Climate change and urbanization have led to an increase in extreme rainfall events. This increase often causes flooding due to insufficient drainage systems and sewer capacity, resulting in infrastructure failures, lack of public safety, and losses fo...
Climate change and urbanization have led to an increase in extreme rainfall events. This increase often causes flooding due to insufficient drainage systems and sewer capacity, resulting in infrastructure failures, lack of public safety, and losses for emergency response and recovery. It is essential to prepare management strategies to minimize the loss of people and properties. The South Korean government is implementing structural management techniques such as increasing the design frequency of sewage pipes to address this issue.
Despite the efforts made to prevent urban flooding caused by extreme rainfall events through structural management, such methods still have limitations. For instance, they are often resource-intensive and may not allow for rapid response. Non-structural management, on the other hand, is an effective approach that is often more cost-effective. It also allows for real-time monitoring, forewarning, and response. Therefore, it's crucial to consider non-structural methods in addition to structural methods when managing urban flooding caused by extreme climate change. This study proposes a methodology for selecting the optimal water level monitoring points for non-structural management of urban flooding and a process for urban flood response using monitoring.
Initially, the probable rainfall of the study area was calculated through hydrological analysis using RFAHD. Since there is an absence of Automatic Weather Station(AWS) in the study area, the rainfall data was constructed for 28 years (1995~2022) by weighting the nearby AWS through a Thiessen polygon. To identify the rainfall characteristics of the study area, Inter-Event Time Definition(IETD) was calculated by autocorrelation analysis, average annual rainfall occurrence analysis, and coefficient of variation analysis. The coefficient of variation analysis method can determine the point where the mean and standard deviation are close to 1, so it was used to determine the IETD for rainfall event separation, which was found to be 18 hours. After separating the rainfall events, 603(32.52%) events with a duration of 1hour were calculated, followed by 368(19.85%) events lasting for 2hours, and 223(12.03%) events with a duration of 3hours. To determine the probable rainfall intensity formula, the probable rainfall was calculated using the rainfall data. In this study, the polynomial regression type was adopted, as the coefficient of determination was higher than 0.999 for all return periods.
Next, to consider various characteristics of rainfall from normal to extreme events, this study evaluated the design frequency and regional features of the study area.
Although sewage facilities in South Korea are presently proposed for a frequency of 10, 30, and 50 years, there is a need to consider the 100-year frequency as well due to the increasing frequency of extreme rainfall events. The study area has a higher frequency of short-duration rainfall events, and therefore 12 scenarios were set by combining four return periods(10, 30, 50, and 100 years) and three durations(1, 2, and 3 hours) to reflect the rainfall characteristics of the area. The temporal distribution of Huff's four quartiles, specifically the third quartile which is used for flood response, was then applied based on the scenarios.
Third, a methodology for selecting optimal water level monitoring points for urban flood response was developed. To select the optimal water level monitoring points, we derived the points of overflow for each scenario and determined the principal points. Considering the results of all scenarios, a total of 67(3.43%) maintenance holes were expected to experience overflow out of a total of 1,956 maintenance holes, and the principal points were concentrated downstream. This is a reflection of the close hydrological relationship between upstream and downstream and the accumulation of rainfall downstream.
To perform efficient monitoring of principal points, grouping was conducted by considering the maximum distance between maintenance holes and the connectivity of pipes. Since about 84.35% of the pipes in the study area are 600mm or less, 75m was considered as the maximum distance in consideration of the maintenance holes installation conditions provided in the sewage design standards. At this time, even if the distance between maintenance holes exceeds 75 meters, if they are connected by a single pipe, they were attributed to the same group because there is a straightforward interconnection. Therefore, a total of 17 groups ([Group 1]~[Group 17]) were derived, and the downstream group was [Group 15].
[Manhole 52] in [Group 15], which is the downstream group, was determined to be the optimal water level monitoring point that satisfied the maximum similarity of the water level pattern and the minimum distance of the flood traces in all scenarios. Based on this point, 12 points were derived by synthesizing the scenarios, taking into account the maximum similarity of the water level pattern and the minimum distance from the flood traces, as well as the non-redundancy of the groups and the number of monitoring points according to the budget. Based on this, four points were selected, prioritizing the points with no redundancy within a radius of 200 meters and the shortest minimum distance to the flood traces, considering the runoff time.
Consequently, this study proposed a real-time monitoring-based non-structural response process utilizing the selected optimal water level monitoring points. This study utilized the ultra-short-term precipitation forecast data provided by the Korea Meteorological Administration(KMA) to propose a specific forecast method for related organizations when the rainfall is predicted to be above the threshold. Based on the predicted rainfall, the simulation results derived from the forecasted rainfall indicate the predicted point of occurrence of overflow. At the same time, the optimal evacuation route is proposed based on real-time location, which can be utilized to build an emergency response system. The process used the Dijkstra algorithm to find the shortest route to the evacuation shelter by setting the flood risk area as a constraint.
The methodology suggested in this study can be used to enable efficient and flexible monitoring, even with limited resources. The proposed method of monitoring and response can adapt to the unique features of each region and is expected to be used for effective decision-making in preemptive response to urban flooding.