Urban flooding has had a strong negative impact on many cities around the world for most of human history and certainly in recent decades [1
]. More than half of the global population lives in urbanised areas, and the frequency as well as the intensity of hydro-meteorological extremes are on the rise [4
]. Urban flooding is therefore likely to cause greater losses in the coming decades. For example, urban flooding and associated property damage accounted for 73% of the total damage caused by flooding in the USA from 1960 to 2016, amounting to USD 107.8 billion [6
]. Of the total damage caused by flooding in Japan’s three largest cities, namely Tokyo’s 23 wards, Osaka City, and Nagoya City, from 2006 to 2013, 82% of the total damage was caused by urban pluvial flooding [7
]. Urban pluvial flooding is one of the typical urban floodings and flood damage that occurs when rainwater is not discharged to mainstream or tributary rivers because rainfall exceeds the design capacity of a drainage facility. In recent years, severe urban pluvial flooding has been caused by the reduction of rainwater soil penetration amounts because of changes in land use and increased strong downpour occurrence frequency. Furthermore, according to the IPCC assessment report [8
], the risk of urban pluvial flooding in urban areas will increase in the future as a result of increased strong downpours due to climate change. Therefore, it is important to clarify the characteristics of past urban pluvial flooding areas in large cities and urbanising cities, where damage from urban pluvial flooding will become more apparent in the future, and take efficient countermeasures against urban pluvial flooding.
Managing urban flood risk is a high priority worldwide, as suggested by a large number of cities, from all continents, taken as case studies in recent research papers dedicated to urban flood modelling [9
]. In Japan, it has become mandatory for local authorities to publish urban pluvial flooding hazard maps as a measure against urban pluvial flooding. On the other hand, the spatial and temporal characteristics of flooding in urban areas are complex due to extensive changes to land use [10
] that introduce micro-urban features such as buildings, roads, and drainage networks [11
], and most inundation simulations cannot accurately simulate flooding in urban areas [13
]. Thus, flooding simulations using physical models represent the flooding mechanism through known physical analyses. Therefore, if the unknown factors are mainly important for the inundation mechanism, the accuracy will be low. On the other hand, the statistical analytical approach is considered to be able to elucidate the distribution of areas where urban pluvial flooding frequently occurs (hereafter referred to as frequent urban pluvial areas) based on past flooding area records and to quantitatively evaluate their characteristics, including unknown factors that cannot be considered in flooding simulations [14
A statistical analytical approach has been greatly developed by utilising several appropriate factors, such as bedrock geology, soil properties, land use, drainage networks, road networks, building, and precipitation. Some previous studies have identified flood mechanisms. For example, Fariza et al. (2019) used fuzzy multi-criteria decision making (FMCDM) to assess urban flood risk levels in Sidoarjo, Indonesia [15
]. Sato and Hayashi (2014) used principal component analysis (PCA) to analyse the main topographic characteristics of inundation in the Musashino Plateau of Tokyo and Saitama [16
]. However, no report of an earlier study describes the quantitative assessment of the universal characteristics of flood-prone areas because they analysed areas where inundation has occurred at least once [14
], and most of the earlier studies were limited to case studies of one city or one case study.
Djamres et al. (2021) identified the frequent urban pluvial flooding areas using seven years of urban pluvial flooding area records during 2008–2015 and analysed, using the PCA, their topographical characteristics in Tangerang, Indonesia [14
]. Results showed that frequent urban pluvial flooding areas in Tangerang emerged because of a slope in the upstream condition, the correlation between concave and flow length conditions, the correlation of the slope condition and distance to a river, and relationships among flow length in upstream characteristics and distance to a pond. Furthermore, 29% of frequent urban pluvial flooding areas had low topographical similarity because of anthropogenic factors such as changes in overland flow directions due to the change of a slope in the upstream condition by land-use change and trapping of flood water by “dominant structures” that may influence inundated water flows in and around frequent urban pluvial flooding areas. Mignot and Dewals (2022) argued that a particular impediment to progress in urban flood modelling science is that the conclusions of most studies remain genuinely site-specific formulations and that significant progress could be made by attempting to extract general knowledge from the collection of existing case studies [9
]. They also suggested that this may be achieved by designing appropriate metrics for classifying and standardizing the definition of flooding scenarios, investigated processes, and effects of analysed factors [9
]. As the newly defined urban pluvial flooding frequency areas contain primary factors related to flooding mechanisms, statistical analysis of frequent urban pluvial flooding areas may reveal previously unknown factors such as anthropogenic factors and their relative influence on the flooding mechanism.
In Japan, flooding area records have been available since 1993, and Kakehashi et al. (2014) [17
] used these records to identify frequent river flooding areas throughout Japan and to elucidate their formation mechanisms. However, there have been no studies that have utilised these records for urban pluvial flooding. Therefore, the objectives of this study were to identify the frequent urban pluvial flooding area using these records for 20 years during 1993–2012 in Osaka and Nagoya Cities, the two cities with the largest proportion of urban pluvial flooding damage from 2006 to 2013 [7
], and to analyse the topographical characteristics of frequent urban pluvial flooding areas and their distribution at both cities by applying methods reported by Djamres et al. (2021) [14
]. This study also aimed to elucidate anthropogenic characteristics of urban pluvial flooding from the views of the location of “dominant structures” and the impact of drainage system improvement.
4. Discussion and Summary
This study clarified the distribution of frequent urban pluvial flooding areas in Osaka and Nagoya Cities by using urban pluvial flooding area records. The identified frequent urban pluvial flooding areas were 70 in Osaka City and 108 in Nagoya City, and their proportion to the area of each city was 0.34% in Osaka City and 0.33% in Nagoya City. Analyses of their characteristics revealed the following:
The PCA of frequent urban pluvial flooding areas using the eleven topographical factors showed high cumulative contribution rates for Osaka and Nagoya Cities, which indicated that the topographical factors used in this study were appropriate for describing frequent urban pluvial flooding areas. The results of the PCA quantitively showed that the topographical characteristics of the frequent urban pluvial flooding areas in both cities were different.
Using the results of PCA of topographical characteristics in the frequent urban pluvial flooding area, the similarity with these topographical characteristics at a 100 m mesh scale in both cities was quantified. Although many urban pluvial flooding areas were located in areas with similar topographical characteristics, especially in Osaka City, the urban pluvial flooding areas were also distributed in areas without similar topographical characteristics. This suggested that factors other than topographical characteristics that caused urban pluvial flooding were largely responsible for such areas in many parts of Osaka City and some parts of Nagoya City.
Anthropogenic factors such as “dominant structures” and drainage system improvements as characteristics other than topographical characteristics on the occurrence of urban pluvial flooding were shown to be influential. The results also showed that the impact of anthropogenic factors was greater in Osaka City, which is on a larger urban scale than Nagoya City.
In general, urbanisation increases the accumulation of assets in topographically flood-prone areas, and the risk of urban pluvial flooding increases. On the other hand, urban pluvial flood risk has been reduced through the improvement of drainage systems in such areas; however, urban pluvial flooding has also been observed in topographically less-flood-prone areas due to changes in land use and land cover. In other words, as urbanisation progresses, the main cause of urban pluvial flooding is likely to shift from topographical factors to anthropogenic factors. The results of this study quantitatively showed this paradigm shift of urban pluvial flooding factors by the statistical analysis of newly defined urban pluvial flooding frequency areas. This is demonstrated as one of the methods for a major advance in urban flood modelling science proposed by Mignot and Dewals [9
]. Furthermore, this study showed that it is difficult to describe past urban pluvial flooding areas in Osaka and Nagoya Cities solely based on topographical characteristics. This was consistent with the findings of the numerical experiment [22
] and statistical analyses [14
] and showed that in some cases that artificial structures formed areas vulnerable to urban pluvial flooding even outside topographically flood-prone areas.
In addition, it is particularly difficult to collect flow velocity and depth observations during urban floods, as they are usually of short duration. The recent proliferation of mobile phones and online video-sharing platforms gives access to countless amateur videos [23
], which are being used in most geophysical sciences, but difficulties with retrieving the location and time of the scenes impair the use of these data for detailed model validation [9
]. The urban pluvial flooding area record used in this study and the newly defined frequent urban pluvial flooding area could be important sources for this validation.
On the other hand, although this study succeeded in quantitatively assessing the differences in the impact of topographical characteristics on the formation of frequent urban pluvial flooding areas due to urban scale by comparative analyses in Osaka and Nagoya Cities, this study only compared two cities, and the relationship between urban scale and frequent urban pluvial flooding areas has not been quantitively clarified. Further quantitative comparison with the characteristics of frequent urban pluvial flooding areas in other cities of different urban scales would help to understand the characteristics of urban pluvial flooding and their transition from topographical factors to anthropogenic factors, which is associated with urbanisation. Furthermore, socio-hydrology [26
], which deals with the interaction between water systems and human activities, has been recently applied worldwide for flooding and water resources management [28
]. Further quantitative comparative analysis of the characteristics of frequent urban pluvial flooding areas among different urban scale cities would provide a quantitative understanding of the system dynamics of urban pluvial flooding interacting with urbanisation, namely human activities.