| Transnational accessibility analysis is the pivotal work to promote the regional communication,economic development,well-being of the people,and trade contacts.Amongst,airline network is the basis of coarse-grained and macroscopic analysis between airports or cities.The development of stereo-traffic network has made the intermodal transport network become the focus of accessibility research,which can provide more comprehensive and accurate evaluation results.Particularly,the air-land intermodal transport(ALIT)network is the core network for rapid movement of people and goods,and also promote the development of grid-level refined and full-coverage accessibility analysis.The overall accessibility analysis,summarized by grid-level evaluation results,can also be more reliable.The settlement area is the core region for human activities and major service object for accessibility analysis,and only occupies a small proportion of land cover,where extraction of settlement areas can get the effective coverage for accessibility analysis.The integration of remote sensing and websourced data can provide the strong data support for settlement area extraction and grid-level accessibility analysis.China has launched “The Belt and Road Initiative”(B&R),which aims to promote the regional development and realize common prosperity,and the South Asia and Southeast Asia(SASEA)region is the key area in B&R.Therefore,it is of great value and practical significance to systematically perceive the accessibility between China and SASEA for B&R implementation and connectivity promotion.Settlement area is defined as regions for human daily activities.Existing settlement area extraction rely on remote sensing(RS)data and land cover information,which cannot represent human activities region.The geolocation data contain human activity information,and could be applied for settlement area extraction.As the objective capability for accessibility evaluation index,grid-level travel time calculation rely on transport network analysis which need large amounts of basic geographical data to support.As the subjective intentions,travel population estimation is mainly based on the data analysis method,or spatial interaction models.While data analysis methods have low effect for large-scale research,and models for transnational travel population estimation are also rare.Meanwhile,how to integrate the grid-level indicators for multi-level comprehensive analysis is also a problem that worth considering.To solve these problems,the framework of “Settlement area extraction-Travel time calculation-Travel population estimation-Multi-scale analysis / suggestion” for accessibility systematic perception is proposed.The web-sourced geolocation data is introduced to represent the intensity of human activity,and conduct the integration with multi-source RS data to enhance the extraction accuracy.The airlines,road network information and related properties are crawled from Internet,and the ALIT network is constructed to connect settlement areas.More specifically,the modified model is proposed to estimate the travel population by ALIT network between China and SASEA.Multi-angle suggestions for accessibility promotion is also developed based on airport nodes.The main contents of this dissertation are as follows:(1)Settlement area grid extraction from integration of remote sensing and web-sourced geolocation data.Features,which can represent human activity intensity,are evacuated from web-sourced Tencent location request density(LRD),point of interest(POI)data,road network,and route planning data by similarity analysis,grid statistics,and other techniques.Then geolocation features,combined with RS features,form the 15 low-redundancy input features.The GEE platform and multi-source data are applied to label the high quality training and validation samples,RF classifier is constructed for China and SASEA respectively to deal with these multi-source RS-geolocation features.The optimal parameter setting for RF classifier are also conducted respectively in China and SASEA,including the decision tree scale k and maximum feature number m.The extraction results show that there are about 192 thousand grids with 1 kilometer resolution in China,and about 189 thousand in SASEA,with an overall accuracy of 90.79% and 88.07% respectively.Analysis on the feature relative importance and absence accuracy show that geolocation features exert significant influence on the promotion of classifier’s accuracy.Amongst,VIIRS nighttime,Tencent LRD,POI density,and road network density features show large contribution and small absence overall accuracy in settlement area extraction.Compared with other LC products and reference boundaries,our extraction results by RS and geolocation joint features show a higher accuracy,and also perform better than single RS or single geolocation features.(2)Travel time and travel population between settlement area grids in ALIT network.A web crawler is applied to gather 2017,2018 airline information between 65 China airports and63 SASEA airports from web ticket booking platform.Compared with 2017 summarized data,China arrival/departure airline and flight number increase slightly in 2018,both for direct and transit airlines.The web mapping data is used to get the travel time from each grid to the nearest airport in road network.The overall land travel time is 2.39 hours in China,and show better accessibility than SASEA,where 4-hour accessibility range cover about 89.43% population.Then the ALIT network is constructed to connect airports,settlement areas,road network and airline network.The grid-level minimum travel time between settlement areas is calculated as the travel time index and the threshold of temporal range in radiation model.The gravity value in gravity model is also introduced as the second threshold for temporal range.Then “GravityRadiation” model is constructed to adapt to the ALIT network.Grid population estimation result is input into the model to calculate the grid-level travel population.The summarized results for districts and countries(provinces)show that China has larger travel population with SEA than SA.The outbound cities with largest travel population are Beijing,Shanghai,Chengdu,Guangzhou,and Shenzhen,and most-visited countries are Thailand,Vietnam,and Indonesia.The simulation results by “Gravity-Radiation” model show a stronger consistency with real statistics data,and a higher accuracy than gravity model and radiation model.(3)Multi-scale ALIT accessibility analysis and promotion suggestions between China and SASEA regions.With the combination of travel time and travel population between settlement areas,the grid level,district level and country(province)level comprehensive travel time are calculated to represent the overall accessibility,and their spatial patterns are also analyzed.Hongkong,Beijing,and Shanghai have the largest provincial accessibility,with comprehensive travel time of 16.85,20.13,and 21.10 hours.Thailand,Singapore,and Malaysia have the largest national accessibility,with comprehensive travel time of 16.72,17.56,and 18.39 hours.Combining travel time and flight number,the radiating capacity analysis of ALIT network show that the capacity promotion in 2018 can be attributed to the expansion of transit airlines.Then the conjoint analysis of airport-level airline travel time,land travel time,and travel population is conducted to find the airport with poor accessibility.Finally,multi-angle suggestions for airport nodes are concluded to improve the overall accessibility,including expansion of direct airlines,reduction of transit times,increment of road network construction in specific regions,and reasonable flight schedule considering travel population.In addition,the national land accessibility index is calculated based on the above-mentioned framework,and higher correlation is found between accessibility index and national socio-economic index,which expand the accessibility application.In summary,the innovation of this research rely on three perspectives.First,we propose the settlement area extraction method based on the integration of RS and web-sourced geolocation data.Second,we construct the ALIT network from diverse web-sourced data and“Gravity-Radiation” model that adapt to ALIT network.Finally,we propose the framework of“Settlement area extraction-Travel time calculation-Travel population estimation-Multiscale analysis / suggestion” for accessibility systematic perception. |
PDF Full Text Request