The Empirical Relationships Between Green Space Characteristics and Flood Events

Abstract

Countermeasures for increasing resilience to urban flooding should consider long-term perspectives because climate change impacts are unpredictable and complex. Recent approaches to climate change adaption have emphasized disaster control, sustainable development, and urban green spaces. For flood control, green spaces can evolve dynamically depending on the physical environment of an urban flood

therefore, identifying the regional features of green spaces is necessary to maximize their effect. In this study, flood vulnerable areaflooded areas in Seoul, Korea, were divided into four flooded area types, and statistical analysis was performed to determine how the flooding probability change with green space area, type, and pattern. In this way, regional features that maximize the effects of green spaces on flood resilience were identified and can now be reflected in the planning and design of green spaces. First, a model to evaluate flood vulnerable areas in Seoul city was developed using MaxEnt. The variables selected for model simulation included those related to the physical environment, climate environment, green space environment, and flood risk management infrastructure (FRMI). The model was simulated by extracting random points 1000 times considering uncertainty. Flood was not taken place in 43 of 239 drainage basins in Seoul. On this basis, the flood vulnerable areas identified were: Seocho4, Gildong, Shinwol3, Bangbae1, and Hwagok2 drainage basins. Second, flooded area types were divided into 4 types based on features of flooded area by using multivariate statistics. Type 1 included regions where flooding occurred in a drainage basin that had a FRMI. These basins were located around the Han River and major streams and were bordered by mountains. Basin slopes were gentler than the slope of Seoul city and the was the second highest identified. These basins were characterized by residential and commercial mixed land use. Type 2 is the regions with steep slopes, low TWI, and the best drainage identified. Compared with the other types, the green space ratio was high. These basins were bordered by steep mountains allowing the downward flow of water without attenuation, which was identified as regional feature of flood resistance. Type 3 represented the gentlest sloping areas, and these were associated with the highest TWI value, and the worst soil drainage. In contrast to type 2, the dominant regional feature was the attenuation of standing water. Type 4 had features that were intermediate to those in type 2 and type 3 (e.g., moderate slopes, imperfect soil drainage, and lower than average TWI value). Third, differences in flooding probability based on green space area, type, and pattern for each flooded area type was comparatively analyzed using logistic regression analysis. We found that green spaces exerted a considerable influence on urban flooding probabilities in Seoul and flooding probabilities could be reduced by over 50% depending on the green space area and the locations where green spaces were introduced. Increasing the area of green spaces was the most effective method of decreasing flooding probability in type 3 areas. In type 2 areas, the maximum hourly precipitation affected the flooding probability significantly, and the flooding probability in these areas was high despite the extensive green space area. On the basis of the results, a formula was developed to identify the green space areas required to reduce flooding probability. Green spaces were categorized as planted area, grassland, wetland, paddy field, field, orchard, or forest based on their CN value, and the contributions of green space areas to flood control for each flooded area type were analyzed. For type 1, grassland showed the highest contribution, followed by forests and then planted areas. For type 2, only the forest type was analyzed with respect to flood control. For type 3, paddy fields showed the highest contribution, followed by fields, planted areas and forests. As most farmland in Seoul is located on gentle slopes bordered by mountains, natural rainwater is often retained in the basin as confined water. For type 4, forests showed the highest contribution, followed by planted areas and fields. For the green space patterns of types 1 and 2, the area-weighted Mean Shape Index (AWMSI) represented as significant variable, with complexity increases correlated with increased flooding probability. Type 3 contained an area in which the flood control efficiency of the green space area was high, and the green space area (CA), number of green space patches (NumP), MPS, and AWMSI were found to be significant variables that exerted a positive influence on flooding probability reduction. In Type 4, increases in NumP were correlated with reduced flooding probability. The results of this study show that green spaces in urban areas can impact upon flooded area type

however, flood control functions also correspond to topographic factors (i.e., slope, TWI, soil drainage)

therefore, green spaces should be introduced to areas that will ensure maximum efficiency for flood control. Green space area, type, and pattern were suggested as a factor to reduce flooding probability according to the properties of the flooded area type. In addition, guidelines for increasing flood resilience were developed to assist with the spatial planning of green spaces as countermeasures for urban flooding. In the case of artificial FRMI such as rainwater retention basins, their value may decrease over time, but increasing the green space area is an eco-friendly solution that will benefit humans and nature over a long period of time. The role of existing green spaces is often limited to the production of ecological benefits for wildlife and aesthetically pleasing landscapes for human residents, but functionally, proper design plans for green space locations could maximize their impact on flood control. Therefore, this study recommends that urban areas devote planning resources for green spaces, and such efforts should determine where the best areas are for their introduction. It is expected that the approach used in this study and the results obtained will provide a framework for diverse research on green spaces in the future. Furthermore, the techniques employed may be useful for predicting flood probabilities in urban areas, i.e., the models, which were based on empirical data, had a high explanatory capability.