Access the full text.
Sign up today, get DeepDyve free for 14 days.
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
International Journal of Biodiversity Science, Ecosystem Services & Management, 2013 Vol. 9, No. 2, 136–145, http://dx.doi.org/10.1080/21513732.2013.782342 An evaluation of ﬂood control and urban cooling ecosystem services delivered by urban green infrastructure a a, b Simon Farrugia , Malcolm D. Hudson * and Lindsay McCulloch Faculty of Engineering and the Environment, Centre for Environmental Sciences, University of Southampton, Highﬁeld, Southampton, SO17 1BJ, UK; Southampton City Council, Civic Centre, Southampton SO14 7LY, UK To inform planning decisions and address climate change impacts in expanding cities, it is desirable to quantify urban ecosys- tem services like ﬂood control and urban cooling. By comparing with a purpose-built habitat map, this study ground-truthed a method to assess ﬂood control, which was developed by Southampton City Council from surface maps. It was conﬁrmed that inﬁltration capacity is a good proxy for ﬂood control, leaf area index could represent urban cooling, and thereby both could be used to score urban surface types. A two-tiered system was proposed so that surface maps would be used for city-wide scale, and as they produce similar results that are more accurate at ﬁne scales, habitat maps are used at site level. These surrogates were integrated to produce a Green Space Factor for ﬂood control and urban cooling, wherein a combined score can be generated for particular locations. This could be extended further to include other ecosystem services. The new integrated multi-scale ecosystem service quantiﬁcation tool could be used by developers and policy-makers to identify target areas in their projects and policies that could beneﬁt from enhanced green infrastructure. Keywords: GIS; urban cooling; ecosystem services; planning ecology; ﬂood control; cities 1. Introduction 2. A background to past developments in urban green infrastructure Although most cities are customarily grey with buildings and other conventional infrastructure like roads (Wolch 2.1 Beneﬁts of green areas in cities 2007), citizens and policy-makers are becoming more con- Green infrastructure is the interconnected network of green scious of the need for these cities to be greener (Young areas and open spaces (Benedict & McMahon 2006; et al. 2009). Greener cities are better places to dwell in Finlay 2010). It involves the incorporation of natural life- on several counts (Lafortezza et al. 2009), including the support systems into built-up areas such as cities (Finlay quality of life of their citizens (Bolund & Hunhammar 2010). Such infrastructure may include combinations of 1999; Chiesura 2004). Many local authorities, especially watercourses, parks, green roofs and trees on pavements. in more economically developed countries, are looking Research has shown that such places can host a wide for innovative ways to increase the amount of green variety of ecosystems, which are of beneﬁt to human pop- areas and infrastructure in their cities (Florida Department ulations (Angold et al. 2006). Understanding these beneﬁts of Environmental Protection and the Florida Greenways would enable planners to devise policies that incorporate Coordinating Council 1999; Rosenzweig et al. 2006; Finlay green areas in a sustainable way while minimising their 2010). Despite this interest, the important role of green costs. infrastructure is not sufﬁciently recognised and hence lacks The beneﬁts of ecosystems on human well-being and integration into spatial planning systems (Kruuse 2011). economic stability are called ecosystem services (De Groot Various attempts have been made to quantify the beneﬁts et al. 2002; Fisher et al. 2009; TEEB, The Economics of green areas in urban contexts and to inform planning of Ecosystems and Biodiversity 2010). Apart from their processes accordingly. Such attempts have, however, failed ecological or biophysical value of performance (De Groot to permeate fully into the policy-making system either et al. 2002), documents such as the Stern Review (Stern because they lack transparency in the methodology used 2006) and TEEB (2010) have ascribed economic value or because they are too site- or service-speciﬁc (see exam- to these ecosystem services, while some authors (Bolund ples in Section 2.3). This article explores a simple and & Hunhammar 1999; Gómez–Baggethun & Barton 2013) generally applicable scoring tool for green areas in cities, are also ascribing them social value. Currently, some of which could be applied in wider contexts to enlighten these services are of urgent relevance to society in view the policy-making process. This is explored by examining of rapidly changing climatic conditions, obliging cities to ﬂood control and urban cooling ecosystem services. adapt their conﬁgurations to the new weather phenomena *Corresponding author. Email: email@example.com © 2013 Taylor & Francis International Journal of Biodiversity Science, Ecosystem Services & Management 137 and extremes (IIP Digital | U.S. Department of State 2010). potential. The BAF is thus a ratio of the green area with In this regard, research has shown that green spaces in cities the total land area. It also formulated target BAFs for spe- provide ﬂood control and urban cooling regulatory services ciﬁc urban functions, which developers were obliged to (Wong & Yu 2005; Rosenzweig et al. 2006), both of which meet so that any development proposal could be approved have an important economic and social value, besides their (Landschaft Planen & Bauen and Becker Giseke Mohren biophysical value, which shall be the subject of this article Richard 1990). (Gómez-Baggethun & Barton 2013). A similar system was adopted by Malmo City Council in Sweden, which focussed on the incorporation of green and blue infrastructure among existing land uses, while minimising the extent of sealed or paved surfaces in any 2.2 Linking urban green areas to particular ecosystem development (Kruuse 2011). Similar to the Berlin BAF, the services Malmo Green Space Factor (GSF) scored surfaces accord- It is known that vegetation reduces surface run-off, fol- ing to their green areas and sealed proportions. While the lowing precipitation events, by intercepting water through scores in the BAF ranged from 0 for sealed surfaces to the leaves and stems (Villarreal & Bengtsson 2005). The 1 for dense vegetation connected to a deep soil, those for underlying soil would also help in reducing inﬁltration the Malmo GSF were initially scored from 0 to 20. In fact, rates, since it can act as a sponge by storing water in the the original GSF system also required developers to choose pore spaces until it percolates as through-ﬂow and base- from a list of possible green infrastructure, such as green ﬂow. Thus, habitats, having a thick soil layer covered with roofs, allotments and water surfaces, to be incorporated in dense vegetation, would have a high inﬁltration capacity their projects so as to achieve the desired number of ‘Green as the water would take longer to pass through and would, Points’. The system was later revised to rescale the weight- therefore, help in preventing ﬂood events downstream ings to 0–1 while including the same green infrastructure (Villarreal & Bengtsson 2005). Besides the underlying soil, (Kruuse 2011). the same vegetation also absorbs water acting as a natural The results of these European systems were applied by reservoir after precipitation events. Some of this moisture the GRaBS (Green and Blue Space Adaptation for Urban would later be released through transpiration (Miglietta Areas and Eco Towns) project (Community Forest North et al. 2011). West 2011). This project developed a user-friendly tool to Green infrastructure also helps to reduce ambient tem- be employed by developers and planners based on similar peratures in cities (Hardin & Jensen 2007). During the weighting factors to the above systems. GRaBS launched process of evapotranspiration, the plant absorbs sensible the North West Green Infrastructure Factor, which is more heat and releases latent heat, when converting water from focussed on the actual infrastructure than the generic sur- a liquid state to a gaseous state, leading to a lowering of face types. Moreover, it splits the surfaces into three air temperature in the nearby areas. This effect increases horizontal layers consisting of a ground-level layer, which with the area of leaves and foliage density, since more includes grasses and water surfaces, a secondary layer hav- water is transpired from a larger surface area (Hardin & ing shrubs, hedges and trees and a tertiary layer with roofs Jensen 2007). Furthermore, the process of photosynthe- covered by plants. sis in green plants requires the absorption of heat energy A similar attempt to quantify the generic value of from the incoming solar radiation to synthesise glucose, green areas was made in Kent Thameside (Defra 2008). thereby cooling the surrounding environment (Rosenzweig However, this project looked at multiple ecosystem ser- et al. 2006). Moreover, trees provide shade to the near- vices but scored them separately. These services, which ground habitat by their canopies, consequently, reducing included recreation and ﬂood control, were scored using the proportion of insolation reaching and heating ground predetermined surrogates and data. Such a concept was level (Kawashima 1991). Therefore, vegetation also helps also explored in Florida urban forests, with a system of to reduce the transmittance of heat energy to the underlying a much more detailed indicators based on literature and surface by absorbing insolation through its relatively low ﬁeldwork (Dobbs et al. 2011). albedo, while reducing the amount of heat radiated from These ﬁve approaches share common ﬂaws. The the ground (Rosenzweig et al. 2006). approaches that they used cannot be replicated across different spatial scales nor in other ecoregions, since quan- tiﬁcation was scale-speciﬁc and occurred within temperate 2.3 Quantifying urban ecosystem services climates. Furthermore, with the exceptions of the Kent Several attempts have been made in different European Thameside and the Florida urban forests, no scoring cri- cities at quantifying the biophysical value of green areas teria have been veriﬁed by other studies. therein, so as to direct planning procedures. One of the ﬁrst Given such limitations, there is a need to ground-truth attempts was made in Berlin with the Biotope Area Factor the existing systems on several scales and possibly extend (BAF) (Landschaft Planen & Bauen and Becker Giseke them to more ecosystem services. Moreover, it is impor- Mohren Richard 1990). This system scored land surface tant to maintain a user-friendly decision support tool that types in development sites according to their ecological is transparent in its functionality, so as to trace the driving 138 S. Farrugia et al. factors of speciﬁc results and be able to make the necessary Enhanced Thematic Mapper Plus (ETM +). These ﬁndings amendments in either planning policy or development were used in our study to score different surface types proposals. Having identiﬁed these needs, this article devel- according to their contribution to urban cooling. Land ops an improved ecosystem services quantiﬁcation tool that surface temperatures were found to correlate strongly (r addresses these issues. = 0.7805) with air temperatures in cities (Guthrie 2006), implying that the scores to be assigned for land tempera- ture data could be extrapolated to air temperature data and 2.4 Southampton Green Space Factor vice versa. One of the partners within the GRaBS project is As GIS was used to calculate the area of each surface Southampton City Council (SCC). As part of its project type, it was also possible to visualise and locate the areas work, SCC started to develop a version of the GSF tool that need to be targeted in planning policies to maintain or to evaluate the extent to which green areas contribute to improve the ecosystem services they deliver. This scoring ﬂood control based on scoring systems used in other North system also attempts to assess the generic value of small European cities (Landschaft Planen & Bauen and Becker habitats on small-scale development sites to help the devel- Giseke Mohren Richard, 1990; Finlay, 2010; Kruuse 2011). oper make the best use of potential ecosystem services on The system works by assigning high scores to areas consid- site. ered as having a high inﬁltration capacity and hence a sub- stantial contribution to ﬂood control. These areas are iden- tiﬁed after combining the Ordnance Survey MasterMaps , 3. Methodology which are maps with a comprehensive view of landscape 3.1 Site of the study in Great Britain (Ordnance Survey 2012), and the SCC Open Spaces, which is a map layer showing all unbuilt Southampton, in the United Kingdom, is a coastal temper- areas in Southampton. Upon scoring these two maps in ate city, which is simultaneously facing rising population the Geographic Information System (GIS), a surface type density levels and changing climate. Having experienced a approach has been adopted. rapid urbanisation process, the city had to construct res- While acknowledging the strong relationship between idential and commercial units on previously virgin land, surface type, inﬁltration capacity and ﬂood control, SCC thus reducing the amount of green space (Southampton is aware that the above-mentioned maps were not created City Council 2009). Given the increase in unpredictable for these purposes and may thus fail to show other char- weather events requiring rapid infrastructural adaptations, acteristics of green areas, which could also contribute to preserving such green spaces is of the utmost impor- ﬂood control. To test this surface type approach, this study tance (IIP Digital | U.S. Department of State 2010). Being therefore replicated the same GSF tool using Joint Nature situated in a low lying coastal area at the conﬂuence Conservation Committee Phase 1 habitat maps (JNCC of two major rivers (the Test and Itchen, which join to 2010) (hereafter referred to as ‘habitat type’ approach in form Southampton Water), Southampton is prone to ﬂood- contrast to the ‘surface type’ approach). The JNCC Phase ing following heavy rainstorms. Thus, it requires ongo- 1 maps were speciﬁcally designed to broadly map vege- ing investment in the maintenance of ﬂood control sys- tated areas for environmental assessment purposes and may tems to protect the existing inhabitants and infrastructure thus be more accurate in capturing details about inﬁltra- (Southampton City Council 2009). Rising temperatures, tion capacity. By comparing the scores obtained from the particularly if sustained over long periods of time, will be surface type and habitat type methods, it is possible to crit- especially hazardous. The existing housing design mostly ically evaluate them and identify the most accurate and lacks green roofs and walls, and Southampton and the cost-effective approach. United Kingdom in general have an ageing population vul- As well as ﬂood control, urban cooling is another nerable to temperature extremes (Lafortezza et al. 2009). desirable ecosystem service in dense cities (Rosenzweig This study was conducted in 50 randomly selected one- et al. 2006). Following the setting up of a GSF for ﬂood hectare sites in Southampton City Centre, each of which control, attempts were made to ﬁnd suitable surrogates comprised at least a part of terrestrial habitat. Southampton on which to construct a similar GSF for urban cooling. City Centre was selected for this case study as it has the Temperature regulation within cities is highly dependent highest building and population density in the whole city on the greenness of habitats: greenness can therefore be (Ofﬁce for National Statistics 2001). The mean study site used to estimate the urban cooling potential (Akbari et al. area was found to be 9948.47 m and ranging from 9773 to 1990; Wong & Yu 2005; Rosenzweig et al. 2006). Research 10,008 m Precise boundary delineation of study sites was carried out by Gibson (2009) has shown clear relationships complicated by the largely heterogeneous land uses, hence between surface types and land surface temperatures. This the given range. A cumulative frequency curve of habitat study was based on satellite-derived data consisting of clas- classiﬁcation with the number of sampling sites showed siﬁed Corine Land Cover categories, the more quantitative a curve ﬂattening out after the 12th site, indicating that Normalised Difference Vegetation Index (NDVI) and Soil the sample size was more than adequate in capturing most Adjusted Vegetation Index (SAVI) worked out from the habitat types. International Journal of Biodiversity Science, Ecosystem Services & Management 139 3.2 Habitat mapping Mapping was carried out using site surveys on Ordnance Survey base maps at a scale of 1:1250 and aerial The ground-truthing of the existing system, based on imagery (Google Earth 2011), with a preference for the Ordnance Survey surface types, was conducted using a former in case of discrepancies. Inaccessible areas could modiﬁed JNCC Phase 1 Habitat mapping system (JNCC only be mapped using aerial photography, and assumptions 2010). The JNCC system was modiﬁed in a number of favouring scarce undergrowth were made since most of ways. To account for the particular ecological situations these areas occurred within private dwellings usually sur- of cities with several small-scale disturbed habitats, tar- rounded by some form of managed grassland. Each habitat get notes were not used. All habitats having an area of was digitised with its standardised code in a GIS pack- 100 m or more were mapped, even if only a minor part of age on top of a 1:10,000 Ordnance Survey MasterMap the total area fell within a study site. Any habitats smaller base (Table 1). Impermeable surfaces such as buildings and than 100 m were not recorded, following JNCC (2010). roads were considered to be impractical to digitise sepa- Such small habitats would have unnecessarily complicated rately in urban settings because of their sizeable extent, and the mapping exercise without any signiﬁcant contribution thus, their area was worked out from the same MasterMap to the ﬁnal score, and many would have been in private basemap. It was decided to classify all the imperme- property, so access was difﬁcult for ground-truthing. able surfaces as either ‘structures’ or ‘transportation’ for The extent of habitats was deﬁned by visible ecological planning purposes, even though they all have a Green boundaries (Fagan et al. 1999), which make species migra- Space Score (GSS) of zero (Table 1). ‘Structures’ refer to tion between habitats difﬁcult, such as a 1-m high brick artiﬁcially built structures, which host some form of sta- wall in amenity grasslands or a main road in woodlands. tionary function other than the transportation of matter. Another modiﬁcation implemented was the exclusion of ‘Transportation’ denotes all those structures that serve to speciﬁc details, namely grassland type, water trophic con- link places for the carrying of matter from one place to ditions and species richness. Instead, the additional sub- another (Table 1). categories of dense or sparse undergrowth were added Habitat recognition, using indicator species, was facil- to woodland and scrub habitats since this was consid- itated by surveying all sites in June 2011, since June ered to provide the best indication of inﬁltration capacity marks the end of spring in the United Kingdom when most (Villarreal & Bengtsson 2005). Although this is a simpli- ﬂowering species are easily identiﬁable. ﬁed version of the original JNCC system, this classiﬁca- tion is sufﬁcient for the purposes of inﬁltration capacity. Moreover, this simpliﬁed version could be easily adapted from existing JNCC Phase 1 habitat maps without the need 3.3 Scoring ﬂood control for further site visits. Using the GIS software, the total area of each broad habi- While the JNCC system does not require the map- tat type in each sample site was calculated and transferred ping of individual trees, but as they have high inﬁltration into a spreadsheet. A GSS (ω), adopted from the SCC sys- capacity (Nijssen et al. 2001), we investigated the value tem and assigned to each habitat depending on inﬁltration of accounting for these trees in parklands. Consequently, capacity (see below), was then multiplied by the respective the locations of individual tree crowns in six sites were area, resulting in a weighted area factor for each habitat. mapped. The sites were purposefully chosen, which con- The following formula was applied to calculate the GSF tained parklands consisting of a total of 12 habitat units for inﬁltration capacity: equally split by high and low tree density. Because of the high level of detail required to distinguish individual trees, GSF = αω/A (1) the mapping was undertaken on site by recording the tree location relative to other landmarks. Aerial imagery from Google Earth (Google Earth 2011) was then used to aid where GSF = Green Space Factor for speciﬁc ecosys- the digitisation process. All trees were assumed to have an tem service, a = total area of speciﬁc habitat type; ω = equal crown diameter of 11.31 m as worked out from cal- Green Space Score of speciﬁc habitat type for inﬁltration culations of nine British broadleaf tree species by Hemery capacity, and; A = total surface area of the site. et al. (2005). This value was ground-truthed by measuring The resulting GSF score is on a scale from 0 to 1, the crown diameter utilising aerial photography of a sam- indicating a range from negligible to maximum inﬁltration ple of three trees from each of the chosen sites. The same capacity of the habitat type. The ω score (GSS) for inﬁltration capacity was tree species used in Hemery et al. (2005) were found to assigned using qualitative expert judgement based on be present in the studied sample sites. No signiﬁcant dif- other similar projects done in Malmo, Sweden (Kruuse ference was found between both mean diameters using an 2011), Berlin (Landschaft Planen & Bauen and Becker independent sample t-test (p = 0.616) (Field 2005). A non- Giseke Mohren Richard 1990) and by the North West parametric Friedman test with α = 0.05 was then utilised Development Agency (Community Forest North West to test for signiﬁcant differences between the habitat map- 2011). Inﬁltration capacity was considered to be a sound ping with individual trees, habitat type without individual surrogate for the ﬂood water regulating ability of the trees and surface type approaches. 140 S. Farrugia et al. Table 1. Deﬁnition of weighing factors (ω and θ) for the habitat type approach. Green Space Score Green Space Score Green Space Score for urban cooling (θ) for urban cooling (θ) JNCC code Habitat description for ﬂood control (ω) (Gibson 2009) (LAI) A1.1.1D Semi-natural broad-leaved woodland 1 0.8 1 with dense undergrowth A1.1.1S Semi-natural broad-leaved woodland 0.8 0.8 0.9 with sparse undergrowth A1.1.2D Planted broad-leaved woodland with 0.9 0.8 1 dense undergrowth A1.1.2S Planted broad-leaved woodland with 0.8 0.8 0.9 sparse undergrowth A2.1D Dense scrub with dense undergrowth 0.9 0.4 0.9 A2.1S Dense scrub with sparse undergrowth 0.8 0.4 0.8 A2.2S Scattered scrub with sparse 0.8 0.4 0.7 undergrowth A3 Parkland/scattered trees 0.7 0.7 0.7 AT Individual trees 0.9 0.8 1 B2.1 Unimproved neutral grassland 0.6 0.3 0.6 B2.2 Semi-improved neutral grassland 0.6 0.3 0.6 G1 Standing water 1 1 1 G2 Running water 1 0.9 1 H1 Intertidal coastland 1 0.6 0.5 I1.2 Scree 0.1 0.1 0.4 I2.2 Spoil 0.2 0.1 0.4 J1.1 Arable land 0.3 0.5 0.7 J1.2 Amenity grassland 0.5 0.5 0.5 J1.3 Ephemeral/short perennial 0.3 0.2 0.3 J2.1 Intact hedge 0.7 0.6 0.8 J2.2 Defunct hedge 0.7 0.6 0.8 J2.3 Hedge and trees 0.8 0.7 0.9 J3.6 Buildings 0 0 0 J4 Bare ground 0.2 0 0 J5 Other habitat (artiﬁcial sea-wall with 0.3 0.1 0.3 interlocking bricks) Structures Impermeable surfaces which in the 000 surface typology were classiﬁed as Building, Glasshouse, Made surface, Structure and Unclassiﬁed. Transport and Impermeable surfaces which in the 000 infrastructure surface typology were classiﬁed as Trafﬁc calming, Paths, Rail, Road, Track, Pylon, Overhead construction, Upper level of communication and steps. Note: LAI, Leaf area index. Adapted from JNCC (2010). surface, since it directly relates to the proportion of per- the former Parkland habitat (code A3, ω = 0.7), which meable surfaces in the city (Whitford et al. 2001). It was consisted of grasslands with sparse trees (JNCC 2010). assumed that more intensively managed habitats, like The above scoring procedure was run for both the sur- amenity grasslands, have lower inﬁltration capacity than face type and habitat type, and the results were compared similar natural habitats such as neutral grassland (Braun to check for consistency. A Spearman Rank correlation & Kruijne, 1994; Yüksek et al. 2010), which is neither was calculated to determine the strength of the relationship acidic nor calcareous (JNCC 2010). Habitats with taller between both scoring methods. To determine the signif- vegetation, such as trees, were assumed to require deeper, icance of the observed differences between each score, more permeable soil to accommodate an extensive root ranks were compared using a Wilcoxon signed test (Field system and would thus have a higher inﬁltration capac- 2005) at α = 0.05. ity (Whitford et al. 2001). When testing for the effect of individual trees in parklands, a ω of 0.9 was assigned to 3.4 Scoring urban cooling individual trees (code AT) in such a way that this score, together with that for amenity grassland (code J1.2, ω = Following the determination of a credible surface classi- 0.5) would theoretically balance each other out in replacing ﬁcation for ﬂood control, an attempt was made to ﬁnd a International Journal of Biodiversity Science, Ecosystem Services & Management 141 comparable system for urban cooling. As in the case of at averaging these scores to produce an Aggregate Green inﬁltration capacity for ﬂood control, surrogates based on Space Factor (AGSF), which tries to quantify these ecosys- scientiﬁc ﬁndings, which are not speciﬁc to any geographi- tem services. In theory, this AGSF could be extended to cal region (Whitford et al. 2001), were explored. The qual- any number of ecosystem services just by averaging. The itative surface type categories identiﬁed by Gibson (2009) AGSF could be worked out by were arbitrarily converted to both the surface and habitat typologies using the Corine Land Cover 2000 Database (GSF + GSF + ... + GSF ) w c AGSF = (produced by the European Environment Agency). These #GSF were ranked according to the ﬁndings of Gibson (2009) and scored a θ (climate control) factor between 0 and 1 where, (Table 1). In cases where it was difﬁcult to split this classi- ﬁcation into the different surfaces used in the surface type AGSF = Aggregate Green Space Factor approach, substitutes such as evapotranspiration, surface GSF = Green Space Factor for ﬂood control albedo, photosynthetic activity and shade were applied by w GSF = Green Space Factor for urban cooling assuming that surfaces having high values in these substi- c #GSF = Number of Green Space Factors. tutes would result in lower temperatures (Kawashima 1991; n Rosenzweig et al. 2006; Hardin & Jensen 2007). A typi- cal case is the category ‘Scrub and herbaceous vegetation’ This assumes that the ecosystem services have equal in Gibson (2009), which for the purposes of the modi- weight, but this could be adjusted on the basis of ﬁed Phase 1 approach was divided into the broad habitats their importance to policy or society or weighted by of ‘scrub’ (A2) and ‘grassland’ (B2) in order of decreas- stakeholders. ing importance for temperature regulation. When multi- layered surfaces occurred in the surface type approach, the topmost layer was used for both the water regulation and 4. Results urban cooling scores, as it was considered to have more effect on the proxies than the ground layer (Kalluri et al. 4.1 Flood control 1998). Almost all the study sites scored a GSF smaller than 0.7, Since literature shows strong relationships between leaf with just two sites scoring higher than 0.9 upon employing area index (LAI) and temperature (Wong & Yu 2005; both methodologies. Using the habitat mapping, the GSF Hardin & Jensen 2007), another attempt was made using for the 50 sites in Southampton City Centre was found to this surrogate to score both the surface and the habi- be 0.23, while on using the surface type data, it was found tat typologies using the presumed LAI. In this approach, to be 0.24. standing and running water habitats were both scored as The Spearman Rank test resulted in a very strong pos- 1, since they were equally the coldest according to Gibson itive correlation of 0.966 between both mapping types for (2009). GSF . The surface type yielded higher mean and median The GSF for each of the four approaches was calcu- (0.2404 and 0.1626, respectively) than the habitat type with lated using Equation (1) and replacing ω with a different a mean of 0.2260 and median of 0.1348. However, the θ for each temperature surrogate. Finally, these tempera- Wilcoxon signed test showed that these differences are not ture regulation approaches were compared to GSFs based statistically signiﬁcant (p = 0.657). on actual temperature data from September 2008 derived The Friedman test found signiﬁcant differences in the from Landsat ETM+ Band 6. This data set was chosen in ranks of habitat type with the individual trees approach order to be consistent with the ﬁndings of Gibson (2009) when compared to that without trees and the surface type about correlation with surface types and because it was approaches (p = 0.028). Further exploration of the data the most recent data set used (Gibson 2009). The orig- reveals that in all six sample sites with parklands, mapping inal temperature values were categorised into 10 equal individual trees yielded a higher GSF for both ﬂood con- classes, and each class scored a progressively decreas- trol and urban cooling than mapping the habitats without ing score of 0.1, starting at 1 for the coldest. Because individual trees or just the surface types. of the large amount of data processing required to score such quantitative surface typologies, other indirect surro- gates suggested by Gibson such as NDVI and SAVI were 4.2 Urban cooling not considered. The correlation between the temperature data and the ﬁnal scores for our sample sites derived from All the Spearman Rank tests, comparing land surface tem- each of the four approaches used was calculated using a perature with the four surrogates used to represent urban Spearman Rank test. A Wilcoxon signed rank test was cooling potential, found a moderate to strong positive cor- then applied to check the signiﬁcance of the different relation at p ≤ 0.001 for each surrogate (Table 2). The scores. Wilcoxon signed tests comparing the same surrogates with After having obtained two GSFs for ﬂood control land surface temperature found no signiﬁcant difference (GSF ) and urban cooling (GSF ), an attempt was made between their GSFs with all at p ≤ 0.001. w c 142 S. Farrugia et al. temperature scores, indicating that all the approaches can Table 2. Spearman Rank correlation coefﬁcients (r )ofdiffer- ent surrogates with land surface temperature. suitably represent urban land surface temperature (Table 2). The strongest correlation with temperature is exhibited by Surface type approach Spearman rho the LAI mapped on the habitat system (r = 0.788) closely MasterMap (Gibson) 0.646 followed by the habitat system according to Gibson (2009) Modiﬁed Phase 1 (Gibson) 0.785 (r = 0.785). This could indicate that habitat mapping is a MasterMap (leaf area index) 0.642 more accurate indicator of urban cooling potential than the Modiﬁed Phase 1 (leaf area index) 0.788 surface type data. Note: All coefﬁcients were found to be statistically signiﬁcant at p < 0.001. 5.3 Applications in urban spatial planning Signiﬁcant differences in mapping individual trees suggest 5. Discussion that such a mapping approach would be more accurate, 5.1 Flood control particularly for smaller scales. Given that individual trees A comparison of the results of both approaches indicates contribute so much to these two, and other ecosystem ser- that there is a minor difference in the GSF bothona vices, their inﬂuence would be more signiﬁcant in smaller site-speciﬁc scale and on a larger scale across the city areas such as gardens than larger portions of land such as centre. This difference is not statistically signiﬁcant and parklands. The fact that parkland habitats always yielded could therefore have occurred by chance. This indicates higher scores than the other approaches might serve to that either approach could be used to work out the GSF for highlight the importance of individual trees in ﬂood control ﬂood control. The very strong positive correlation between and urban cooling. This could be explained by the observa- the observations of both approaches (Figure 1) further tion, suggested by aerial imagery and site inspection, that conﬁrms this notion. more than half the area of the managed parks in the city It could also be noted that the JNCC habitat type centre is covered by tree canopies (Google Earth 2011). approach yields lower scores than the surface type Unlike the case of ﬂood control, there were signiﬁcant approach, possibly because of its wider span of scores. differences between temperature and all the qualitative In this case, this could have been exacerbated by the chosen habitat types. Such a ﬁnding is meaningful since it area having a lot of surfaces with low inﬁltration capacity indicates that temperature cannot be used directly in our and thus low scores. Two extremely elevated results, higher system to score the cooling potential since the data has a than 0.9, were measured for sites containing water bodies signiﬁcantly different distribution. In fact, the mean and like the River Itchen and Southampton Water. median of temperature GSF scores were higher than the other variables, while the range and standard deviation were much smaller (Table 3). 5.2 Urban cooling Since there is no signiﬁcant difference in the results of both approaches for ﬂood control, one could easily use In the case of urban cooling, high scores in all the approach that proves to be the most cost-effective and four approaches correlate with elevated land surface adaptable to multiple purposes. The surface type approach is surely accurate enough for planning purposes on a regional scale. In addition, it does not require further data collection as, in the United Kingdom at least, this is readily available in digital format through the Ordnance Survey. On the other hand, the habitat type approach could be more suitable for small scale development purposes, especially if it maps the individual tree coverage. In theory it would be easy to adopt since it only requires minor modiﬁcations and reclassiﬁcations of Phase 1 habitat maps, which would be required for certain projects of high ecological sensitivity (JNCC 2010). The statistically signiﬁcant difference found for the modiﬁed habitat type approach with individual trees vis-á-vis the other two approaches might suggest a higher accuracy for this typology. This approach could thus be recommended, especially in view of the ecological nature of the JNCC maps, which might be closer to representing some other ecosystem services. Because of the impracti- cality of producing such a map on a city-wide scale, the adoption of a two-tiered system is suggested, whereby the habitat map would be used for individual development Figure 1. Correlation between Green Space Factor scores gen- sites, whilst the surface maps could be used for larger areas erated by the habitat type and surface type approaches. such as cities. International Journal of Biodiversity Science, Ecosystem Services & Management 143 Table 3. Summary statistics for computed Green Space Factor scores using four surface type approaches and land surface temperature. Surface type Habitat type Surface type Habitat type temperature temperature temperature temperature Land surface regulation (Gibson) regulation (Gibson) regulation (LAI) regulation (LAI) temperature ( C) Mean 0.24 0.20 0.23 0.23 0.48 Median 0.16 0.08 0.15 0.12 0.48 Range 0.96 0.96 0.96 0.96 0.47 Standard deviation 0.23 0.23 0.24 0.25 0.10 Note: LAI, Leaf area index. This two-tiered approach could also be applied for ﬂood control services offered by different ecosystems in urban cooling, with the surface typology used for large a city. Similarly, LAI could be used to score urban cooling scale urban or regional purposes and the habitat type potential. These surrogates can be incorporated into dif- used for small scale purposes like development proposals. ferent mapping systems depending on the required spatial For the habitat typology, there is a signiﬁcant difference scale. The surface types in MasterMaps from Ordnance between the Gibson approach and the LAI, with the latter Survey could be utilised for strategic planning purposes showing a stronger correlation with surface temperature. on a large city-wide scale, while habitat maps derived LAI could therefore be suggested as a surrogate for urban from a modiﬁed JNCC Phase 1 system could be used for cooling potential at a small site-speciﬁc scale. In the case of individual development projects. the surface typology, both approaches are not signiﬁcantly The ecosystem services could be scored by inputting different, and the discrepancy in the Spearman Rank fac- these maps into a GIS package, working the area of each tor is only 0.004 (Table 2). For consistency and simplicity surface type and multiplying it by a weighted GSS depend- purposes, the adoption of the LAI as a surrogate for large ing on the surrogate used. Such a system could yield a scale scoring purposes (Hardin & Jensen 2007) could be single GSF between 0 and 1 for each service, the average suggested. of which could provide a combined GSS for all the cho- The habitat maps could be incorporated in any form sen ecosystem services. This average could be weighted of impact appraisal or assessment and be utilised to pre- to account for the local situation, for example ﬂooding is dict the change in the ecosystem services produced by more of an issue in some cities than others. Target GSFs a speciﬁc development project. Target GSF scores could could then be incorporated into development policy to be set as indicators for the appropriateness of a project. make informed planning decisions. On a strategic level, surface type ecosystem service maps Although this work acknowledges the fact that no sin- could indicate target areas in need of further investment gle surrogate could completely represent any ecosystem in green infrastructure. Such target areas could consist of service, it has shown that surrogates could provide a valu- those bridging existing high scoring sites, which would able tool to aid policy-makers and developers in making also serve to create stepping stones between ecologi- better informed judgements. This tool is also intended to cal islands to facilitate species migration (Reid et al. be ﬂexible in its approach, giving scope for the scores to be 2002). modiﬁed after better proxies have been found and weighted On a project level, target GSF scores could be set to account for the value or importance of the ecosystem as indicators of the need for green infrastructure in cer- services. Further research could shed more light on the fac- tain project types. As suggested by the Mersey Forest tors affecting these ecosystem services, thereby enabling a Team (Short 2011), planned development projects should more reﬁned and accurate scoring system. Other ecosys- demonstrate a GSF score of 0.2 higher than the existing site tem services, such as air quality and recreational value, score. Planned projects on green ﬁeld sites should score could also be added to this tool to produce a more holistic at least 0.6 to ensure that a minimum level of ecosys- understanding. tem services is maintained. This could be either prescribed With the increasing threat of climate change and the for speciﬁc ecosystem services or for the aggregate score, growing interest in stakeholder participation approaches, depending on the policies of the local authority. Such there will be an increasing need for a more accurate measure would couple development projects with improve- yet user-friendly tool to measure the services produced ments in ecosystem services and the multi-functionality freely by nature. The utility of such tools might even associated with such green areas. challenge traditional views that development is naturally opposed to environmental conservation by introducing an innovative system whereby development is directly linked with improvement in ecosystem services. This would be 6. Conclusion especially useful in brownﬁeld sites and urban locations It is possible to score the ecosystem services of a densely where through better design choices, it will be possible developed city using speciﬁc surrogates for particular ser- to protect and possibly enhance the remaining ecosystem vices. Inﬁltration capacity can be used to quantify the services. 144 S. Farrugia et al. Acknowledgments Google Earth [Internet]. 2011. Aerial image of Southampton City Center. Google; [cited 2011 Jul 20]. Available from: http:// The research work disclosed in this publication is partially funded maps.google.com by the Strategic Educational Pathways Scholarship Scheme Guthrie P. 2006. A study of the development of the heat (Malta), part-ﬁnanced by the European Union – European Social island effect in local microclimates and its relationships Fund. The authors are grateful to Southampton City Council with human comfort [master’s project]. Southampton: School (a partner in the INTERREG IVC project GRaBS) for provid- of Civil Engineering and the Environment, University of ing the tree list and spatial data. Our deepest gratitude also goes Southampton. to Dr P.E. Osborne and B. Gibson for the remote sensing data and Hardin PJ, Jensen RR. 2007. The effect of urban leaf area on sum- K. Farrugia for his assistance in the ﬁnal production of this tool. mertime urban surface kinetic temperatures: a Terre Haute case study. Urban Forest Urban Green. 6:63–72. Hemery GE, Savill PS, Pryor SN. 2005. Applications of the crown diameter-stem diameter relationship for different species of References broadleaved trees. Forest Ecol Manag. 215:285–294. IIP Digital. U.S. Department of State [Internet]. 2010. U.S. Akbari H, Rosenfeld AH, Taha H. 1990. Cooling urban Cities Lead the Way on Climate Change Policies; heat islands. In: Proceedings of the 4th Urban Forestry [cited 2013 Feb 26]. Available from: http://iipdigital. Conference; [unknown date of conference]; St. Louis (USA). usembassy.gov/st/english/article/2010/07/20100716144330 Washington (DC): Amer Forestry Assoc. ksevir0.5723078.html#axzz2M2HYvns6 Angold PG, Sadler JP, Hill MO, Pullin A, Rushton S, Austin K, JNCC. 2010. Handbook for Phase 1 habitat survey – a technique Small E, Wood B, Wadsworth R, Sanderson R, et al. 2006. for environmental audit. Peterborough: JNCC. Biodiversity in urban habitat patches. Sci Total Environ. Kalluri SNV, Townshend JRG, Doraiswamy P. 1998. A simple 360:196–204. single layer model to estimate transpiration from vegetation Benedict M, McMahon ET. 2006. Green infrastructure: linking using multi-spectral and meteorological data. Int J Remote landscapes and communities. Washington (DC): Island Press. Sens. 19:1037–1053. Bolund P, Hunhammar S. 1999. Ecosystem services in urban Kawashima S. 1991. Effect of vegetation on surface tempera- areas. Ecol Econ. 29:293–301. ture in urban and suburban areas in winter. Energ Buildings. Braun HMH, Kruijne R. 1994. Soil conditions. In: Ritzema HP, 15:465–469. editor. Drainage principles and applications. Wageningen: Kruuse A. 2011. GRaBS Expert Paper 6: the green space factor International Institute for Land Reclamation and and the green points system. In: GRaBS editor. The GRaBS Improvement; p. 77–111. Project. London: Town and Country Planning Association & Chiesura A. 2004. The role of urban parks for the sustainable city. GRaBS; p. 1–14. Landsc Urban Plan. 68:129–138. Lafortezza R, Carrus G, Sanesia G, Davies C. 2009. Beneﬁts Community Forest North West. 2011. Toolkit for developers. In: and well-being perceived by people visiting green spaces in Gill S, editor. Green infrastruture to combat climate change periods of heat stress. Urban Forest Urban Green. 8:97–108. part of the north west climate change action plan. Bolton: PJ Landschaft Planen & Bauen, Becker Giseke Mohren Richard. Web Solutions Ltd.; p. 1–4. 1990. The biotope area factor as an ecological parameter – De Groot RS, Wilson MA, Boumans RMJ. 2002. A typology principles for its determination and identiﬁcation of the tar- for the classiﬁcation, description and valuation of ecosystem get. Berlin: Senate Department for Urban Development in functions, goods and services. Ecol Econ. 41:393–408. Berlin. Defra. 2008. Case study to develop tools and methodologies to Miglietta F, Peressott A, Viola R, Körner C, Amthor JS. 2011. deliver an ecosystem based approach – Thames Gateway Stomatal numbers, leaf and canopy conductance, and the Green Grids. London: Defra. control of transpiration. Proc Natl Acad Sci. 108:E275. Dobbs C, Escobedo FJ, Zipperer WC. 2011. A framework for Nijssen B, Schnur R, Lettenmaier DP. 2001. Global retrospec- developing urban forest ecosystem services and goods indi- tive estimation of soil moisture using the variable inﬁl- cators. Landsc Urban Plan. 99:196–206. tration capacity land surface model, 1980–93. J Climate. Fagan WF, Cantrell RS, Cosner C. 1999. How habitat edges 14:1790–1808. change species interactions. Am Nat. 153:165–182. Ofﬁce for National statistics. 2001. Neighbourhood Statistics: Field AP. 2005. Discovering statistics using SPSS. London: Sage thematic map. Southampton. Publications. Ordnance Survey [Internet]. 2012. Southampton: Ordnance Finlay E. 2010. Green infrastructure to combat climate change. Survey MasterMap Topography Layer; [cited 2012 Aug 28]. Northwest Climate Change Action Plan and GRaBS project. Available from: http://www.ordnancesurvey.co.uk/oswebsite/ Edinburgh: Community Forest Northwest. products/os-mastermap/topography-layer/index.html Fisher B, Turner RK, Morling P. 2009. Deﬁning and classi- Reid W, Ash N, Bennett E, Kumar P, Lee M, Lucas N, Simons H, fying ecosystem services for decision making. Ecol Econ. Thompson V, Zurek M. 2002. Millennium ecosystem assess- 68:643–653. ment: ecosystems and human well-being – synthesis. Penang: Florida Department of Environmental Protection and the ICLARM. Florida Greenways Coordinating Council. 1999. Connecting Rosenzweig C, Solecki WD, Slosberg RB. 2006. Mitigating New Florida’s communities with greenways and trails: a sum- York City’s heat island with urban forestry, living roofs, and mary of the Five Year Implementation Plan for The Florida light surfaces. In: Savi P, editor. New York City Regional Greenways and Trails System. Tallahassee, FL: Florida Heat Island Initiative Final Report. New York (NY): New Department of Environment Protection. York State Energy Research and Development Authority; Gibson B. 2009. The effect of land cover and topography p. 4–8. upon satellite derived land surface temperature, and its rela- Short M. 2011. Liverpool green infrastructure strategy – techni- tionship with urban air pollution and population [master’s cal document. Liverpool: Liverpool City Council Planning project]. Southampton: School of Civil Engineering and the Department. Environment, University of Southampton. Southampton City Council. 2009. Submission core strategy Gómez-Baggethun E, Barton DN. 2013. Classifying and valu- public examination hearing statement on other issues (b). ing ecosystem services for urban planning. Ecol Econ. Southampton: Southampton City Council. 86:235–245. International Journal of Biodiversity Science, Ecosystem Services & Management 145 Stern N. 2006. Stern review: the economics of climate change. Wolch J. 2007. Green urban worlds. Anna Assoc Am Geogr. London: HM Treasury. 97:373–384. TEEB. 2010. The economics of ecosystems and biodiversity: Wong NH, Yu C. 2005. Study of green areas and urban heat island mainstreaming the economics of nature: a synthesis of in a tropical city. Habitat Int. 29:547–558. the approach, conclusions and recommendations of TEEB. Young C, Jarvis P, Hooper I, Trueman I. 2009. Urban landscape Geneva: TEEB. ecology and its evaluation: a review. In: DuPont A, Jacobs H, Villarreal EL, Bengtsson L. 2005. Response of a Sedum green- editors. Landscape ecology research trends. New York (NY): roof to individual rain events. Ecol Eng. 25:1–7. Nova Science Publishers; p. 45–69. Whitford V, Ennos AR, Handley JF. 2001. “City form and natural Yüksek T, Kurdoglu ˘ O, Yüksek F. 2010. The effects of land use process” – indicators for the ecological performance of urban changes and management types on surface soil properties in areas and their application to Merseyside, UK. Landsc Urban Kafkasör protected area in Artvin, Turkey. Land Degrad Dev. Plan. 57:91–103. 21:582–590.
International Journal of Biodiversity Science, Ecosystem Services & Management – Taylor & Francis
Published: Jun 1, 2013
Keywords: GIS; urban cooling; ecosystem services; planning ecology; flood control; cities
Access the full text.
Sign up today, get DeepDyve free for 14 days.