overall accuracy of 94.3% based on the 1253 reference pixels. For an areal assessment of impervious surfaces, the HyMap based map is first compared to results from the field survey. The RMSE based on the nine survey areas within the HyMap data frame is 8.8 and 10.4 for the situation with trees and based on the true ground cover underneath the trees, respectively. The higher value in the case of true ground cover shows how TIA is underestimated by the HyMap data.

Similar results are achieved by the comparison of the HyMap based map to estimates for the 37 UEIS polygons. These polygons cover a wider range of surface compositions than the survey areas. RMSEs of 14.3 and 20.7 are calculated with and without tree cover respectively. Again, an offset results from comparing HyMap based results and true ground cover, while almost a 1:1 relationship is achieved including the tree cover that shows the sensor view situation (Fig. V-7).

↓104

Figure V-7: Impervious surface estimates based on HyMap data compared to impervious surface fractions derived from aerial photographs for 37 UEIS polygons. Values from ground mapping relate to sensor view including tree cover (left) and to the true ground cover (right).

Chapter V:6 Discussion

In the following, results from the land cover classification of vegetation, buildings, paved surfaces, soils and water and the impervious surface map are discussed. The three research questions are sequentially addressed with regard to the different accuracy assessments.

Chapter V:6.1 Accuracy of the urban land cover classification

The results clearly indicate that hyperspectral data allow the spectral differentiation of basic land cover classes in urban areas. Despite some remaining confusion between buildings and paved areas, as well as soil patches mistakenly labeled as paved, the user's accuracy of all classes is high. Even for dark areas, which are often treated as a meaningless shadow class (e.g.Shackelford and Davis, 2003), there is 89.2% accuracy. A detailed assessment of misclassified reference pixels shows that mixed pixels and phenomena, such as cars on streets or sand heaps on industrial grounds, account for some of the confusion.

↓105

The comparison of land cover in the classified image and of the surface categories based on the field survey reveals other potential sources of error (Table V-5): the low accuracies for derelict and construction sites can be explained by the spectral similarity of sand or sandy soils in open pits and concrete surfaces on the one hand; on the other, objects like vehicles or heaps of rubble, which are spectrally more similar to competing land cover, exist on these surface categories. This underlines that land cover does not directly relate to spectral properties and that issues of land use should better be treated separately (Cadenasso et al., 2007). The fraction of vegetation is relatively little on lawn surfaces, which may be confused with soil, due to non-irrigated lawns or sparse plant cover. This confusion of organic soil and sparse vegetation also exists the other way around and a clear distinction between the two is not possible. It is also obvious from field observations that the two classes form transitional spectral classes.

The spectral information content of airborne hyperspectral imagery is of great value in separating and describing vegetation, built structures and other surface materials. Inaccuracies that have been reported for separating buildings from non built-up impervious areas using multispectral data (e.g.Shackelford and Davis, 2003) exist to a lesser extent. Remaining confusion related to spectral ambiguity has to be judged against the general advantages of remote sensing, i.e. synchronous coverage of large area at relatively low cost (Mathieu et al., 2007). Considering the number of dynamic surfaces like construction sites and derelict areas, the possibility of regular monitoring is an important asset of remote sensing approaches. These points apply especially for regions in the world with incomplete, inaccurate or missing cadastre information and a lack of detailed maps on urban environmental indicators (Miller and Small, 2003).

The SVM classification in this work neither requires building spectral sub-classes nor incorporating texture measures (e.g.Benediktsson et al., 2005) or segment-based analysis (e.g.Shackelford and Davis, 2003). It only uses the full hyperspectral information of original pixels. This way, the classification approach is very simple and requires no additional and often time intensive processing steps like image segmentation or feature extraction. Taking into account the high accuracies within areas of the different urban structure types, SVM classifications of HyMap data are expected to generally perform well for urban areas.

Chapter V:6.2 Spatial accuracy of land cover and impervious surface maps

↓106

The difference between the high accuracy for the class buildings based on the set of reference pixels (Table V-3) and the small portion of overall 60% of pixels classified as buildings within the building polygons from the cadastral information (Table V-4) is obvious. Since reference pixels do not account for geometric inaccuracies, this discrepancy can be related to insufficient orthorectification and object displacement. The value of 65.5% buildings in polygons near nadir is explained by the 4 m resolution of the image data, the general inaccuracy of the orthorectification (compare Section 4.2) and by a slight displacement of very high buildings within this interval. Thus the buildings' positions from remote sensing based mapping and polygon outlines from cadastral data never match perfectly (Fig. V-8, upper-right). This general disparity between the orthorectified image data in raster format and the vectorized polygons can also be observed by comparing the HyMap based map to those from the field survey. Accuracy of the land cover map for surface categories that mainly exist as small patches is lower than for those categories of rather large spatial extent. The latter contain less mixed pixels and are less influenced by the mentioned inaccuracies. For example, single trees are mapped at significantly lower accuracy than tree groups or lawn surfaces. Areas with shrubs often form narrow corridors and are many times located close to high buildings.

Figure V-8: Building positions from land cover mapping compared to polygons from cadastre. Roof-tops at large view-angles north of the nadir region are shifted northwards (upper-left), south of the nadir region they are shifted southwards and façades are not illuminated (bottom). Buildings near nadir exhibit no shift
(upper-right).

The impact of the building displacement alone can be assessed by comparing the portion of building pixels within polygons of the nadir region to areas acquired at large view-angles. Off-nadir accuracies are generally lower. The decreases of 13.2% and 3.1% for southern and northern parts of the image, respectively, can be explained by the influence of shade: on the one hand, non-illuminated façades that are visible in the southward sensor view are classified as paved (Fig. V-8, bottom); the illuminated façades to the north, on the other, can be differentiated from paved surfaces and the classification within the building polygon is more accurate, although the roof pixels appear north of the buildings' actual position (Fig. V-8, upper-left). Vegetation increases at larger view-angles, due to trees in front of the façades. The spectral signal of vegetation is very dominant on dark areas with low reflectance. Therefore non-illuminated mixed pixels are mostly labeled vegetation. This explains the greater increase in vegetation to the south. In general, the decrease of 13.2% rather reflects the impact of the displaced roofs than the decrease of 3.1% that is compensated by the "correctly" assigned façades. The impact of buildings occluding adjacent surfaces is also indirectly shown by the other reference products: in the maps from the field survey, 26% of the surface categorized as courtyard is classified as buildings. The portion of pixels classified as building within the street network polygon increases at larger view-angles where the buildings exhibit more displacement. In general, the displacement is a function of view-angle and building height. A building of 20 m height, for example, exhibits about 3.5 m or ~1 pixel offset at 10° off-nadir and 11.5 m or 3-4 pixel offset at 30°. Even when a DSM is available, occluded surfaces can only be reconstructed when at least one additional image acquired from a different position is available (Zhou et al., 2005).

↓107

The spatial accuracy of the land cover map based on airborne hyperspectral data varies significantly for different areas in the image, i.e. different view-angles (Schläpfer and Richter, 2002). For most urban environmental studies, the detailed information on patterns of impervious areas and the abundance of roof-top areas is required at larger units like the UEIS blocks for the city of Berlin or for an entire watershed (Tong and Chen, 2002). In these cases, the total area of such objects is mapped with sufficient accuracy by airborne hyperspectral data. Whenever configuration and connectivity of built-up land or land cover in general is required (e.g.Alberti et al., 2007), the object displacement is not expected to be of great negative impact due to its regular and homogeneous increase by view-angle. In such cases, it is rather the missing information on the occluded surfaces behind buildings that may hamper analyses. This is especially important when trees are underestimated in narrow street canyons due to the high relevance of trees for modeling urban environmental quality (Nichol and Wong, 2005). Similarly, the overestimation of built-up areas at large view-angles to the north, which is caused by both roof-tops and façades being classified as building, may have negative impact.

In urban climate models, buildings are often explicitly addressed. Sometimes, this information is averaged to grid cell variables (e.g.Martilli et al., 2002), other times a level of detail is required, that would be influenced by building displacement (e.g.Harman and Belcher, 2006). In either case, however, information on building height or roof inclination is needed. This underlines the need to combine land cover products derived by means of remote sensing, with additional digital information. In such cases, the geometric inaccuracy of the hyperspectral information at large view-angles appears most critical and the use of a precise DSM during geometric processing is required to produce spatially accurate information.

Chapter V:6.3 Influence of tree cover on impervious surface estimates

The delineation of impervious areas by spectral information works well. Most of the remaining misclassification relates to confusion of soil and the two impervious classes. By aggregating buildings and non built-up impervious areas, some of the described inaccuracies become obsolete. For example, the misclassification of dark façades as paved at large view-angles does not change results when buildings and paved areas are treated as one. However, the phenomenon of buildings occluding non impervious areas at large view-angles remains.

↓108

The phenomenon of trees obscuring impervious areas, on the other hand, is shown to be of great relevance in all accuracy assessments. More than 30% of the pixels within the area of the street network of the UEIS are classified as vegetation. Given the distinct spectral properties of vegetation, this error can not be explained by spectral classification errors. Differences between the nadir region and large view-angles are generally low due to the little height of trees. The increase in the northern parts is explained by the dominance of the very brightly illuminated portions of trees in mixed pixels.

The fact that image data represents the sensor view situation with tree cover and the general underestimation of impervious area are also shown by the comparison of imperviousness values based on HyMap and values derived for subsets from the field survey and for the UEIS polygons (Fig. V-7). In both cases, the analysis underlines that the sensor view situation correlates well, whereas true ground cover is generally underestimated by analysis from remote sensing image data.

The rate of error can also be shown in the reference data itself: Comparing the estimates of impervious surface coverage for the 37 UEIS polygons with and without trees obscuring paved areas shows an average difference of 7.5%. For three polygons, estimates differ by 30% and differences exist at all degrees of imperviousness (Fig. V-9). The same assessment based on all 17 field survey areas leads to an underestimation of only 3.8% in average.

↓109

Figure V-9: Distribution of impervious surface estimates for 37 UEIS polygons. Values relate to sensor view including tree cover and to the true ground cover.

The influence of trees obscuring impervious surface is very critical and, more importantly, hard to predict. The discrepancy between 33.3% vegetation obscuring the area of the street network, 7.5% offset on residential UEIS blocks (where streets and building objects have been excluded), and 3.8% for survey areas, which contain a great portion of open spaces, is great. Given the high variation of values, it appears impossible to apply a standard correction to underestimated impervious surface coverage. We could not discover a direct relation between TIA and the amount of impervious surface covered by trees based on the available reference data.

Detailed spatial information on streets, the geometry of the street canyon, and the distribution of trees is needed in many boundary layer climate models, for example to model airflow and pollutant distribution (Tsai and Chen, 2004), the surface-atmosphere energy exchanges (Pearlmutter et al., 2007), or in UHI simulations (Hirano et al., 2004). For cities comparable to Berlin, a semi-automated mapping of streets from remote sensing data appears infeasible considering error rates of about 30%. In such cases, additional digital information on the street network is crucial. Whenever digital information on streets exists, remote sensing can provide important additional information on the distribution of trees above the digitized surface.

Chapter V:7 Conclusion

↓110

The quality of urban surface mapping based on airborne hyperspectral remote sensing data was investigated in this work. The high spatial and spectral resolution of the image data and the diverse reference information allowed for the performance of intensive accuracy assessments and the identification of sources of inaccuracy. The precision of the land cover classification, the influence of geometric inaccuracy caused by the complex urban geometry and the remote sensing perspective, and the impact of trees obscuring surface underneath were addressed.

By classifying the hyperspectral data with an SVM classifier, high classification accuracy was achieved for five land cover classes including the spectrally critical classes buildings and paved. The spectral information of multispectral sensors like Ikonos does not allow the differentiation of such surfaces (Small, 2003). Classification accuracy is high for all urban structure types and the setup of the classification approach is simple. We assume that the combination of HyMap data and SVM can be successfully used in other urban environments. More instruments with similar characteristics, for example the Airborne Reflective Emissive Spectrometer (ARES) (Müller et al., 2005), will be available in near future and the number of applications in this field is expected to increase. This way, similar data will be more and more available for urban regions with less additional information.

The orthorectification of airborne hyperspectral data will often include the use of a DEM similar to the one used in this work. The detailed analysis of map accuracy revealed the negative impact of a missing DSM. The decision, of whether the work with commonly available DEMs is sufficient, must be based on the scope and scale of analysis. In the environmental context, the missing information on occluded surfaces behind high buildings at large view-angles appears more critical than the homogeneous and linear displacement of objects. Information on patterns of impervious surface, including a differentiated treatment of built-up and non built-up areas, is possible at relatively fine scales. The combination of image information and additional digital data is probably limited to regions near nadir or to sensors with generally small swath widths. This issue is of high relevance for remote sensing based analysis because spaceborne instruments with high spatial resolution, like Quickbird or Ikonos, frequently use off-nadir acquisition at view-angles of up to 30°. The availability of an accurate DSM will further increase the value of airborne hyperspectral data for urban environmental analyses.

↓111

The impact of trees obscuring possibly impervious surfaces underneath was shown to be highly variant and of important relevance, especially in the case of the street network. This impact is a general problem in remote sensing based analyses. The continuous mapping of the street network will be influenced by trees in sensor view and hence information on the connectivity and configuration of impervious areas. By performing multitemporal analyses at leaf-on/leaf-off situation, the quantification of this impact in relation to different urban structure types might be possible. In this case, the information derived from airborne hyperspectral remote sensing data is expected to serve as a useful input for an increased number of urban environmental analyses.

Chapter V:8 Acknowledgements

The author is grateful to T. Scheuschner for the great help during GIS analysis and the effort he put into visualizing results from the ground mapping. B. Kleinschmit, B. Coenradie, L. Haag, A. Damm, and the Berlin City Department for Urban Development are thanked for providing the reference data based on the UEIS and the digital cadastre. The contribution of the students from Humboldt-Universität who performed the field survey is greatly appreciated. P. Griffiths helped with the GIS integration of the survey data and together with M. Langhans he performed great parts of the preprocessing of the HyMap data. S. van der Linden was funded by the scholarship programs of the German Federal Environmental Foundation (DBU) and the German Academic Exchange Service (DAAD). The cost of the HyMap data was covered by the DBU the German Research Foundation (DFG) under project number no. HO 2568/2-1, and the DFG research training group 780/2.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
27.05.2008