An Exploratory Optimal Framework of Low Impact Development Measures Spatial Arrangement based on Source Tracking for Urban Flood Mitigation


 Urban areas are vulnerable to flooding as a result of climate change and population growth and thus rainstorm-induced flood losses are becoming increasingly severe. Low impact development (LID) measures are a storm management technique designed for controlling runoff in urban areas, which is critical for solving urban flood hazard. Therefore, this study developed an exploratory simulation-optimization framework for the spatial arrangement of LID measures. The proposed framework begins by applying a numerical model to simulate hydrological and hydrodynamic processes during a storm event, and the urban flood model coupled with the source tracking method was then used to identify the flood source areas. Next, based on source tracking data, the LID investment in each subcatchment was determined using the inundation volume contribution ratio of the flood source area (where most of the investment is required) to the flood hazard area (where most of the flooding occurs). Finally, the resiliency and sustainability of different LID scenarios were evaluated using several different storm events in order to provide suggestions for flooding predictions and the decision-making process. The results of this study emphasized the importance of flood source control. Furthermore, to quantitatively evaluate the impact of inundation volume transport between subcatchments on the effectiveness of LID measures, a regional relevance index (RI) was proposed to analyze the spatial connectivity between different regions. The simulation-optimization framework was applied to Haikou City, China, wherein the results indicated that LID measures in a spatial arrangement based on the source tracking method are a robust and resilient solution to flood mitigation. This study demonstrates the novelty of combining the source tracking method and highlights the spatial connectivity between flood source areas and flood hazard areas. Further, the framework acts as a strategic tool for the effective spatial arrangement design of LID measures.

Finally, the resiliency and sustainability of different LID scenarios were evaluated using 35 several different storm events in order to provide suggestions for flooding predictions 36 and the decision-making process. The results of this study emphasized the importance 37 of flood source control. Furthermore, to quantitatively evaluate the impact of inundation 38 volume transport between subcatchments on the effectiveness of LID measures, a 39 regional relevance index (RI) was proposed to analyze the spatial connectivity between 40 different regions. The simulation-optimization framework was applied to Haikou City, 41 China, wherein the results indicated that LID measures in a spatial arrangement based 42 on the source tracking method are a robust and resilient solution to flood mitigation. 43 This study demonstrates the novelty of combining the source tracking method and highlights the spatial connectivity between flood source areas and flood hazard areas.
Optimization allows researchers to identify the optimal solution set from a large number 133 of results. However, some research gaps remain. First, as an optimization model is a 134 "black-box" approach, it lowers the confidence of city planners in the optimization 135 results. Second, optimization often leads to non-unique solution sets. Third, previous 136 studies mostly focused on coupled simulation-optimization methods, which normally 137 require large computational burdens (particularly for two-dimensional flood modeling). process that combines flood mitigation strategies with spatial connectivity and uses the 157 regional relevance index (RI) to quantitatively measure the connection between flood 158 source areas and flood hazard areas based on source tracking. By applying this 159 framework to the urban watershed of Haikou (China), we identified the potential 160 prioritization of LID spatial arrangement using source tracking data as a driving force.

161
Furthermore, the framework output is especially important as it highlights the spatial 162 connectivity between the flood source area (requiring most of the LID measures) and 163 the beneficiary area (the areas where flooding is mostly reduced), thereby creating a 164 basis for strengthening cooperation between these areas. The overall framework of the proposed method is illustrated in Fig. 1. First, an 168 urban flood coupled model was established using a hydraulic model, hydrologic model, 169 and source tracking module. Second, using a coupled model, the inundation volume 170 was simulated under typical scenarios combining rainfall and tide level. Third, 171 according to the inundation volume, the regional flood transfer coefficient (A) and RI 172 were calculated for the drainage district. Finally, the spatial arrangement ratio for LID   For example, during a storm event, the urban watershed (as shown in Fig. 2), 189 which consists of three subcatchments (S1, S2, and S3), can flood in response to rapid 190 runoff. The arrows represent the preferred direction of water flow. The runoff generated 191 by subcatchment S1 flows into S2 and is mixed with the inundation volume generated by S2. Subsequently, the inundation volume of S2 divides into two parts. Some of the 193 water flows into S3, while the rest remains in S2. Accordingly, the inundation volume 194 of the flood hazard area is composed of the runoff of each subcatchment. We evaluated 195 the runoff from subcatchments S1, S2, and S3 using tracers A, B, and C, respectively, where C1, C2, and C3 are the initial tracer concentrations for subcatchments S1, S2, and where C1, C2,….Cn are the initial tracer concentrations for subcatchments S1, S2, ….Sn, where A represents the cross-sectional area (m 2 ), l represents the distance along the  The inundation volume contribution from the source area can be quantified using 256 the source tracking data. If the inundation volume in the hazard area comes from 257 multiple subcatchment runoffs, then the regional relevance is strong. Conversely, if the 258 inundation volume comes from a smaller number of subcatchments, the regional 259 relevance is weak. To quantify regional relevance, the regional relevance index (RI) 260 was developed to determine the importance of inundation volume transfer between 261 flood source and hazard areas during urban flood mitigation. The following method can 262 be adopted to quantify the RI for coastal cities. First, the regional flood transfer 263 coefficient (A) is calculated as follows: where Vi,j and Wi,j are the transferred and generated inundation volumes, respectively, However, the calculation of Ai,j must be adjusted as the design periods of rainfall 270 and tide levels do not coincide, and the revision can be resolved in two cases.

271
(1) Regarding rainfall, the following revision should be included:

283
(2) Regarding tide level, the revision is defined as follows: ,,  Generally, the larger the ratio of flood mitigation projects, the higher the total 327 volume detained, the stronger the flood reduction, and thus, the higher the cost. Eqs.
(19)-(21) were adopted to identify the urban flood mitigation plans at a budget 329 constraint.
where Pk is the area of the LID measures in subcatchment k, which was retrofitted with   The main districts of Haikou City (Fig. 3) were selected as the study area. The

357
The data adopted for the urban flood model included rainfall, tide level, digital 358 elevation model data, river data, and pipe network data, which were provided by the 359 Haikou Municipal Water Authority. The urban flood model (Fig. 4) Table 2. Furthermore, with increasing return period, the effective reduction rate of scenario A3 was higher than that of scenario A2. At the return periods of 2 years, 10 years, and 489 20 years, scenario A3 reduced the peak inundation volumes by 7.44%, 9.03%, and 490 13.17%, respectively, as compared with those of scenario A2. This is because with 491 increasing design return period, the RI increases, thereby increasing the regional 492 inundation volume transfer ratio, which makes the flood source area control strategy 493 more effective. This validates the effectiveness of the proposed framework. 506  To quantify regional relevance, a regional relevance index (RI) was developed to 507 determine the importance of inundation volume transfer between flood source and 508 hazard areas during urban flood mitigation. These results show that the regional 509 inundation volume transfer greatly impacts the efficacy of LID measures.