The Influence of Urban Flooding on Residents' Daily Travel

water

Article The Inﬂuence of Urban Flooding on Residents’ Daily Travel: A Case Study of Macau with Proposed Ameliorative Strategies

Kehong Li and Long Zhou *

Faculty of Innovation and Design, City University of Macau, Macau 999078, China * Correspondence: [email protected]; Tel.: +853-85902862

Received: 18 July 2019; Accepted: 29 August 2019; Published: 31 August 2019

Abstract: Climate change has resulted in more extreme weather events in coastal cities, and understanding how daily life is impacted is crucial to make effective adaptation measures. Using Macau as a testbed, this research describes examines the impacts of flooding caused by storm surges on residents’ daily travel and proposes measures to ameliorate disaster risks. Spatial extents of urban floods were modelled through inundation simulations using geographical information data. An analysis of the travel distance increases from residents’ homes to common types of destinations was performed both under normal conditions and during flood events in GIS (Geographic Information System) to assess the influence of urban flooding on residents’ daily travel. The results show that one third of the land is threatened by floods in Macau. People’s average travel distance increases as the warning levels escalate, and travel distance is predicted to rise by up to 64.5%. Based on the findings, the study proposes mitigation strategies to minimize urban flooding’s impacts. It suggests that the area more densely populated is not necessarily the one requiring the deployment of preventative measures with the highest priority, as a traffic analysis is identified as the key area which demands disaster prevention measures.

Keywords: urban ﬂooding; daily travel distance; network analysis; disaster prevention and mitigation; Macau

1. Introduction Changes in global climate and atmospheric circulation frequently cause extreme weather events, such as typhoons, rainstorms, and drought [1], with the resulting harmful effects often exacerbated in cities with high urbanization rates [2]. Coastal cities are particularly subject to storm surges and floods. Research data from the Centre for Research on the Epidemiology of Disasters indicate that natural disasters related to climate change in 2016 caused global economic losses of about USD 66.5 billion, and China ranked second with total losses of about USD 13.6 billion [3]. At present, about 40 million people (0.6% of the global population or one-tenth of the total population of port cities) are currently exposed to a 1 in 100 year coastal flood event [4]. During the past decade, floods were the natural disaster that most affected people, and their intensity and frequency have been increasing [3]. Currently, flood disasters mostly occur in estuary and delta regions, and the Pearl River Delta (PRD) region has the highest risk of urban floods in China [5,6]. In the south of the PRD region, the Macau Special Administrative Region (SAR) has capitalized on its convenient geographical advantages and resources to achieve continuous population and economic growth [7]. Macau is one of the centers in the PRD area with a 100% urbanization rate and a population density (21,055/km2) ranked 1st in the world [8]. However, Macau’s current disaster prevention and mitigation systems are insufficient, and its early warning and prevention

Water 2019, 11, 1825; doi:10.3390/w11091825 www.mdpi.com/journal/water Water 2019, 11, 1825 2 of 21 mechanisms need improvements [9,10]. On average, Macau suffers five-to-six urban flood disasters caused by typhoon and storm surge every year [11]. In 2017, urban floods triggered by Typhoon Hato, the most devastating to hit the PRD in the past 50 years, resulted in the death of ten people, a 1.5 billion dollar loss, and traffic paralysis for three days [8]. Public transport was suspended, and some center streets were completely impassible [12]. Consequently, the Macau government has aimed to improve the city’s capacity to prevent and defend against disasters to protect residents’ life and activities [11]. To this end, Macau urgently needs to develop a sustainable and scientific system to manage urban flooding, especially during the event of typhoon and storm surges. To support these efforts, this study aims to employ a quantitative data analysis on the impacts of urban flooding on Macau residents’ urban travel and to identify the areas that might cause the most severe traffic network problems in disasters. These areas are then prioritized for the installation of preventative and recovery measures to overcome the effects of disasters in order to minimize negative impacts. Risk prevention and adaptation measures are proposed and demonstrated at the end of this research. Factors triggering floods in cities include natural factors, such as rainfall, storm surge and seawater intrusion, and human factors, such as deforestation, drainage system blockage and improper land use. Additionally, population increases accelerate the urbanization process that converts natural lands to built-up hard surfaces with more impervious pavement, lesser infiltration rates, and higher flood peaks and runoff volumes. Floods can have a severe impact on residents’ travel. Severe floods make streets impassable, cause lane closures and reduce transportation system capacity, all of which decrease traffic connectivity and paralyze transport [13]. People have to detour to their destination, which increases their travel distance and results in more traffic congestion [14]. Measures to prevent and mitigate floods include structural measures such as the construction of dams, river dikes, and drainage networks, as well as non-structural measures such as flood forecasting and warning systems [15]. Recent studies on preventing, controlling and evaluating flooding disaster risks have mostly focused on digitalization, technicalization, and informatization [16–19]. Mark (2004) showed how urban flooding could be simulated by one-dimensional hydrodynamic modelling incorporating the interaction between the buried pipe system, the streets (with open channel flow) and the areas flooded with stagnant water. In order to visualize flood extent and impact, the modelling results were presented in the form of flood inundation maps produced in GIS (Geographic Information System) [20]. Mathews (2015) performed a GIS-based spatial analysis of U.S. population census data and siren locations to assess the disaster warning system’s protection coverage in Stillwater, Oklahoma [21]. The findings were used to improve the community’s disaster prevention and early warning capability to reduce disasters’ effects on residents’ lives and properties. Hasnat (2017) used a GIS network analysis to locate the vulnerable areas of Dhaka, the capital of Bangladesh, and vulnerability was assessed by evaluating the by disaster prevention facilities’ emergency response time and the distances to evacuation points [22]. In previous studies, most scholars used urban flooding models to find engineering solutions based on survey sample data without a comprehensive analysis of the overall population. These studies were not effectively supported by social data and could not realistically reflect urban residents’ needs in disaster situations [16]. Understanding how daily life is impacted by natural disasters is crucial to make effective adaptation measures [23]. From a planning point of view, the transport system is the backbone of cities [24], and floods’ impacts on transportation result in more travel distance that disrupts social production, logistics, and business; puts the urban environment under stress; and increases emissions [25]. Low impact development (LID) means systems and practices to manage storm water as part of green infrastructures through infiltrating, filtering, storing, and detaining runoff close to its source [26]. LID has been deemed an innovative approach to prevent flooding in previous studies [27,28]. However, the effectiveness of LID in practice varies depending on a variety of urban design characteristics and patterns [29]. Therefore, this study took a locational and directional approach using hydrological modelling, a GIS-based network analysis, and the social profile of Macau residents to analyze: (i) What are the impacts of urban flooding on Macau’s traffic network; (ii) how is people’s travel distance affected by the flood event; and (iii) where and how the prevention and adaptation Water 2019, 11, x FOR PEER REVIEW 3 of 21

Waterflood2019 levels, 11, 1825and identifies the areas which should be prioritized for the deployment of prevention3 of 21 and adaptation measures. Section 3 explains and discusses the results, and it proposes a prevention measuresand adaptation should bedesign taken, plan. in terms Section of priority 4 summarize and effectiveness.s the research Section and2 describes points out the methodsthe research and modelslimitations. applied to explore the spatial extent affected by different flood levels and identifies the areas which should be prioritized for the deployment of prevention and adaptation measures. Section3 explains2. Materials and and discusses Methods the results, and it proposes a prevention and adaptation design plan. Section4 summarizesThe spatial the researcharea and and percentage points out affected the research by urban limitations. floods on different warning levels were calculated with the coastal flood hazard model [30], Macau geographic information data, and the 2. Materials and Methods digital elevation model. The changes to the residents’ travel distance due to traffic paralysis caused by urbanThe spatialfloods were area andcalculated percentage applying affected a GIS by origin–destination urban floods on di(OD)fferent network warning analysis levels [31]. were A calculatedscenario analysis with the and coastal forecasting flood hazardwere conducted model [30 to], identify Macau geographicthe area that information should be prioritized data, and thefor digitalthe deployment elevation model.of preventative The changes measures. to the residents’ Finally, travelthe study distance developed due to tra affi riskc paralysis prevention caused and by urbanadaptation floods design were calculatedplan as demonstration applying a GIS based origin–destination on a hydrological (OD) analysis network in analysis GIS [32]. [31 The]. A research scenario analysisframework and is forecastingshown in Figure were conducted1. A flood toinundation identify themodel area was that first should applied be prioritized to map the for areas the deploymentinundated by of preventativefloods and calculate measures. the Finally, percentage the study of developedbuildings and a risk roads prevention affected and by adaptation different designwarning plan levels. as demonstration Then, network based analysis on a hydrologicalwas performed analysis to compute in GIS [the32]. residents’ The research shortest framework travel isdistances shown from in Figure home1. to A their flood common inundation destinations model was under first normal applied conditions to map theand areasduring inundated flood events. by floodsThe parish and calculatewith the thehighest percentage priority of for buildings the deployment and roads of aff ectedpreventative by different and warningadaptation levels. measures Then, networkwas identified analysis by a was scenario performed analysis to and compute forecastin the residents’g based on shortest the extent travel to which distances flood from avoidance home toin theireach commonparish scenario destinations would under alleviate normal floods’ conditions influence and on during the city’s flood overall events. road The network. parish with Finally, the highestprevention priority and adaptation for the deployment measures ofwere preventative proposed in and the adaptation parish with measures the highest was priority identified based by on a scenariothe hydrological analysis analysis. and forecasting based on the extent to which flood avoidance in each parish scenario would alleviate floods’ influence on the city’s overall road network. Finally, prevention and adaptation measures were proposed in the parish with the highest priority based on the hydrological analysis.

Figure 1. Research framework. Figure 1. Research framework. 2.1. Study Area 2.1. StudyMacau Area is on the west bank of the PRD region. Its northern border is Zhuhai, Mainland China, and itsMacau three is other on the sides west are bank surrounded of the PRD by region. the South Its northern China Sea border (Figure is 2Zhuhai,). Macau Mainland is close toChina, the shallowand its three side ofother the river,sides are and surrounded severe ﬂooding by the with South storm China surges Sea tends (Figure to occur2). Macau when is local close tropical to the typhoons occur [33].

Water 2019, 11, x FOR PEER REVIEW 4 of 21 shallow side of the river, and severe flooding with storm surges tends to occur when local tropical Water 2019, 11, 1825 4 of 21 typhoons occur [33].

Figure 2. Macau Special Administrative Region (SAR). (a) Parish in Macau; (b) Population density [34,35].

FigureFor several2. Macau centuries, Special MacauAdministrative has been Region a historical (SAR). city (a) inParish which in ChineseMacau; ( andb) Population Portuguese density culture co-exist.[34,35]. Macau comprises the Macau Peninsula on the north, and Taipa and Coloane Island on the south. The Macau Peninsula is Macau’s political, economic, and cultural center where the resident populationFor several and centuries, government Macau are concentrated has been a historical [34]. The Macaucity in Peninsulawhich Chinese occupies and 8.5 Portuguese km2 of Macau’s culture 2 co-exist.total 32.8 Macau km land comprises [35]. There the areMacau 7 main Peninsula parishes on in Macau. the north, Our and Lady Taipa Fatima and (OL) Coloane Parish, St.Island Anthony on the south.(SA) The Parish, Macau St. Lazarus Peninsul (SU)a is Parish, Macau’s Cathedral political, (CA) economic, Parish and and St. Lawrence cultural center (SC) Parish where are the located resident in populationthe Macau and Peninsula. government Taipa are (T) concentrated Parish and St. [34]. Francisco The Macau (SF) Parish Peninsula are located occupies in Taipa 8.5 km and2 of Coloane, Macau’s totalrespectively 32.8 km2 (seeland Figure [35]. 2There). OL Parish,are 7 main SA Parish parishes and in SC Maca Parishu. haveOur higherLady Fatima population (OL) densities, Parish, St. Anthonyand the (SA) west Parish, parts along St. Lazarus the coast (SU) in SC Parish, Parish andCath SAedral Parish (CA) have Parish stronger and propensitiesSt. Lawrence for (SC) flooding Parish due to their lower land elevation and closeness to the sea. Eyewitnesses have stated that floods rose are located in the Macau Peninsula. Taipa (T) Parish and St. Francisco (SF) Parish are located in Taipa from ankle level to chest-high in less than 20 min during Typhoon Hato in 2017 [36]. and Coloane, respectively (see Figure 2). OL Parish, SA Parish and SC Parish have higher population densities,2.2. Simulations and the of west the Spatial parts Scopealong of the Urban coast Flooding in SC Parish and SA Parish have stronger propensities for flooding due to their lower land elevation and closeness to the sea. Eyewitnesses have stated that Simulations were performed using GIS with a high-precision digital elevation model (DEM) of floods rose from ankle level to chest-high in less than 20 minutes during Typhoon Hato in 2017 [36]. Macau (with 3 3 m cell resolution) and geographical information data including buildings and the × roads network obtained from the Macau Cartographic and Topographical Bureau [35]. The simulations 2.2. Simulations of the Spatial Scope of Urban Flooding were used to estimate the spatial range of floods and the percentage of buildings and the roads affectedSimulations by five stormwere performed surge warning using levels GIS (SSW) with (Tablea high-precision1) based on digital the flood elevation inundation model model (DEM) [ 37]. of MacauAccording (with to3 × the 3 m geographic cell resolution) features, and thegeographical historical record information of water data level including and the buildings meteorological and the roadscharacteristics network ofobtained Macau, stormfrom surgethe Macau warnings Cartographic were divided intoand blue,Topographical yellow, orange, Bureau red and [35]. black The simulationsby the Macau were Weather used to Service estimate Bureau the spatial [33]. Based rang one of the floods data weand obtained, the percentage buildings of were buildings categorized and the roadsby type:affected Residential by five (stormRE), commercial surge warning (C), olevelsffice (O (SSW)), school (Table (S), and1) based hospital on (theH). flood The simulation inundation modelmodel [37]. was According operated in to GIS the and geographic generated afeatures, raster map the showing historical land record area that of couldwater belevel inundated and the meteorologicalfor the given watercharacteristics level. The neighborhoodof Macau, storm analysis surge was warnings also operated were in divided GIS to simulate into blue, the floodsyellow, orange,spreading red and from black the coast by the onto Macau the connecting Weather land.Servic Thee Bureau flood water [33]. levelsBased predicted on the data under we di obtained,fferent buildings were categorized by type: Residential (RE), commercial (C), office (O), school (S), and hospital (H). The simulation model was operated in GIS and generated a raster map showing land

Water 2019, 11, x FOR PEER REVIEW 5 of 21

area that could be inundated for the given water level. The neighborhood analysis was also operated

Waterin GIS2019 to, 11 simulate, 1825 the floods spreading from the coast onto the connecting land. The flood 5water of 21 levels predicted under different SSW in Table 1 were the model input parameters. The neighborhood operation identified the land area along the flood flow path with DEM values lower than the input SSWwater in levels. Table1 This were process the model was inputrepeated parameters. continually The within neighborhood a 3 × 3 raster operation cells until identified the full the potential land areainundation along the land flood was flow identified. path with Therefore, DEM values instead lower of thansimply the supposing input water all levels. the cells This lower process than was the repeatedinput parameter continually were within inundated, a 3 3 the raster model cells generated until the full the potential flood inundated inundation area land map was with identified. a logical × Therefore,connecting instead flow path of simply [37]. supposing all the cells lower than the input parameter were inundated, the model generated the flood inundated area map with a logical connecting flow path [37]. Table 1. Surge warning levels (SSW) in Macau. Table 1. Surge warning levels (SSW) in Macau. Storm Surge Warning Levels Water Level Storm SurgeSSW-1/Blue Warning Levels The storm surge water Waterlevel is Level expected to be below 0.5 m. SSW-1SSW-2/Yellow/Blue The The storm storm surge surge water water level level is expected is expected to be to bebetween below 0.5–1.0 0.5 m. m. SSW-2SSW-3/Orange/Yellow The The stormstorm surgesurge waterwater levellevel isis expectedexpected toto bebe betweenbetween 0.5–1.01.0–1.5 m.m. SSW-3SSW-4/Red/Orange The The stormstorm surgesurge waterwater levellevel isis expectedexpected toto bebe betweenbetween 1.0–1.51.5–2.0 m.m. SSW-4SSW-5/Black/Red The stormstorm surgesurge waterwater levellevel isis expectedexpected toto bebe betweenbetween 1.5–2.02.0–3.0 m.m. SSW-5/Black The storm surge water level is expected to be between 2.0–3.0 m. 2.3. Comparison of the Impacts of Urban Flooding by Parish 2.3. Comparison of the Impacts of Urban Flooding by Parish With network analysis in GIS, we could build an origin–destination (OD) cost matrix from multipleWith networkorigins to analysis multiple in GIS,destinations we could [31]. build Using an origin–destination Macau’s road network (OD) cost as the matrix base, from the multiple network originsanalysis to origin–destination multiple destinations matrix [31]. was Using built Macau’s to comp roadute the network residents’ as the shortest base, the travel network distances analysis from origin–destinationhome (origins) to matrixtheir common was built destinations to compute (des thetinations) residents’ under shortest normal travel conditions distances fromand during home (origins)flood events to their using common Equations destinations (1) and (destinations) (2), respectively. under The normal flood conditions inundation and land during simulated flood events above usingwas Equationsinput as the (1) andrestriction (2), respectively. polygon Thebarrier flood in inundation the GIS network land simulated analysis above that wasprohibits input astravel the restrictionanywhere polygonthe polygon barrier interacts in the the GIS road network network—ther analysis thatefore, prohibits people have travel to anywhere travel a longer the polygon distance interactsto destinations the road (Figure network—therefore, 3). people have to travel a longer distance to destinations (Figure3).

Figure 3. Travel route simulation in GIS. (a) Shortest route from origins to destinations under normalFigure condition;3. Travel route (b) shortest simulation route in from GIS. origins(a) Shortest to destinations route from during origins floods to destinations events. under normal condition; (b) shortest route from origins to destinations during floods events. We obtained residential, commercial, office, school and hospital buildings’ patterns and characteristicsWe obtained information residential, from thecommercial, Macau SAR office, Cartography school andand Cadastrehospital Bureaubuildings’ [35]. Inpatterns Equations and (1)characteristics and (2), building information type is from represented the Macau by SARF and Cart theography travel distance and Cadastre from Bureau a residential [35]. In origin Equations to a non-residential(1) and (2), building destination type is represented byby DF. and There the were travel seven distance parishes from under a residential observation origin in thisto a study:non-residentialCA, OL, SA destination, SC, SU, T ,is and representedSF. Travel by under D. There the normal were seven conditions parishes and under during observation flood events in arethis labeledstudy: BCAand, OLA,, SA respectively., SC, SU, T The, and five SF SSW. Travel warning under levels the normal re represented conditions by sand, and during the warning flood events level (blue,are labeled yellow, B orange, and A, respectively. red, and black) The re fiveBL, YSSW, OR warning, R, and Dlevels, respectively. re represented The numbers by s, and of the residential warning originslevel (blue, and travelyellow, destinations orange, red, (C ,andO, S black), and Hre) useBL, mY, andOR,n R, respectively., and D, respectively. Based on The these numbers symbols, of FresidentialC, O, H, originsS , s BLand, Y travel, OR, R destinations, D and P (CCA, O, OL, S,, andSA, SCH), useSU, mT, andSF . n, respectively. Based on these symbols,∈ { 𝐹∈} 𝐶,𝑂,𝐻,𝑆∈ { , 𝑠 ∈ 𝐵𝐿,𝑌,𝑂𝑅,𝑅,𝐷} ∈ { and 𝑃∈𝐶𝐴, 𝑂𝐿,} 𝑆𝐴, 𝑆𝐶, 𝑆𝑈,𝑇,𝑆𝐹. 1 XmBnB FDsB = FiDj, F C, O, H, S , s BL, Y, OR, R, D (1) mB nB i=1 j=1 ∈ { } ∈ { } × Water 2019, 11, 1825 6 of 21

1 XmAnA FDsA = FiDj, F C, O, H, S , s BL, Y, OR, R, D (2) m n i=1 j=1 ∈ { } ∈ { } A × A FD FDsB LITDI = sA − 100%, F C, O, H, S , s BL, Y, OR, R, D (3) FDsB × ∈ { } ∈ { } A comparative analysis and increase percentage (LITDI) were performed evaluating the travel distances under normal conditions (FDsB) and during flood events (FDsA) at the five levels of intensity using Equation (3). Equation (4) was used to estimate the individual contributions of floods in each of the seven parishes to the overall network travel distances increase and the scenarios of flood avoidance in each of the seven parishes to alleviate the overall network paralysis during the flooding events. Equation (5) was used to single out the parish to apply prevention with first priority that would most alleviate the overall network paralysis.

= 1 PmBnB FDsP m n i=1 j=1 FiDj, F C, O, H, S , s BL, Y, OR, R, D , B× B ∈ { } ∈ { } (4) P CA, OL, SA, SC, SU, T, SF ∈ { }

PP = Max(FD FDsP), F C, O, H, S , s BL, Y, OR, R, D , sA − ∈ { } ∈ { } (5) P CA, OL, SA, SC, SU, T, SF ∈ { } Based on the scenarios analysis and forecasting, the parish with the highest priority for the deployment of preventative and adaptation measures was used to build a demonstration proposal based on the hydrological analysis in GIS.

3. Results and Discussion

3.1. Spatial Extent and Percentage Affected by Urban Floods under Different Warning Categories Flooding simulations were performed based on the high-precision digital elevation model (DEM) (Figure4a) and the hydrological analysis on di fferent storm surge warning categories provided by the Macau Cartography and Cadastre Bureau [35]. Figure4b shows that the area a ffected by SSW-1 storm surge flood is on the west part of the Macau Peninsula and in the middle of Taipa. The total area is 315,189.98 m2 (0.315 km2), which accounts for 0.9% of the total land in Macau. The affected areas simulated under SSW-2 storm surge flood would cover 1,076,183.78 m2 (1.076 km2), which accounts for 3.2% of the total land in Macau (Figure4c). Areas in addition to these affected by SSW-1 would be in the northern part of the Macau Peninsula. In the event of SSW-3, the size of the affected area would increase to 2,138,833.10 m2 (2.138 km2) (Figure4d), which accounts for 6.5% of the total land in Macau. The western part of Coloane would be affected in addition to the areas affected by SSW-1 and SSW-2. However, Taipa’s affected areas would not be substantially different from the areas affected by SSW-2. This might be attributed to the new urban drainage system and a higher percentage of green space [38]. Figure4e shows that the areas affected by SSW-4 storm surge floods are located on the eastern part of the Macau Peninsula and western Taipa in addition to those affected by SSW-1, SSW-2, and SSW-3 events. The total area is 5,307,022.97 m2 (5.307 km2), which accounts for 16.1% of the total land in Macau. Figure4f shows that the areas affected by SSW-5 storm surge flood are located on the southern part of the Macau Peninsula and on the west of Taipa. The total area is up to 9,560,885.68 m2 (9.561 km2), which accounts for 29.1% of the total land in Macau. Water 2019, 11, 1825 7 of 21 Water 2019, 11, x FOR PEER REVIEW 7 of 21

Figure 4. Spatial extent affected by urban floods. (a) Digital elevation model (DEM) of buildings and roads in Macau [35]; (b) SSW Level 1 (blue) flooding simulation results (0–0.5 m floods); (c) SSW Level 2

(yellow) flooding simulation results (0.5–1 m floods); (d) SSW Level 3 (orange) flooding simulation results (1–1.5 m floods); (e) SSW Level 4 (red) flooding simulation results (1.5–2.5 m floods); (f) SSW Level 5 (black) flooding simulation results (2.5–3.0 m floods). Water 2019, 11, x FOR PEER REVIEW 8 of 21

Figure 4. Spatial extent affected by urban floods. (a) Digital elevation model (DEM) of buildings and roads in Macau [35]; (b) SSW Level 1 (blue) flooding simulation results (0–0.5 m floods); (c) SSW Level 2 (yellow) flooding simulation results (0.5–1 m floods); (d) SSW Level 3 (orange) flooding simulation Water 2019, 11, 1825 8 of 21 results (1–1.5 m floods); (e) SSW Level 4 (red) flooding simulation results (1.5–2.5 m floods); (f) SSW Level 5 (black) flooding simulation results (2.5–3.0 m floods). Figure5 shows the number of buildings located in the areas inundated by flood water under the five SSWFigure levels 5 shows considered the number in this of study. buildings Overall, located the ratein the of areas buildings inundated of all types by flood affected water by under the flood the eventfive SSW increased levels inconsidered the same mannerin this study. than the Overall, impact the rate rate of theof buildings SSW event. of Oallffi cetypes buildings affected were by the mostflood aeventffected, increased increasing in fromthe same 7 to 145manner buildings than (fromthe impact about rate 3% to of about the SSW 56% event. of the total)Office as buildings the SSW increasedwere the most in severity affected, from increasing 1 to 5. Commercial from 7 to 145 buildings buildings were (from second about most 3% aff toected, about increasing 56% of the from total) 18 toas 125the surroundedSSW increased by floodin severity water (fromfrom 1 about to 5. 7%Commercial to about 51% buildings of the total).were second Residential most buildings affected, rankedincreasing third, from increasing 18 to 125 from surrounded 113 to 589 buildingsby flood (fromwater about(from 8% about to 40% 7%of to the about total). 51% The of numbers the total). of aResidentialffected hospitals buildings and ranked schools third, were similar,increasing increasing from 113 from to 1589 to buildings 7 (from about (from 4% about to 29% 8% of to the 40% total) of andthe total). from 1The to 20 numbers (from about of affected 1% to26% hospitals of the and total), schools respectively. were similar, Figure 6increasing shows the from number 1 to of 7 roads(from locatedabout 4% in to the 29% areas of the inundated total) and by from flood 1 water, to 20 (from increasing about from 1% to 194 26% to 3508of the roads total), with respectively. higher rate Figure from SSW6 shows 1 to the 5 (from number about of roads 3% to located about 46% in the of theareas total), inundated and the by increasing flood water, rate increasing was higher from when 194 the to warning3508 roads levels with escalate higher from rate 3from to 5. SSW 1 to 5 (from about 3% to about 46% of the total), and the increasing rate was higher when the warning levels escalate from 3 to 5.

Water 2019, 11Figure, x FOR 5. PEER Numbers REVIEW of buildings (grouped by type) a ffectedﬀected by ﬂoodingflooding atat each SSWSSW level.level. 9 of 21

Figure 6. Number of roads aﬀected by ﬂooding at each SSW level. Figure 6. Number of roads affected by flooding at each SSW level.

3.2. Impact of Urban Floods on Residents’ Distance of Travel Analyzed for Each Parish The distance of travel plays an important role in assessing the ease of movement [39,40]. In this study, the simulated total travel distances from residential origins to daily destinations gradually increased as the SSW level rose in each parish (Figure 7a–g and Appendix A). This finding shows that more residents need to detour from their regular route when travelling to their destinations as the situation worsens, which indicates deterioration in the ease of movement.

Travel distance for residents of Cathedral Travel distance for residents of St. Lawrence Parish (feet) Parish (feet) 38,000 40,000 36,000 38,000 34,000 36,000 32,000 34,000 32,000 30,000 30,000 28,000 28,000 26,000 26,000 24,000 24,000 22,000 22,000 20,000 20,000 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5

C H O S C H O S

(a) (b)

Water 2019, 11, 1825 9 of 21

3.2. Impact of Urban Floods on Residents’ Distance of Travel Analyzed for Each Parish The distance of travel plays an important role in assessing the ease of movement [39,40]. In this study, the simulated total travel distances from residential origins to daily destinations gradually increased as the SSW level rose in each parish (Figure7a–g and AppendixA). This ﬁnding shows that Water 2019, 11, x FOR PEER REVIEW 10 of 22 more residents need to detour from their regular route when travelling to their destinations as the situation worsens, which indicates deterioration in the ease of movement.

C H O S C H O S

(a) (b)

Travel distance for residents of Our Lady Travel distance for residents of St. Anthony Fatima Parish (feet) Parish (feet) 38,000 36,000 36,000 34,000 34,000 32,000 32,000 30,000 30,000 28,000 28,000 26,000 26,000 24,000 24,000 22,000 22,000 20,000 20,000 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5

C H O S C H O S

Figure 7. Cont.

Water 2019, 11, 1825 10 of 21 Water 2019, 11, x FOR PEER REVIEW 11 of 22

Travel distance for residents of Taipa Parish Travel distance for residents of St. Francisco (feet) Xavier Parish (feet) 62,000 84,000 58,000 81,000 54,000 78,000 50,000 75,000 46,000 72,000 42,000 69,000 38,000 66,000 34,000 63,000 30,000 60,000 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5

C H O S C H O S

(e) (f)

Travel distance for residents of St. Lazarus Parish （feet）

32,000 30,000

28,000

26,000 24,000

22,000

20,000 B SSW- SSW- SSW- SSW- SSW- 1 2 3 4 5

C H O S

(g)

Figure 7. Travel distances from residential areas to commercial, office, educational, and hospital Figure 7. Travel distances from residential areas to commercial, oﬃce, educational, and hospital destinations by parish. (a) Cathedral Parish; (b) St. Lawrence Parish; (c) Our Lady Fatima Parish; (d) destinations by parish. (a) Cathedral Parish; (b) St. Lawrence Parish; (c) Our Lady Fatima Parish; (d) St. St. Anthony Parish; (e) Taipa Parish; (f) St. Francisco Xavier Parish; and (g) St. Lazarus Parish. Anthony Parish; (e) Taipa Parish; (f) St. Francisco Xavier Parish; and (g) St. Lazarus Parish.

TravelTravel distances distances for for the the residents residents in in CathedralCathedral ParishParish would would be be slightly slightly affected affected under under SSW-1, SSW-1, SSW-2, and SSW-3 conditions; however, starting at SSW-4, travel distances to schools and offices SSW-2, and SSW-3 conditions; however, starting at SSW-4, travel distances to schools and offices would would be particularly affected (Figure 7a). This might be attributed to the fact that the elevation of be particularly affected (Figure7a). This might be attributed to the fact that the elevation of Cathedral Cathedral Parish is high, but large sections are occupied by streets and parking lots with low or no Parish is high, but large sections are occupied by streets and parking lots with low or no permeability, permeability, and streets can easily become flooded and impassable, which restricts residents’ ability and streets can easily become flooded and impassable, which restricts residents’ ability to travel [40]. to travel [40]. InIn St. St. Lawrence Lawrence Parish, Parish, all all types types of travel of travel would would be affected be aff wheneverected whenever storm surges storm occur surges (Figure occur (Figure7b). 7Theb). Thetravel travel distance distance would would increase increase steadily steadily with with the the flood flood level. level. Travelling Travelling to to commercial commercial buildingsbuildings and and schools schools would would be more be amoreffected, affected, probably probably because because of the lack of ofth commerciale lack of commercial destinations anddestinations schools in St.and Lawrence schools in Parish St. Lawrence and its poor Parish road and network its poor connectivity road network to otherconnectivity parishes to [41other]. parishesResidents [41]. of Our Lady Fatima Parish face similar travel problems across the SSW levels (Figure7c). Travel distancesResidents of would Our Lady gradually Fatima increase Parish face as floodingsimilar travel worsened, problems particularly across the SSW to commercial levels (Figure and school7c). Travel destinations. distances This would finding gradually might increase be related as flooding to the parish’s worsened, relatively particularly higher to population commercial density and andschool construction destinations. density This compared finding might to the be otherrelated parishes to the parish’s [42,43]. relatively Consequently, higher population storm water density rapidly

Water 2019, 11, 1825 11 of 21 accumulates, and flooding immediately follows [44]. If storm surges escalate, water easily inundates this parish and impedes residents’ travel. Flooding in St. Anthony Parish would not seriously affect the daily travel route of its residents with floods levels between SSW-1 and SSW-3 (Figure7d). However, at SSW-4 or higher, flooding would seriously interfere with travel, particularly the travel to school destinations. St. Anthony is almost as densely populated as Our Lady Fatima [38]. During the twentieth century, Chinese merchants wanted to improve land-use efficiency to stimulate socioeconomic developments. This led to an overwhelming construction of commercial and enclosed buildings at Porto Interior in St. Anthony Parish [45]. During a storm surge at all levels, storm water would rapidly flow towards the relatively low-lying areas (including Patane and the vicinities of Rua de Fai Chi Kei, Mercado Vermelho, and Porto Interior). Flooding and road closures would ensue and interfere with residents’ travel. The situation in Taipa and St. Francisco Xavier Parishes are similar (Figure7e,f). In both cases, when the areas are affected by storm surges with levels between SSW-1 and SSW-4, the impacts would not be particularly serious. However, at SSW-5, daily routines would be seriously disrupted. Both parishes are on the outlying islands, and the residents depend on the Macau-Taipa, Amizade, and Sai Van Bridges to connect to the Macau Peninsula. During a storm surge at SSW-5, urban floods impede the travel activities of residents of both parishes to the Macau Peninsula through bridges. As shown in Figure7g, buildings and roads in St. Lazarus Parish would be rarely a ffected by floods, which could be attributed to its mountainous terrain and higher elevation [33]. However, flooding at SSW-4 and SSW-5 would restrict residents’ travel from St. Lazarus Parish to destinations in other parishes. The rapidly increasing travel distance for people living in St. Lazarus Parish indicates that these residents’ travel activities would mostly be restricted by road conditions on the ways to destinations in other parishes.

3.3. Parishes Prioritization to Install Preventative Measures to Alleviate The Impacts of Urban Floods on Daily Travel Risk prevention measures had to be urgently applied in Macau after Typhoon Hato [9]. The simulation results revealed that all the parishes would be affected by flooding except for St Lazarus Parish (Figure4). To alleviate the impacts of urban floods, the key task consists of the rational organization of the parishes’ prevention applications in order to maximize efficiency to prevent floods. To that end, this study employed scenario analysis and forecasting, and it assigned ranking scores to the seven parishes based on the extent to which flood avoidance in each parish scenario would alleviate floods’ influence on the city’s overall roads network based on Equation (4). The ranking score ranged from one through six, in which one and six represented the lowest and highest alleviation need of the roads network paralysis (travel distance), respectively. The parish with the higher cumulative score would be more in more urgent need of flood prevention measures and vice versa. The assessment result is shown in Table2. The results show the extent in each scenario to alleviate roads network paralysis, from most to least: CA (155), T (126), SF (102), SA (83), SC (68), and OL (54). Therefore, Cathedral Parish was identified as the area with the highest need for the deployment of prevention and adaptation measures to alleviate the impacts of floods. This indicates that, contrary to the results discussed in other studies [46,47], the more densely populated area is not necessarily the one requiring the more urgent deployment of prevention measures. It also demonstrates that traffic analysis is an important factor to consider for the identification of the key area requiring disaster prevention. Water 2019, 11, 1825 12 of 21

Table 2. Assessment of priority parish to install prevention measures.

Scenario Analysis and Forecasting CA SC OL SA T SF Origin- Ranking Ranking Ranking Ranking Ranking Ranking Destination Score (Travel Score (Travel Score (Travel Score (Travel Score (Travel Score (Travel Alleviation Distance Distance Distance Distance Distance Distance Increase Increase Increase Increase Increase Increase Alleviation) Alleviation) Alleviation) Alleviation) Alleviation) Alleviation) CA- C 6 (10,582 ) 3 (6416 ) 1 (5715 ) 4 (6039 ) 5 (8093 ) 2 (5965 ) ↓ ↓ ↓ ↓ ↓ ↓ H 6 (9921 ) 2 (5711 ) 1 (2847 ) 3 (5813 ) 5 (7329 ) 4 (5878 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (10,139 ) 4 (6263 ) 2 (6074 ) 3 (6196 ) 5 (7720 ) 1 (5984 ) ↓ ↓ ↓ ↓ ↓ ↓ S 6 (9098 ) 2 (4981 ) 1 (4462 ) 3 (4992 ) 4 (6084 ) 5 (7254 ) ↓ ↓ ↓ ↓ ↓ ↓ SC- C 6 (9939 ) 3 (6660 ) 2 (6414 ) 4 (7354 ) 5 (9543 ) 1 (3669 ) ↓ ↓ ↓ ↓ ↓ ↓ H 6 (9624 ) 3 (6396 ) 1 (2777 ) 4 (7189 ) 5 (8566 ) 2 (3555 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (9388 ) 2 (5945 ) 3 (6229 ) 4 (7092 ) 5 (8505 ) 1 (3505 ) ↓ ↓ ↓ ↓ ↓ ↓ S 6 (8183 ) 2 (5041 ) 3 (5132 ) 4 (5925 ) 5 (7477 ) 1 (4666 ) ↓ ↓ ↓ ↓ ↓ ↓ OL- C 5 (3426 ) 1 (1983 ) 2 (2458 ) 3 (2839 ) 4 (3315 ) 6 (3448 ) ↓ ↓ ↓ ↓ ↓ ↓ H 4 (2831 ) 1 (1366 ) 2 (1607 ) 3 (2708 ) 5 (3096 ) 6 (3539 ) ↓ ↓ ↓ ↓ ↓ ↓ O 2 (3136 ) 1 (2541 ) 4 (3440 ) 3 (3420 ) 6 (3838 ) 5 (3620 ) ↓ ↓ ↓ ↓ ↓ ↓ S 4 (1615 ) 1 (539 ) 2 (1153 ) 5 (1976 ) 3 (1609 ) 6 (4547 ) ↓ ↓ ↓ ↓ ↓ ↓ SA- C 6 (6113 ) 3 (3782 ) 1 (2714 ) 2 (2934 ) 4 (3991 ) 5 (4019 ) ↓ ↓ ↓ ↓ ↓ ↓ H 6 (5540 ) 2 (2680 ) 1 (1026 ) 3 (2834 ) 4 (3566 ) 5 (3849 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (5597 ) 4 (4034 ) 2 (3419 ) 1 (3005 ) 3 (3957 ) 5 (4052 ) ↓ ↓ ↓ ↓ ↓ ↓ S 5 (4257 ) 4 (2172 ) 1 (1169 ) 3 (1958 ) 2 (1702 ) 6 (4926 ) ↓ ↓ ↓ ↓ ↓ ↓ T- C 6 (16,607 ) 4 (8073 ) 3 (7952 ) 2 (7725 ) 5(12,497 ) 1 (5075 ) ↓ ↓ ↓ ↓ ↓ ↓ H 6 (14,918 ) 2 (7429 ) 1 (233 ) 3 (7778 ) 5(10,531 ) 4 (7877 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (16,893 ) 3 (8068 ) 2 (7683 ) 4 (8080 ) 5 (12,031 ) 1 (5008 ) ↓ ↓ ↓ ↓ ↓ ↓ S 6 (17,044 ) 4 (8647 ) 2 (8243 ) 3 (8602 ) 5 (12,776 ) 1 (7257 ) ↓ ↓ ↓ ↓ ↓ ↓ SF- C 6 (13,079 ) 3 (6907 ) 4 (7731 ) 2 (6730 ) 5 (8589 ) 1 (6630 ) ↓ ↓ ↓ ↓ ↓ ↓ H 6 (23,420 ) 2 (6134 ) 1 (4718 ) 3 (6470 ) 4(18,912 ) 5(19,628 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (13,300 ) 1 (6863 ) 3 (7045 ) 2 (6988 ) 5 (8597 ) 4 (7608 ) ↓ ↓ ↓ ↓ ↓ ↓ S 6 (15,860 ) 1 (8258 ) 4 (9267 ) 2 (9007 ) 5(11,181 ) 3 (9050 ) ↓ ↓ ↓ ↓ ↓ ↓ SU- C 6 (4880 ) 3 (2716 ) 1 (2224 ) 2 (2258 ) 4 (3329 ) 5 (3337 ) ↓ ↓ ↓ ↓ ↓ ↓ H 4 (3117 ) 2 (1964 ) 1 (522 ) 3 (2232 ) 5 (3391 ) 6 (4518 ) ↓ ↓ ↓ ↓ ↓ ↓ O 6 (4485 ) 3 (3027 ) 2 (2,893 ) 1 (2837 ) 5 (3558 ) 4 (3335 ) ↓ ↓ ↓ ↓ ↓ ↓ S 5 (3135 ) 2 (1276 ) 1 (798 ) 4 (1447 ) 3 (1319 ) 6 (4306 ) ↓ ↓ ↓ ↓ ↓ ↓ Total Score Prevention and 155 68 54 83 126 102 Adaptation 1 5 6 4 2 3 Priority Order

3.4. Floods Prevention and Adaption Plan Demonstration The low impact development (LID) approach is highly recommended in published studies and reports as an effective set of techniques for the prevention and control of urban floods [48,49]. LID includes sunken green spaces, vegetation ditches, storm water wetlands, and permeable pavement. [50]. Based on results of the hydrological analysis, the scenario forecasting analysis, and the field examination in Macau, this study proposes an LID plan for CA Parish as a way to combat the impact of floods. In Cathedral Parish, the land surfaces are mostly hard pavement, structural rooftops and green parks (Figure8). Generally, there is a clear di fference in development between the newer and older urban districts in Macau [35]. In Cathedral Parish, the older urban district is in the northwestern region, and the new urban district is along the water on the south. The development and building densities are high, resulting in high proportions of impervious pavement and rooftops. In this context, residential, office, and commercial blocks need to reduce storm water discharge by implementing flood prevention measures. Water 2019, 11, x FOR PEER REVIEW 13 of 21

urgent deployment of prevention measures. It also demonstrates that traffic analysis is an important factor to consider for the identification of the key area requiring disaster prevention.

3.4. Floods Prevention and Adaption Plan Demonstration The low impact development (LID) approach is highly recommended in published studies and reports as an effective set of techniques for the prevention and control of urban floods. [48,49]. LID includes sunken green spaces, vegetation ditches, storm water wetlands, and permeable pavement. [50]. Based on results of the hydrological analysis, the scenario forecasting analysis, and the field examination in Macau, this study proposes an LID plan for CA Parish as a way to combat the impact of floods. In Cathedral Parish, the land surfaces are mostly hard pavement, structural rooftops and green parks (Figure 8). Generally, there is a clear difference in development between the newer and older urban districts in Macau [35]. In Cathedral Parish, the older urban district is in the northwestern region, and the new urban district is along the water on the south. The development and building densities are high, resulting in high proportions of impervious pavement and rooftops. In this context, residential, office, and commercial blocks need to reduce storm water discharge by implementing flood prevention measures. A hydrological analysis was performed in GIS, and the source, channels, and endpoints of the surface runoff in Cathedral Parish were identified for the application of LID techniques. The LID demonstration plan proposed is shown in Figures 9 and 10. Subsection 3.4.1 and 3.4.2 describe the Water 2019, 11, 1825 13 of 21 proposals at the points of storm water generations and along the flow channels with details, respectively.

Water 2019, 11, x FOR PEER REVIEW 14 of 21 FigureFigure 8.8. Map of the existing surface types of of Cathedral Cathedral Parish. Parish. such as the Sai Van Lake Square, Garden of the Arts, Ferreira Amaral Square, and Hong Kung Temple Square.3.4.1.A LID hydrological These Technique features analysis Proposal enable was atthe Storm performed collection, Water instoring, Runoff GIS, and Pointsand the purificati source,on channels, of storm andwater. endpoints Storm water of the surfacerunoff from runo ffrooftopsin Cathedral could be Parish conveyed were identifiedto these ponds for the via application rainwater downspouts of LID techniques. and piping The [51]. LID demonstrationThe largeRain gardensbio-retention plan were proposed pondsadded fitto is theinto shown map the innewin Figuresthe and Ma lacau9rge-scale and Tower 10. publicConvention Sections space 3.4.1 and in andthe Entertainment south 3.4.2 ofdescribe Cathedral Center the vicinity; they were also added to Avenida do Infante D. Henrique, Dynasty Plaza, Macau Science proposalsParish, and at thethey points provide of storm high water ecological, generations aestheti andc alongand amenity the flow benefits channels to with the details,local commercial respectively. property.Center, and the A-Ma Temple. The advantage of rain gardens is managing storm water runoff and reduce runoff pollution onsite with inexpensive installation costs in some old neighborhoods with many low-income families and deficient public storm water facilities in the north of Cathedral Parish. Larger bio-retention ponds were placed in areas with low-lying terrain (i.e., the runoff endpoints),

FigureFigure 9. 9. LowLow impactimpact developmentdevelopment (LID) technique propos proposalal at at storm storm water water onsite onsite runoff runoﬀ pointspoints and and channelschannels in in CathedralCathedral Parish.Parish.

At the renovated midstream runoff points (such as Avenida da Amizade, Avenida do Infante D. Henrique, and Rua do Guimaraes), sunken vegetation spaces were proposed to arrange as scattered strips for storm water to permeate and be transported farther downstream. The layout and design of these sunken green spaces would include storm water storage and infiltration space on roadsides. Usually, inception gullies are in sunken green spaces to ensure storm water overflow discharge [52– 54]. The sunken vegetation is low cost and easy to incorporate into existing roadside landscaping, and its daily maintenance can be carried out at the same time by general landscape management, which is effective in Macau where labor cost is much higher than most other cities in the PRD region [55].

3.4.2. LID Technique Proposal along Storm water Runoff Channels and on Hard Surfaces in Cathedral Parish. Permeable brick pavements would be used at the square, on sidewalks, and roads along the runoff channels with relatively low loading capacities [56]. Cobble stone, gravel, or reinforced grass could be used at Alameda Dr. Carlos D’Assumpcao, Garden of the Arts, and the automobile lanes at Ferreira Amaral Plaza and in the vicinities of Dynasty Plaza. Macau Science Center is proposed to be paved with permeable asphalt or concrete (Figures 9 and 10). Perforated pipes connected to drainage pipes would be installed beneath these permeable road surfaces. Due to the subtropical climate and high population density, the weather is generally hot from April to September in Macau [57]. The permeable pavement allows the soil underneath to breathe, reduces heat islands, and brings temperatures down.

Water 2019, 11, 1825 14 of 21 Water 2019, 11, x FOR PEER REVIEW 15 of 21

FigureFigure 10.10. LID technique proposal along the the storm storm wa waterter runoff runoff channelschannels in in Cathedral Cathedral Parish. Parish. 3.4.1. LID Technique Proposal at Storm Water Runoff Points To direct storm water appropriately, road intersections should be designed along the transverse slope,Rain and gardens vertical wererelationships added toshould the mapbe establ in theished Macau between Tower the Conventionroad surfaces and and Entertainment the roadside Centervegetation vicinity; strips. they Curb were gaps also should added be toadded, Avenida and dodrainage Infante troughs D. Henrique, should Dynastybe installed Plaza, on roads’ Macau Sciencesurfaces Center, to facilitate and the rapid A-Ma storm Temple. water The channeling advantage into of rain LID gardens facilities is [58,59]. managing The storm overflow water height runoff andshould reduce be runocontrolledff pollution to limit onsite the withdepth inexpensive of retained installation storm water, costs and in some intercepting old neighborhoods embankments with manyshould low-income reduce the families impacts and of deficientroads’ longitudinal public storm slopes water on facilities the extent in the of north storm of water Cathedral retention. Parish. LargerMoisture-resistant bio-retention and ponds drought-resistant were placed in vegetati areas withon should low-lying be terrainplanted (i.e., in roadside the runo ffstrips.endpoints), Bio- suchretention as the ponds Sai Van and Lake shallow Square, vegetation Garden ditches of the Arts,should Ferreira be established Amaral Square,in low-lying and Hongareas where Kung Templestorm Square.water collects, These features and vegetation enable the ditches collection, should storing, be created and purification in parking oflots storm [60,61]. water. Storm water runoff from rooftops could be conveyed to these ponds via rainwater downspouts and piping [51]. The large bio-retention4. Conclusions ponds fit into the new and large-scale public space in the south of Cathedral Parish, and theyScientific provide and high sustainable ecological, management aesthetic and strategi amenityes benefitsare needed to the to localprotect commercial Macau from property. floods disastersAt the in renovated the current midstream context of runoglobalff points climate (such change as Avenidaand frequent da Amizade, extreme Avenidaweather events. do Infante This D. Henrique,study quantitatively and Rua do examined Guimaraes), thesunken impacts vegetation of urban spacesfloods wereon residents’ proposed travel to arrange in Macau. as scattered The stripssimulation for storm results water revealed to permeate that up andto one be transportedthird of the land farther is threatened downstream. by floods The layout under and different design ofstorm these surge sunken warnings, green spaces and the would travel include distances storm between water residential storage and areas infiltration and common space destinations on roadsides. Usually,increase inception by up to 64.5% gullies as are storm in sunken surge warning green spaces levels to escalate. ensurestorm The scenarios water overflow analysis discharge and forecasting [52–54 ]. Thereveal sunken that vegetationthe areas with is low higher cost andpopulation easy to incorporatedensities are into not existingnecessarily roadside the ones landscaping, requiring andmost its dailyurgently maintenance the installation can be of carried preventative out at themeasures. same time We demonstrated by general landscape that traffic management, network analysis which is is etheffective key area in Macau to tackle where to laborensure cost disaster is much prevention higher than in terms most otherof alleviating cities in overall the PRD traffic region paralysis. [55]. The analysis described here was based on Macau’s digital elevation, hydrological analysis, and 3.4.2.network LID analysis Technique models, Proposal as well along as geographic Storm Water info Runormationff Channels data. It andprovides on Hard scientific Surfaces evidence in and Cathedral Parish support in decision-making and the design of flood prevention measures which could be recommendedPermeable to brick manage pavements flood risk would in other be used coastal at the cities square, where on sidewalks,there are limited and roads land along resources. the runo ff channelsDue with to the relatively lack of data, low loadingthere are capacities some limitations [56]. Cobble with stone,this study. gravel, First, or reinforcedthe use of grassa DEM could and a be usedhydrological at Alameda analysis Dr. Carlos model D’Assumpcao, to simulate flooding Garden disaster of the Arts,risks andon the the city automobile scale cannot lanes accurately at Ferreira Amaralexpress Plazathe total and extent in the vicinitiesof the affected of Dynasty areas Plaza.in Macau Macau in every Science case, Center because is proposed urban flooding to be paved is withinfluenced permeable by numerous asphalt or factors concrete in (Figuresaddition9 toand variations 10). Perforated in topography, pipes connected such as towind drainage speed pipes and underground drainage systems. Improved precision with a fine scale in the hydrodynamic models

Water 2019, 11, 1825 15 of 21 would be installed beneath these permeable road surfaces. Due to the subtropical climate and high population density, the weather is generally hot from April to September in Macau [57]. The permeable pavement allows the soil underneath to breathe, reduces heat islands, and brings temperatures down. To direct storm water appropriately, road intersections should be designed along the transverse slope, and vertical relationships should be established between the road surfaces and the roadside vegetation strips. Curb gaps should be added, and drainage troughs should be installed on roads’ surfaces to facilitate rapid storm water channeling into LID facilities [58,59]. The overﬂow height should be controlled to limit the depth of retained storm water, and intercepting embankments should reduce the impacts of roads’ longitudinal slopes on the extent of storm water retention. Moisture-resistant and drought-resistant vegetation should be planted in roadside strips. Bio-retention ponds and shallow vegetation ditches should be established in low-lying areas where storm water collects, and vegetation ditches should be created in parking lots [60,61].

4. Conclusions Scientific and sustainable management strategies are needed to protect Macau from floods disasters in the current context of global climate change and frequent extreme weather events. This study quantitatively examined the impacts of urban floods on residents’ travel in Macau. The simulation results revealed that up to one third of the land is threatened by floods under different storm surge warnings, and the travel distances between residential areas and common destinations increase by up to 64.5% as storm surge warning levels escalate. The scenarios analysis and forecasting reveal that the areas with higher population densities are not necessarily the ones requiring most urgently the installation of preventative measures. We demonstrated that traffic network analysis is the key area to tackle to ensure disaster prevention in terms of alleviating overall traffic paralysis. The analysis described here was based on Macau’s digital elevation, hydrological analysis, and network analysis models, as well as geographic information data. It provides scientific evidence and support in decision-making and the design of flood prevention measures which could be recommended to manage flood risk in other coastal cities where there are limited land resources. Due to the lack of data, there are some limitations with this study. First, the use of a DEM and a hydrological analysis model to simulate flooding disaster risks on the city scale cannot accurately express the total extent of the affected areas in Macau in every case, because urban flooding is influenced by numerous factors in addition to variations in topography, such as wind speed and underground drainage systems. Improved precision with a fine scale in the hydrodynamic models (such as the storm water management model) [62–64] might be useful for future research and confirmation. Second, this study used the mean distance from home to destinations in the parishes to measure the residents’ travel distance. A more realistic approach might be needed to analyze transportation modes, traffic flow, congestion during flood events and destinations to other types of buildings. All of these call for future studies.

Author Contributions: K.L. provided the initial concept, data collection, and wrote the manuscript. L.Z. provided the research design, analysis approach and revision of the manuscript. Funding: This research was funded by National Natural Science Foundation of China (No. 51708399); Macao Foundation (MF 1814) and Macao Cultural Affairs Bureau Academic Research Grant (No. 2018). The APC was funded by City University of Macau Foundation. Acknowledgments: The authors acknowledge the three anonymous reviewers for their constructive comments, which have helped to improve the paper significantly. Conflicts of Interest: The authors declare no conflict of interest. Water 2019, 11, 1825 16 of 21

Abbreviations

A Travel Conditions During Flood Events BL Blue CA Cathedral D Travel Distance F Building’s functional Travel Distances During Flood Events at Five Levels FD sA of Intensity LID Low Impact Development m Numbers of Residential Origins OOﬃce OL Our Lady Fatima P Parish PRD Pearl River Delta RE Residential s Five SSW Warning Levels SC St. Lawrence Parish SU St. Lazarus Parish SSW Storm Surge Warning Y Yellow B Travel Conditions Under Normal Condition C Commercial D Black DEM Digital Elevation Model FDsB Travel Distances Under Normal Condition H Hospital LITDI Travel Distance Increase Percentage n Numbers of Travel Destinations OD Origin–Destination OR Orange PP Prevention Priority Order R Red S School SA St. Anthony Parish SF St. Francisco Parish SAR Special Administrative Region T Taipa Parish

Appendix A Travel distances under normal conditions and diﬀerent storm surge warnings.

Table A1. Travel distance from residents in Cathedral Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 21,035 21,189 21,012 23,413 30,042 C 20,905 (0.6% ) (1.4% ) (0.5% ) (12.0% ) (43.7% ) ↑ ↑ ↑ ↑ ↑ 20,698 21,383 22,349 24,087 29,680 H 20,614 (0.4% ) (3.7% ) (8.4% ) (16.8% ) (44.0% ) ↑ ↑ ↑ ↑ ↑ 22,198 22,440 22,111 24,518 33,556 O 22,080 (0.5% ) (1.6% ) (0.1% ) (11.0% ) (52.0% ) ↑ ↑ ↑ ↑ ↑ 22,621 22,925 23,106 26,526 36,053 S 22,476 (0.6% ) (2.0% ) (2.8% ) (18.0% ) (60.4% ) ↑ ↑ ↑ ↑ ↑ Water 2019, 11, 1825 17 of 21

Table A2. Travel distance from residents in St. Lawrence Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 23,412 24,070 23,470 24,342 35,188 C 23,117 (1.3% ) (4.1% ) (1.5% ) (5.3% ) (52.2% ) ↑ ↑ ↑ ↑ ↑ 23,131 24,264 24,795 25,376 31,438 H 22,800 (1.5% ) (6.4% ) (8.8% ) (11.3% ) (37.9% ) ↑ ↑ ↑ ↑ ↑ 24,713 25,292 24,573 25,388 35,560 O 24,449 (1.1% ) (3.5% ) (0.5% ) (3.8% ) (45.4% ) ↑ ↑ ↑ ↑ ↑ 24,802 25,552 25,339 27,194 37,637 S 24,548 (1.0% ) (4.1% ) (3.2% ) (10.8% ) (53.3% ) ↑ ↑ ↑ ↑ ↑

Table A3. Travel distance from residents in Our Lady Fatima Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 27,471 28,578 28,313 30,573 32,473 C 27,062 (1.5% ) (5.6% ) (4.6% ) (13.0% ) (20.0% ) ↑ ↑ ↑ ↑ ↑ 26,698 28,366 28,332 29,919 31,185 H 26,496 (0.8% ) (7.1% ) (6.9% ) (12.9% ) (17.7% ) ↑ ↑ ↑ ↑ ↑ 27,899 29,114 28,886 31,519 32,894 O 27,845 (0.2% ) (4.6% ) (3.7% ) (13.2% ) (18.1% ) ↑ ↑ ↑ ↑ ↑ 30,110 31,402 31,320 34,728 36,556 S 29,951 (0.5% ) (4.8% ) (4.6% ) (15.9% ) (22.1% ) ↑ ↑ ↑ ↑ ↑

Table A4. Travel distance from residents in St. Anthony Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 22,843 22,998 22,844 24,337 30,286 C 22,317 (2.4% ) (3.1% ) (2.4% ) (9.0% ) (35.7% ) ↑ ↑ ↑ ↑ ↑ 21,613 22,556 23,352 24,232 27,812 H 21,288 (1.5% ) (6.0% ) (9.7% ) (13.8% ) (30.6% ) ↑ ↑ ↑ ↑ ↑ 23,723 23,963 23,744 25,420 30,554 O 23,575 (0.6% ) (1.6% ) (0.7% ) (7.8% ) (29.6% ) ↑ ↑ ↑ ↑ ↑ 25,109 25,435 25,692 27,887 33,864 S 24,674 (1.8% ) (3.1% ) (4.1% ) (13.0% ) (37.2% ) ↑ ↑ ↑ ↑ ↑

Table A5. Travel distance for residents in Taipa Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 34,365 34,735 34,492 37,439 53,779 C 34,186 (0.5% ) (1.6% ) (0.9% ) (9.5% ) (57.3% ) ↑ ↑ ↑ ↑ ↑ 36,607 37,290 38,264 40,960 48,483 H 36,501 (0.3% ) (2.2% ) (4.8% ) (12.2% ) (32.8% ) ↑ ↑ ↑ ↑ ↑ 35,683 36,041 35,633 38,410 54,856 O 35,569 (0.3% ) (1.3% ) (0.2% ) (8.0% ) (54.2% ) ↑ ↑ ↑ ↑ ↑ 35,079 35,327 36,032 39,672 57,400 S 34,892 (0.5% ) (1.2% ) (3.3% ) (13.7% ) (64.5% ) ↑ ↑ ↑ ↑ ↑ Water 2019, 11, 1825 18 of 21

Table A6. Travel distance for residents in St. Francisco to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 63,945 64,392 66,085 67,988 81,662 C 63,861 (0.1% ) (0.8% ) (3.5% ) (6.5% ) (27.9% ) ↑ ↑ ↑ ↑ ↑ 64,980 65,781 66,717 70,177 70,617 H 64,916 (0.1% ) (1.3% ) (2.8% ) (8.1% ) (8.8% ) ↑ ↑ ↑ ↑ ↑ 63,572 63,878 63,417 67,055 80,951 O 63,396 (0.3% ) (0.8% ) (0.1% ) (5.8% ) (27.7% ) ↑ ↑ ↑ ↑ ↑ 61,366 61,642 61,804 66,370 81,890 S 61,282 (0.1% ) (0.6% ) (0.9% ) (8.3% ) (33.6% ) ↑ ↑ ↑ ↑ ↑

Table A7. Travel distance for residents in St. Lazarus Parish to destinations.

F B SSW-1 SSW-2 SSW-3 SSW-4 SSW-5 21,852 21,910 21,929 23,787 27,849 C 21,777 (0.3% ) (0.6% ) (0.7% ) (9.2% ) (27.95% ) ↑ ↑ ↑ ↑ ↑ 21,127 21,915 22,553 23,828 25,854 H 21,127 (0.0%) (3.7% ) (6.7% ) (12.8% ) (22.4% ) ↑ ↑ ↑ ↑ 22,465 22,676 22,573 24,674 28,051 O 22,458 (0.0%) (0.1% ) (0.5% ) (9.9% ) (24.9% ) ↑ ↑ ↑ ↑ 23,807 24,102 24,233 27,405 31,262 S 23,705 (0.4% ) (1.7% ) (2.2% ) (15.6% ) (31.9% ) ↑ ↑ ↑ ↑ ↑

References

1. Trenberth, K.E. Changes in precipitation with climate change. Clim. Res. 2011, 47, 123–138. [CrossRef] 2. Grimm, N.B.; Faeth, S.H.; Golubiewski, N.E. Global change and the ecology of cities. Science 2008, 319, 756–760. [CrossRef][PubMed] 3. The OFDA/CRED—International Disaster Database. Available online: www.emdat.be (accessed on 11 November 2018). 4. Hanson, S.; Nicholls, R.; Ranger, N. A global ranking of port cities with high exposure to climate extremes. Clim. Chang. 2011, 104, 89–111. [CrossRef] 5. Kuenzer, C.; Renaud, F.G. Climate and environmental change in river deltas globally: Expected impacts, resilience, and adaptation. In The Mekong Delta System; Springer: Dordrecht, The Netherlands, 2012; pp. 7–46. 6. Yang, L.; Scheﬀran, J.; Qin, H.; You, Q. Climate-related ﬂood risks and urban responses in the Pearl River Delta, China. Reg. Environ. Chang. 2015, 15, 379–391. [CrossRef] 7. Hou, Q.; Li, S.M. Transport infrastructure development and changing spatial accessibility in the Greater Pearl River Delta, China, 1990–2020. J. Transp. Geogr. 2011, 19, 1350–1360. [CrossRef] 8. World Countries by Population Density 2019. Available online: http://worldpopulationreview.com/countries/ countries-by-density/ (accessed on 11 August 2019). 9. Lo, S. Eight Routes to Better Typhoon Management in Macau, as Pakhar Response Reveals Crisis Lessons Learnt from Hato. Available online: https://www.scmp.com/comment/insight-opinion/article/2108608/eight- routes-better-typhoon-management-macau-pakhar-response (accessed on 16 August 2019). 10. Hincks, J. A Battered Macau Struggles to Come to Terms with Deadly Typhoon Hato. Available online: https://time.com/4917981/typhoon-hato-macau-aftermath/ (accessed on 16 August 2019). 11. Governo da Regiao Administrative Especial de Macau. Available online: https://www.dsec. gov.mo/getAttachment/e20c6bab-ada4-4f83-9349-e72605674a42/E_ICEN_PUB_2016_Y.aspx (accessed on 11 November 2018). 12. Wang, P. Hong Kong Picks Itself up after Typhoon Mangkhut. Chaos in Public Transport with 600 Blocked Roads. Available online: http://www.asianews.it/news-en/Hong-Kong-picks-itself-up-after-typhoon- Mangkhut.-Chaos-in-public-transport-with-600-blocked-roads-44950.html (accessed on 16 August 2019). Water 2019, 11, 1825 19 of 21

13. Kim, K.; Pant, P.; Yamashita, E. Integrating travel demand modeling and flood hazard risk analysis for evacuation and sheltering. Int. J. Disaster Risk Reduct. 2018, 31, 1177–1186. [CrossRef] 14. Debionne, S.; Ruin, I.; Shabou, S.; Lutoff, C.; Creutin, J.D. Assessment of commuters’ daily exposure to flash flooding over the roads of the Gard region, France. J. Hydrol. 2016, 541, 636–648. [CrossRef] 15. Guilloux, A. Cross-Boundary Disaster Risk Governance: Lessons from the Pearl River Delta. In Crossing Borders; Springer: Singapore, 2018; pp. 133–151. [CrossRef] 16. Price, R.K.; Vojinovic, Z. Urban flood disaster management. Urban Water J. 2008, 5, 259–276. [CrossRef] 17. Acosta-Coll, M.; Ballester-Merelo, F.; Martinez-Peiro, M.; Hoz-Franco, E. Real-time early warning system design for pluvial flash floods—A review. Sensors 2018, 18, 2255. [CrossRef] 18. Yang, B.; Li, M.; Li, S. Design-with-nature for multifunctional landscapes: Environmental benefit and social barriers in community development. Int. J. Environ. Res. Public Health 2013, 10, 5433–5458. [CrossRef] 19. Kurte, K.R.; Durbha, S.S. High resolution disaster data clustering using Graphics Processing Units. In Proceedings of the 2013 IEEE International Geoscience and Remote Sensing Symposium—IGARSS, Melbourne, Australia, 21–26 July 2013; IEEE (The Institute of Electrical and Electronics Engineers): Piscataway, NJ, USA; pp. 1696–1699. [CrossRef] 20. Mark, O.; Weesakul, S.; Apirumanekul, C.; Aroonnet, S.B.; Djordjević,S. Potential and limitations of 1D modelling of urban flooding. J. Hydrol. 2004, 299, 284–299. [CrossRef] 21. Mathews, A.J.; Ellis, E.A. An evaluation of tornado siren coverage in Stillwater, Oklahoma: Optimal GIS methods for a spatially explicit interpretation. Appl. Geogr. 2016, 68, 28–36. [CrossRef] 22. Hasnat, M.M.; Islam, M.R.; Hadiuzzaman, M. Application of GIS for Disaster Response in Dense Urban Areas: A Case Study for Dhaka City. J. Public Transp. 2018, 8, 1–19. Available online: https://trid.trb.org/view/1496884 (accessed on 15 November 2018). 23. Usman, R.A.; Olorunfemi, F.B.; Awotayo, G.P.; Tunde, A.M.; Usman, B.A. Disaster Risk Management and Social Impact Assessment: Understanding Preparedness, Response and Recovery in Community Projects. Available online: https://www.intechopen.com/books/environmental-change-and-sustainability/disaster- risk-management-and-social-impact-assessment-understanding-preparedness-response-and-recove (accessed on 15 August 2019). 24. Pregnolato, M.; Ford, A.; Wilkinson, S.M.; Dawson, R.J. The impact of flooding on road transport: A depth-disruption function. Transp. Environ. 2017, 55, 67–81. [CrossRef] 25. Hammond, M.J.; Chen, A.S.; Djordjević, S.; Butler, D.; Mark, O. Urban flood impact assessment: A state-of-the-art review. Urban. Water J. 2015, 12, 14–29. [CrossRef] 26. US EPA. Urban Runoff: Low Impact Development. Available online: https://www.epa.gov/nps/urban-runoff- low-impact-development (accessed on 19 August 2019). 27. Jia, H.F.; Yu, S.L.; Qin, H.P. Low Impact Development and Sponge City Construction for Urban Stormwater Management. Front. Environ. Sci. Eng. 2017, 11, 20. [CrossRef] 28. Zhou, L.; Shen, G.Q.; Woodfin, T.; Chen, T.; Song, K. Ecological and economic impacts of green roofs and permeable pavements at the city level: The case of Corvallis, Oregon. J. Environ. Plan. Manag. 2017, 61, 430–450. [CrossRef] 29. Seo, M.; Jaber, F.; Srinivasan, R.; Jeong, J. Evaluating the Impact of Low Impact Development (LID) Practices on Water Quantity and Quality under Different Development Designs Using SWAT. Water 2017, 9, 193. [CrossRef] 30. Ward, P.J.; Marfai, M.A.; Yulianto, F.; Hizbaron, D.R.; Aerts, J.C.J.H. Coastal inundation and damage exposure estimation: A case study for Jakarta. Nat. Hazards 2011, 56, 899–916. [CrossRef] 31. OD Cost Matrix Analysis. Available online: https://desktop.arcgis.com/en/arcmap/latest/extensions/network- analyst/od-cost-matrix.htm (accessed on 30 December 2018). 32. Hydrologic Analysis Sample Applications. Available online: https://pro.arcgis.com/en/pro-app/tool- reference/spatial-analyst/hydrologic-analysis-sample-applications.htm (accessed on 30 December 2018). 33. Macao Meteorological and Geophysical Bureau. Available online: http://www.smg.gov.mo/smg/c_index.htm (accessed on 11 November 2018). 34. Wan, Y.K.P. A comparison of the governance of tourism planning in the two Special Administrative Regions (SARs) of China–Hong Kong and Macao. Tour. Manag. 2013, 36, 164–177. [CrossRef] 35. Governo da Regiao Administrative Especial de Macau. Available online: https://www.dscc.gov.mo/ENG/ knowledge/geo_statistic.html (accessed on 11 November 2018). Water 2019, 11, 1825 20 of 21

36. Li, L.L.; Yang, J.; Lin, C.Y.; Chua, C.T.; Wang, Y.; Zhao, K.F.; Wu, Y.T.; Liu, L.F.; Switzer, A.D.; Mok, K.M.; et al. Field survey of Typhoon Hato (2017) and a comparison withstorm surge modeling in Macau. Nat. Hazards Earth Syst. Sci. 2018, 3167–3178. [CrossRef] 37. Marfai, M.A.; King, L. Potential vulnerability implications of coastal inundation due to sea level rise for the coastal zone of Semarang city, Indonesia. Environ. Geol. 2008, 54, 1235–1245. [CrossRef] 38. Macau SAR Yearbook of Statistics. Available online: https://yearbook.gcs.gov.mo/zh-hant/ (accessed on 11 November 2018). 39. Jenelius, E.; Petersen, T.; Mattsson, L.G. Importance and exposure in road network vulnerability analysis. Transp. Res. Part A Policy Pract. 2006, 40, 537–560. [CrossRef] 40. Cova, T.J.; Johnson, J.P. A network flow model for lane-based evacuation routing. Transp. Res. Part A Policy Pract. 2003, 37, 579–604. [CrossRef] 41. Bhaduri, B.; Minner, M.; Tatalovich, S.; Harbor, J. Long-term hydrologic impact of urbanization: A tale of two models. J. Water. Res. Plan. Manag. 2001, 127, 13–19. [CrossRef] 42. Feng, C.; Wang, H.; Rao, X. The morphological evolution of Macau. In Proceedings of the Eighth International Space Syntax Symposium, 2012. Available online: https://www.academia.edu/3535278/The_Morphological_ Evolution_of_Macau (accessed on 21 November 2018). 43. Xi, X.; Min, L.I. Research on the Increment of Micro-Green Space in High-Density City of Macao Peninsula. Available online: http://en.cnki.com.cn/Article_en/CJFDTotal-CXGH201505016.htm (accessed on 6 December 2018). 44. Qie, Z.J.; Rong, L.L. An integrated relative risk assessment model for urban disaster loss in view of disaster system theory. Nat. Hazards 2017, 88, 165–190. [CrossRef] 45. Porter, J. The transformation of Macau. Pac. Aff. 1993, 66, 7–20. [CrossRef] 46. Tingsanchali, T. Urban flood disaster management. Procedia Eng. 2012, 32, 25–37. [CrossRef] 47. Tenerelli, P.; Gallego, J.F.; Ehrlich, D. Population density modelling in support of disaster risk assessment. Int. J. Disaster Risk Reduct. 2015, 13, 334–341. [CrossRef] 48. Dietz, M.E. Low impact development practices: A review of current research and recommendations for future directions. Water Air Soil Pollut. 2007, 186, 351–363. [CrossRef] 49. Ahiablame, L.M.; Engel, B.A.; Chaubey, I. Effectiveness of low impact development practices: Literature review and suggestions for future research. Water Air Soil Pollut. 2012, 223, 4253–4273. [CrossRef] 50. Qin, H.; Li, Z.; Fu, G. The effects of low impact development on urban flooding under different rainfall characteristics. J. Environ. Manag. 2013, 129, 577–585. [CrossRef][PubMed] 51. Foster, J.; Lowe, A.; Winkelman, S. The Value of Green Infrastructure for Urban Climate Adaptation. Available online: https://ccap.org/resource/the-value-of-green-infrastructure-for-urban-climate-adaptation/ (accessed on 5 December 2018). 52. Ellis, J.B. Managing urban runoff. In Handbook of Catchment Management; Wiley-Blackwell: Oxford, UK, 2009. Available online: https://onlinelibrary.wiley.com/doi/book/10.1002/9781444307672 (accessed on 3 December 2018). 53. Stephenson, D.; Barta, B. Impacts of Stormwater and Groundwater Ingress on Municipal Sanitation Services. Available online: http://www.wrc.org.za/Knowledge%20Hub%20Documents/Research%20Reports/1386-1- 05.pdf (accessed on 5 December 2018). 54. Rossman, L.A. Storm Water Management Model User’s Manual, Version 5.0. Available online: https://cfpub. epa.gov/si/si_public_record_report.cfm?Lab=NRMRL&dirEntryId=114231 (accessed on 10 December 2018). 55. Stotz, J. The Average and Minimum Salary in Macau will be Analyzed in this Post, Featuring also an Economic Outlook of This Chinese Special Administrative Region. Available online: https://checkinprice. com/average-and-minimum-salary-in-macau-china/ (accessed on 17 August 2019). 56. Valinski, N.A.; Chandler, D.G. Infiltration performance of engineered surfaces commonly used for distributed stormwater management. J. Environ. Manag. 2015, 160, 297–305. [CrossRef][PubMed] 57. Macau Weather. Available online: https://www.travelchinaguide.com/climate/macau.htm (accessed on 17 August 2019). 58. Luzar, J.A. A sustainable Design Solution for Stormwater: The Great Northern Mall in North. Available online: http://cardinalscholar.bsu.edu/handle/123456789/193764 (accessed on 5 December 2018). 59. Olmsted. Available online: https://pdfs.semanticscholar.org/6ecf/7fcf81eaaf35644fba46e32d7aa6c2e7fa4d.pdf (accessed on 10 December 2018). Water 2019, 11, 1825 21 of 21

60. Introduction to Stormwater Treatment Practices. Available online: https://www.caciwc.org/library/ presentations/IW%20Training%20Sessions.pdf (accessed on 15 December 2018). 61. Rauch, I. Parking Lot Typical Stormwater Runoff Constituents and Stormwater. Ph.D. Thesis, Stellenbosch University, Stellenbosch, South Africa, 2019. Available online: http://scholar.sun.ac.za/handle/10019.1/105938 (accessed on 11 December 2018). 62. Getter, K.L.; Rowe, D.B.; Andresen, J.A. Quantifying the effect of slope on extensive green roof stormwater retention. J. Ecol. Eng. 2007, 31, 225–231. [CrossRef] 63. Coffman, L.S. Method and Apparatus for Treating Stormwater Runoff. Available online: https://patents. google.com/patent/US6277274B1/en (accessed on 15 December 2018). 64. Pina, R.; Ochoa-Rodriguez, S.; Simões, N.; Mijic, A.; Marques, A.; Maksimović, C.ˇ Semi-vs. fully-distributed urban stormwater models: Model set up and comparison with two real case studies. Water 2016, 8, 58. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).