Scenario forecasts - A Difference-in-Differences Analysis in Finland

A Difference-in-Differences Analysis in Finland

4. Scenario forecasts

This section presents and discusses the urbanization trajectories predicted for each scenario. For the benefit of clarity, each main scenario with its variations is discussed in its own subsection. The discussion of each scenario focuses firstly on the aggregate characteristics of the trajectories for the whole urban region, followed by the local, spatially disaggregate, characteristics of the predicted trajectories in the coastal and near-coastal flood-prone and flood-safe areas.

4.1. BAU scenario

The current trends scenario, named business-as-usual (BAU), is summarized in Figure 4. The simulation of this scenario is based on the last available data layers (cf. Figure 2 and section 2.2) and indicates the future trajectory of current urbanization patterns, if land constraints and the transport network remain unchanged and no serious exogenous shocks happen in population, economic structure, and the way economic growth is currently distributed throughout Helsinki. This scenario also assumes that development behavior inside the floodplain continues in the future without planning interventions specific to flood risks. This is an important assumption as the market response and development restriction scenarios simulate changes in development patterns when planning interventions specific to food risks are applied.

Water bodies Omission error Commission error No error

Figure 4: Simulated output image for the BAU scenario in 2040

Figure 5 summarizes nine growth parameters of the BAU scenario. The growth rate of built-up land is overly stable at approximately 2.7% until 2020 and steadily drops each subsequent year reaching approximately 1.3% by 2040. The growth rate trajectory corresponds to almost a doubling of built-up land(‘pop’), approximately from 39000 to 66000 hectares. At the same time, the length of the urban/non-urban frontier (‘edges’)does not appear to change significantly; it increases rather weakly until 2030 and declines subtly afterwards. The growth of the total volume of built-up land while maintaining the length of the urban/non-urban boundary implies that, in addition to a significant decrease of natural land, progressively fewer urban areas will maintain direct access to natural areas.

More precisely, the number of built-up clusters (‘clusters’) will steadily decrease, while the size of those clusters (‘cluster size’) will concurrently increase, indicating that Helsinki’s built-up morphology will become more consolidated and less fragmented. This particular growth behavior links to typical development practices in Helsinki, which have historically used ample space, with preference on sprawling low-density residential areas and a notable absence of a comprehensive preservation plan for green infrastructure. However, developable land is eventually becoming less and the saturation of built-up areas implies that the overall loss of large masses of natural land, which are found mainly at the urban periphery, is coupled with the progressive loss of small green spaces located between built-up clusters across the entire city.

Edge growth is orders of magnitude larger than the other growth types handled by SLEUTH. In light of the aforementioned growth behavior, this indicates for Helsinki that the spreading out from existing built-up areas, with limited leap-frogging, translates to the filling-in of natural areas between neighborhoods. On the other hand, the emergence of areas from spontaneous and new spreading center growth is nearly absent throughout the forecast timeframe. Road-influenced growth is active throughout the forecast timeframe, but declines steadily. This is reasonable, as no new major transport links are simulated in any of the scenarios and therefore any road-influence growth is gradually saturated around existing high access links.

10 5 0 10

3 1½ 0 3

Water bodies Floodplain (F1000) Flood-safe areas 300m from coast 2000 urban area BAU growth 2012

compared to the images of actual growth. Table 5 provides the results of the accuracy assessment and the metrics indicate a satisfactory performance of the calibrated model in reproducing observed growth (cf. Chaudhuri and Clarke 2014). The maps of Figure 3 display the differences between simulated and observed growth in the three control years (2005, 2010, and 2012).

Table 5: Accuracy assessment for control years 2005, 2010, and 2012.

Year Overall accuracy (%) Kappa coefficient

2005 95.92 .86

2010 93.77 .80

2012 92.19 .77

Figure 3: Differences between simulated and observed growth in control years 2005, 2010, and 2012

4. Scenario forecasts

This section presents and discusses the urbanization trajectories predicted for each scenario. For the benefit of clarity, each main scenario and its variations are discussed in its own subsection. The discussion of each scenario focuses firstly on the aggregate characteristics of the trajectories for the whole urban region, followed by the local, spatially disaggregate, characteristics of the predicted trajectories in the coastal and near-coastal flood-prone and flood-safe areas

4.1. BAU scenario

Water bodies Omission error Commission error No error

Figure 4: Simulated output image for the BAU scenario in 2040

Figure 5 summarizes nine growth parameters of the BAU scenario. The growth rate of built-up land is overly stable at approximately 2.7% until 2020 and steadily drops each subsequent year reaching approximately 1.3% by 2040. The growth rate trajectory corresponds to almost a doubling of built-up land (‘pop’), approximately from 39000 to 66000 hectares. At the same time, the length of the urban/non-urban frontier (‘edges’) does not appear to change significantly; it increases rather weakly until 2030 and declines subtly afterwards. The growth of the total volume of built-up land while maintaining the length of the urban/non-urban boundary implies that, in addition to a significant decrease of natural land, progressively fewer urban areas will maintain direct access to natural areas.

10 5 0 10

3 1½ 0 3

Water bodies Floodplain (F1000) Flood-safe areas 300m from coast 2000 urban area BAU growth 2012

Figure 5: BAU scenario; overall volume and form (left) and growth types (right). One pixel corresponds to an area of 50x50 meters (1/4 hectare).

For Helsinki’s residential areas, the majority of unbuilt land is green infrastructure. Its loss represents a multifaceted increase in disaster risk, as the ecosystem services provided by green infrastructure are central in regulating flooding (De Groot et al. 2002; Davies et al. 2011). At the same time, in addition to the loss of natural flood regulation, the consolidation of impermeable areas, assuming that water-absorbing construction materials are not widely implemented, means exacerbation of flood events or their impacts. This is relevant not only for coastal flooding, as the damages of storm-related flooding have been increasing (cf. the urban flood in Copenhagen on 2.7.2011 with € 800 million of damage; see Gerdes 2012). From an aggregate perspective, it can be therefore suggested that, for Helsinki, the BAU scenario represents an increase in both vulnerability (loss of regulating ecosystem services) and exposure to flood-related hazards (increase and consolidation of urban areas). It is worth noting that the loss of green infrastructure increases vulnerability of the housing market to flooding via an additional route. The loss of economic value associated to the loss of proximity to natural areas is widely reported in hedonic valuation literature (Tyrväinen 1997;

Tyrväinen and Miettinen 2000; Brander and Koetse 2011; Perino et al. 2014; Siriwardena et al.

2016). This implies an increase in the economic vulnerability of households concurrently with an increase in the abovementioned physical vulnerability and exposure to flood impacts.

The abovementioned aggregate characteristics of the BAU scenario can be complemented by a closer examination of the scenario’s local characteristics. In addition to the seven flood risk zones (F5-F1000) of the urban region’s coastal areas, and as the morphology of these zones is fragmented, additional flood-safe areas within a certain distance from the coast were explored. These flood-safe areas were categorized into three indicative zones: 0.3, 0.3-1, and 1-10 km from the coastline. The distance of 0.3 km is grounded in the significant homogeneity of highly expensive coastal properties within this buffer, in terms of market behavior and physical characteristics. Beyond 0.3 km and until 1 km from the coast, one observes a second zone of coastal properties that is still of significant value, but does not belong to the far-right end of the price range. Properties between 1 and 10 km from the coast are assumed as representatives of the inland housing market.

The local characteristics of the BAU scenario were identified by applying a 90% threshold to the scenario’s cumulative urbanization probability map of year 2040 (end of forecast period). The threshold of 90% (10% uncertainty) is borrowed from common practice in statistical analysis. Since

is assumed that 10% is the maximum allowed uncertainty forthe model’s predictions of urbanization.

The total amounts of predicted built-up cells where counted for the flood risk (F5-F1000) and flood-safe zones (0.3, 0.3-1, and 1-10 km from the coastline). It is important to note that counting the growth in these zones as separate from each other represents an assumption behind flood risk mapping and economic analysis. Even though, for instance, an F1000 flood risk zone may well overlap with an F5 flood safe zone, the flood maps represent these cases as independent (i.e. separate inundation maps for different return periods) which, when overlaid, can represent conflicting information to the public, whereas urban economic analysis also assumes that the demand and price responses of the housing market to these flood risk maps is the compound result of their independent characteristics. Clearly, this assumption merits attention in future research, sorting out truly safe areas independently from return period. However, a question following the identification of non-overlapping safe areas would be, what the reaction of the market is (upon which future development depends, among other things) to areas that are flood safe in some return periods but unsafe in other return periods. A further question would be the relation between binary classifications sound for engineering analysis versus fuzzy, overlapping classifications with which the public and markets operate. In light of the above, this study’s counting of urban growth in all the different flood safe and flood prone areas adopted the compound effect assumption for the BAU scenario in order to make the BAU trends comparable to those of the DR and MR scenarios, which certainly contain compound market effects.

Table 5 summarizes the BAU trajectory in the aforementioned flood-prone and flood-safe zones.

Table 5: Local characteristics of the BAU scenario for year 2040 for near-coast areas (90% certainty threshold) Built-up land in 2040 % change from 2012

Zone pixels hectares

Flood-safe (0.3 km from coast) 8226 2057 18.6 Flood-safe (0.3-1 km from coast) 96415 24104 39.9 Flood-safe (1-10 km from coast) 16211 4053 24.0

Of interest are the divergent amounts of growth in built-up land in the various zones. The flood risk areas are set for notably higher growth in built-up land (30-70% relative to 2012) than the waterfront flood safe areas (19% within 0.3 km from coast) and inland (24% within 1-10 km from coast) areas.

The transition zone between coast and inland (0.3-1 km from coast) is the exception, with 40% of growth relative to 2012. The high growth rates in the coastal flood prone areas can be related to prior research in the topic (e.g. Bin et al. 2008a; Daniel et al. 2009) that indicates that coastal amenities overdrive decisions in the real estate sector irrespective of the risks that may be involved. In this case, the BAU simulation confirms that the forces driving new urban development overestimate the amenity dimensions while not reacting in par with the flood risk levels. The high intensity of urban growth in risky areas relative to elsewhere in the city poses challenges for the resilience of Helsinki

Figure 5: BAU scenario; overall volume and form (left) and growth types (right). One pixel corresponds to an area of 50x50 meters (1/4 hectare).

see Gerdes 2012). From an aggregate perspective, it can be therefore suggested that, for Helsinki, the BAU scenario represents an increase in both vulnerability (loss of regulating ecosystem services) and exposure to flood-related hazards (increase and consolidation of urban areas). It is worth noting that the loss of green infrastructure increases vulnerability of the housing market to flooding via an additional route. The loss of economic value associated to the loss of proximity to natural areas is widely reported in hedonic valuation literature (Tyrväinen 1997; Tyrväinen and Miettinen 2000;

Brander and Koetse 2011; Perino et al. 2014; Siriwardena et al. 2016). This implies an increase in the economic vulnerability of households concurrently with an increase in the abovementioned physical vulnerability and exposure to flood impacts.

The abovementioned aggregate characteristics of the BAU scenario can be complemented by a closer examination ofthe scenario’s local characteristics. In addition to the seven flood risk zones (F5-F1000) of the urban region’s coastal areas, and as the morphology of these zones is fragmented, additional flood-safe areas within a certain distance from the coast were explored. These flood-safe areas were categorized into three indicative zones: 0.3, 0.3-1, and 1-10 km from the coastline. The distance of 0.3 km is grounded in the significant homogeneity of highly expensive coastal properties within this buffer, in terms of market behavior and physical characteristics. Beyond 0.3 km and until 1 km from the coast, one observes a second zone of coastal properties that is still of significant value, but does not belong to the far-right end of the price range. Properties between 1 and 10 km from the coast are assumed as representatives of the inland housing market.

is assumed that 10% is the maximum allowed uncertainty for the model’s predictions of urbanization.

Table 5 summarizes the BAU trajectory in the aforementioned flood-prone and flood-safe zones.

Table 5: Local characteristics of the BAU scenario for year 2040 for near-coast areas (90% certainty threshold) Built-up land in 2040 % change from 2012

Zone pixels hectares

Flood-safe (0.3 km from coast) 8226 2057 18.6 Flood-safe (0.3-1 km from coast) 96415 24104 39.9 Flood-safe (1-10 km from coast) 16211 4053 24.0

indicates that a significant portion of the regional economy’s resources is channeled toward growth in risky coastal areas instead of safer areas or instead of being invested into flood insurance or flood protection options. It also indicates an increase in the society’s exposure and vulnerability to flood risks, as large volumes of urban development typically translate to large volumes of residential building stock, associated public infrastructure, and population.

4.2. Market response scenarios

Figure 6 displays an overview of the simulated output of the market adaptation scenarios, MRa and MRb, in the coastal zone. As introduced in Section 1, these scenarios aim to translate the housing market effects of the public disclosure of flood risk information into gradual urban development

In document Space and Price in Adapting Cities : Exploring the Spatial Economic Role of Climate-Sensitive Ecological Risks and Amenities in Finnish Housing Markets (sivua 126-134)