Research On Megalopolitan Changes And Corresponding Precipitation Response

As a result of high-density urbanization and climate change,both the frequency and intensity of extreme urban rainfall are increasing.Drainage systems are not designed to cope with this increase,and as a result,floods are becoming more common in cities,particularly in the rapidly growing cities of China.In recent years,policymakers and city managers have placed hopes on a large-scale integrated blue-green-gray rainwater system to improve the city’s climate adaptability,i.e.,the initiative and implementation of Sponge City Construction(SCC).This research aims to evaluate the statistical characteristics of precipitation in a megalopolis with the dual drivers of climate change and land use change.And there are a few topics explored: the simulation of urban growth in a megalopolis level under the context of nonintervention and/or implementation of volume control,the precipitation downscaling based on a deep learning network,and the gridded precipitation statistical analysis in a changing region.A series of novel and practical methods were developed,and the Yangtze Delta Megalopolis(YDM)was employed as a case study.The main conclusions are as follows:(1)A set of megapolitan land use change simulation methods based on cellular automation including the dynamization of the simulation subregions,the adaptive update of transition rates,and the combination of multi-time steps was developed.This method has the following advantages: the domain is hierarchized into three levels and the subregions of last level are changing with simulation,which eliminates the defects of stationary administrative boundaries;the transfer matrix and the weight matrix are dynamically and adaptively acquired,which reduces human intervention in the simulation process.An adaptive dynamic growth model in the YDM was constructed.The model was tested using historical land use data from 2005 to2015.Compared the simulated land use map with the original,the average fuzzy similarity reaches more than 0.99 when the size of moving window increases is 17 grids(its resolution of5.1 km)and the average similarity for the comparison of change part is 0.65.(2)Based on the indicator of volume control and the design of the four assumptions for SCC modeling(assumption of the occurrence only on one type of land use,assumption of the uniform spatial distribution,assumption of the uniform value of controlling indicator,and assumption of equivalent of existence of land use equivalent to sponge facility performance),a set of simulation method coupled volume control and land change was proposed.Those methods establish the relationship between green rainwater infrastructures and land uses,further,explore the feedback on land use and land cover(LULC)caused by undergoing strategry of volume control.A coupled model in the YDM was constructed and the land uses with a resolution of 5 km under the volume control of 70%,75%,and 80% from 2021 to 2050 was predicted.(3)A deep residual network for precipitation downscaling(DRN-PD)was proposed.The inputs of the network are a variety of environmental variables with inconsistent resolution,and the output is precipitation with highly uneven temporal and spatial distribution,and it would be suitable for downscaling of data from general circulation models(GCMs)with large biases.Compared with the state-of-the-art neural network models for precipitation downscaling,the trained DRN-PD in the YDM(YDM-DRN-PD)has better adaptability and accuracy in the evaluations.(4)A gridded precipitation statistical analysis method that is suitable for sampling region that changes with time,meeting the requirements of a large sample size,and more adaptable to statistics was designed.The samples generated according to this method have the advantages of huge sample size,uniform distribution,and the same grid size.The extracted statistics are based on percentiling,then they have probabilistic significance and have a wider application range.(5)Integrating the simulation of megapolitan land use change,forecasting of SCC under planning,precipitation downscaling,and statistical analysis of rainfall,a framework of assessment of megalopolitan precipitation responses under the conditions of climate change and rapid urbanization was proposed.This framework was used to systematically assess the characteristics of precipitation in the YDM under the dual conditions of climate change and urban growth.The results show that:(i)Most daily precipitation greater than 100 mm are located in the southern part of YDM.(ii)Most daily precipitation greater than 50 mm are located in the southweatern part,while the daily precipitation greater than 50 mm,the annual maximum daily precipitation,and annual total precipitation increase all over the YDM.(iii)Compared with the business-as-usual scenario of ssp585,the increase of daily precipitation greater than 50 mm and annual maximum daily precipitation are slowed down under the ssp245.(iv)the strategies of volume control have little effect on the precipitation rules,and won’t bring significantly heavy storms.The proposed methods and constructed models in this study can provide methodological references for urban planning,SCC,and related scientific research;the prepared datasets and conclusions of the analysis in the YDM can provide data support and knowledge accumulation for the sustainable development of the YDM under the conditions of climate change.