Multi-scale Forecast And Projection Of Urban Waterlogging Risk Based On Coupled Hydro-meteorological Modelling

The dual effects of rapid urbanization and climate warming lead to the increasingly serious urban waterlogging.As an important non-engineering measure for disaster prevention and mitigation,waterlogging prediction has become a research hotspot.However,the current focus of waterlogging prediction is mostly limited to the waterlogging intensity itself,while ignoring the important factors affecting the occurrence,intensity and spatial distribution of waterlogging disaster,namely,the exposure and vulnerability of hazard-affected boby.Therefore,in order to predict the risk caused by urban waterlogging more precisely and quantitatively,it is necessary to further carry out the refined prediction of the exposure and vulnerability of hazard-affected boby on the basis of waterlogging intensity prediction.Thus,it is urgent to conduct the research on urban waterlogging risk prediction so as to realize the further extension from the disaster weather forecast to the disaster risk prediction.This study takes Hefei,the capital of Anhui Province,as the research area.multi-source data was used,including meteorological data(meteorological observation data in recent 39 years,climate model projection,quantitative precipitation forecast),hydrological data(drainage pipe network,drainage facilities,dispatching information of drainage,historical waterlogging observation),geographic data(1:10000 geographic information data,land use,soil type),social and economic data(GDP,population density,price and people's life data),urban management(transportation facilities,medical facilities,building information)and satellite remote sensing data.Firstly,the characteristics and causes of urban waterlogging were analyzed.Secondly we constructed an urban waterlogging numerical model to simulate waterlogging scenario.Based on the simulation,framework of subdistrict waterlogging risk assessment and community residential waterlogging risk assessment is established by using index system method and loss function respectively.Finally,by coupling the numerical forecast and the urban waterlogging numerical model,the waterlogging scenarios in the future period are predicted.Combined with the risk assessment framework,the waterlogging risk projection and forecast of the subdistrict scale and the community scale is realized respectively.Around the above objectives and contents,the following research results have been achieved:(1)Statistics of waterlogging characteristics and waterlogging cause analysis.Most of the heavy rainfall that causes waterlogging occurred in summer,the largest number occured in July(34.8%),and mostly occurred at 14-16 PM(51.9%).There are mainly three types of rain patterns,which are steep single peak type,gentle single peak type and multi peak type.Most of them show the characteristics of steep single peak type.The short-term precipitation intensity exceeds 30mm/h and the duration does not exceed 2 hours.The waterlogging generally occurs within one hour after the short-time heavy rainfall.There are many reasons for the aggravation of waterlogging disasters,mainly including: short-term heavy rainfall has tended to be more extreme,frequent and concentrated in recent years;“Heat Island Effect” and “Rain Island Effect” are becoming increasingly significant;the increase of runoff depth due to the change of underlying surface;the low standard of urban drainage;imperfect drainage network;influence of river water level;the explosive growth of urban population and GDP.(2)The construction and modification of urban waterlogging numerical model.Based on GIS technology and hydrodynamic equation,urban surface hydrological process and drainage system hydrological process is simulated.The evaluation shows that the urban waterlogging prone area,concave interchange,underground passage,urban surface roughness and drainage pipe diameter coefficient have great influence on the simulation accuracy.Through parameter calibration and correction,the simulation accuracy can be significantly improved.The correlation coefficient of observation data is 0.89,the average grade error is 0.38,and the rate of missing report is 0.The fuzzy score of grade I to IV waterlogging is greater than 85.The simulated water evolution trend is consistent with the actual situation.(3)The establishment of multi-scale waterlogging risk assessment framework based on scenario simulation.The waterlogging numerical model driven by 20-year and 100-year return period short-time heavy rainfall was used to simulate the inundation scenarios on an hourly scale under different return periods.Using remote sensing and GIS tools,on the one hand,taking 37 subdistricts in Hefei City as the basic assessment unit,10 risk assessment indicators representing disaster causing factors,exposure and vulnerability were selected.The comprehensive risk value of subjective and objective fusion was calculated by combining fuzzy analytic hierarchy process(FAHP)and principal component analysis(PCA),and the waterlogging risk map under different return periods was compiled.On the other hand,on a smaller scale,taking more than 1000 residential buildings in the community as the basic unit of assessment,the vulnerability curves between the indoor property loss and the depth of waterlogging of residents in four income classes are established by using the synthetic method,and the property loss assessment is carried out through the exposure analysis of the housing inundation status.The results show that: under 20-year and 100-year rainfall,(1)the difference of waterlogging risk among subdistricts is very significant.Generally,the waterlogging risk in the urban-rural fringe is the highest,followed by the more developed old town,the new urban area and the less developed old town,and the waterlogging risk of the old downtown is the lowest.(2)When the high-risk area encounters stronger rainfall,the waterlogging disaster will be magnified.(3)From high to low,the order of waterlogging risk is: new community > shanty simple house > old community >high-grade housing.When the rainfall intensity increases,the risk of residential waterlogging in shantytowns increases significantly.(4)When the depth of inundation increased from 0 to 25 cm,the property loss per unit area increased rapidly.When the depth of inundation exceeded 50 cm,the increase slowed down.The increase of property loss of high-income class was always greater than that of low-income class,and the gap of property loss among different classes was increasing with inundation depth.(4)Multi-scale waterlogging risk forecast and projection based on hydro-meteorological coupling.At the subdistrict scale,coupled the climate model projection in CMIP6(Coupled Model Intercomparison Project Phase 6)with waterlogging numerical model,combined with the subdistirct risk assessment framework,the subdistirct waterlogging risk under three scenarios of Shared Socioeconomic Pathways(SSP)under extreme heavy rainfall is obtained.The results show that:(1)Due to the distribution characteristics of extreme rainfall in the South and less in the north,the waterlogging risk in the southern area is generally higher than that in the northern part of the city.(2)In general,the risk of waterlogging in the urban-rural fringe > the more developed old town > new town > the less developed old town > the old downtown.(3)The spatial distribution characteristics of waterlogging risk under three SSP scenarios are very similar.(4)The risk of waterlogging under SSP1-2.6 scenario(low emission)> SSP 2-4.5(medium emission)> SSP 5-8.5(high emission).(5)In the area with high waterlogging risk,the risk values under three SSP scenarios have obvious differences;While in the areas with low risk,the risk values under different SSP scenarios have little difference.At the community scale,the quantitative precipitation forecast which are updated every 10 minutes was used to drive waterlogging numerical model,combined the results with the risk assessment framework of community housing to obtain the dynamic prediction of community waterlogging risk in the next 6 hours.The test results show that the prediction of inundation depth can indicate the occurrence and development of waterlogging 2 hours in advance.When the forecast lead time is 1hour,the inundation area,depth and occuring time can be accurately predicted.Therefore,the dynamic residential risk forecast product can accurately estimate the number and distribution of residential property loss caused by waterlogging 1 hours in advance.In summary,this study goes deep into the interdisciplinary fields of meteorology,hydrology and risk management,based on numerical weather prediction,urban waterlogging numerical simulation technology and disaster risk assessment technology,the process framework of waterlogging disaster risk prediction was constructed: numerical model rainfall forecast ? waterlogging numerical simulation? waterlogging scenario prediction ? risk assessment based on waterlogging scenario ? waterlogging risk forecast and projection.The risk forecast and projection of waterlogging disaster are based on the prediction of hazard factors,exposure and vulnerability of hazard-effected body.In the context of future climate change,the subdistrict scale waterlogging risk projection is based on the newly proposed future emission scenario-Shared Socioeconomic Pathways(SSP),which has some improvement compared with the Representative Concentration Pathway scenario(RCP)used in the Fifth International coupled model comparison program(CMIP5)and is more suitable for the study of climate impact,adaptation and vulnerability.In addition,the subjective and objective methods were used in the assessment process.Therefore,the results of waterlogging risk projection based on SSP scenario can provide more scientific support for government decision-makers to make urban planning.In the future 6 hours,the waterlogging risk forecast product of community housing can dynamically predict the economic loss.The time resolution is hour,and the forecast unit is residential building,which achieves the quantification,refinement and dynamic of waterlogging risk forecast.