Soundscape Evaluation and ANN Modelling in Urban Open Spaces - Lei Yu _ A thesis submitted to the University of Sheffield in partial fulfi)ment of the requirements for the degree of Doctor of Philosophy _ School of Architecture, University of Sheffield Summary There is an increasing public and academic interest in the environmental qualities of urban open spaces. The study in this thesis focuses on soundscape research in urban open spaces, which is within the paradigm of environmental psychology. It explores how to use the results of soundscape research in aiding the design process of urban open spaces with regard to the sonic environment. It is based on common notions that the acoustic aspect of urban open spaces should be considered in the same way as the visual dimensions. The determinant of a soundscape is subjective evaluations, which depend on two acoustic aspects; one is the sound noise scales and the other is the effects of various sound sources. Based on data collected from a series of field studies and laboratory experiments, the subjective evaluations of sound-level and sound preference have been separately studied using statistical analyses, and the overall evaluations of soundscape and acoustic comfort have been examined. In order to provide a feasible tool to aid soundscape deigns, the study develops a modelling tool, namely artificial neural network (ANN), to present the subjective evaluations of potential users at the design stage. Based on the ANN models, soundscape maps can be produced. The results of statistical analyses suggest that various factors influencing the subjective evaluations of sound level, sound preference and acoustic comfort are different in terms of a variation among case study sites and noticed sounds. Generally speaking, sound physical and psychological characteristics have the most influence on the subjective evaluations. The subjective evaluations of other physical environments are also much relevant to the soundscape evaluations, whereas social/demographical and behavioural factors are insignificant although some relationships have been found for certain factors. In addition to giving useful guidelines and information to soundscape research and design, the results are also crucial in selecting input variables for ANN prediction models. For ANN model predictions, it is found that a general model for all the case study sites is less feasible due to the complex physical and social environments. Practical models for certain type of urban open spaces are more reliable. The performance of acoustic comfort models is considerably better than that of sound level models. It is also found that the key variables to determine the prediction performance of sound preference models are sound meanings and the sounds' physical and psychological characteristics. Furthermore, the prediction maps based on ANN models' outputs have been successfully produced in presenting the potential users' appraisals of a soundscape in developing urban open spaces. Acknowledgements Firstly I wish to express my sincere appreciation to my supervisor, Professor J. Kang, for his valuably supervising and instantly encouraging me since my MSc study. I would then like to thank my colleagues, Dr. M. Zhang and Dr. W. Yang for their supports and useful discussions in the whole study. It has been a pleasure to work with them as well as other colleagues, Dr. Y. Meng, Dr. C. J. Yu, B. Wang and H. Xie, in the acoustics groups of Sheffield University. I also enjoyed the time having spent with the interviewees in the Shanghai surveys for their cooperation. My appreciation is also given to Dr. R. F. Harrison and Dr. K. Triantafyllopoulos for their professional support in the interdisciplinary parts of this study. I specially appreciate the helps from my friend S. Payne for the important discussions. I am greatly indebted to my friends, Dr. A. Orrell, L. Cresswell, S. Payne, P. Allum, S. Bayliss, N. Awan, and ELT staffs who proofread my thesis and publications. Last but not least, my deepest gratitude goes to my mother, my father and my sister for their understanding, encouragement and support. II Table of Contents Summary Acknowledgements ii Contents iii List of tables ix List of figures xiii Abbreviations xvi Chapter 1 -Introduction 1 1.1 Background 1.2 Aims and methodology 2 1.3 Thesis structure 4 Chapter 2 - Literature Review 6 2.1 Urban open spaces 6 2.1.1 Definition of urban open spaces 6 2.1.2 Development of urban open spaces 7 2.1.3 Roles of urban open in modern city 8 2.1.4 Constructing urban open spaces 9 2.2 Physical social environment and its study method 10 2.2.1 Physical and social environment 10 2.2.2 Physical and social environment in urban open spaces II 2.2.3 Visual, thermal, lighting and sound environment 11 2.2.4 What is environmental psychology? 12 2.2.5 Theories of environmental psychology 13 2.2.6 Environmental and noise stress 14 2.2.7 Statistics 14 2.3 Soundscape 15 2.3.1 What is soundscape? 15 2.3.2 Development of soundscape along with noise control 16 2.3.3 Approaches of soundscape study 17 2.3.4 Studies of noise annoyance 18 2.3.5 Sound characteristics 19 2.3.6 Soundscape in urban planning and design 20 2.4 Soundscape in urban open spaces 21 2.4.1 Sound preference 21 2.4.2 Studies of soundscape in urban open spaces 22 2.4.3 Aural and visual interactions 23 2.4.4 Subjective evaluations in the soundscape design 24 iii 2.4.5 Relationships amongst various aspects 25 2.5 Artificial neural networks (ANNs) 26 2.5.1 Introduction 26 2.5.2 Historical perspective 27 2.5.3 Networks 28 2.5.4 ANNs configuration 30 2.5.5 Discussions 31 2.6 Applications of ANNs in various areas 32 2.6.1 Applications in medical and environmental area 32 2.6.2 Applications in building science 33 2.6.3 Applications in building acoustics 33 2.6.4 Applications in design area 34 2.6.5 Applications in soundscape study 34 2.7 ANN software 35 2.7.1 Qnet and NeuroSolutions 35 2.7.2 BrainMaker & STATISTICA 36 2.8 Conclusions 37 Chapter 3 - Methodology 40 3.1 Field studies 40 3.1.1 Questionnaires for the EU field studies 41 3.1.2 Questionnaires for the Chinese field studies 42 3.1.3 Case study sites in the EU 44 3.1.4 Case study sites in Beijing 47 3.1.5 Case study sites in Shanghai 49 3.1.6 Interviewees and duration of field studies 52 3.1.7 Sound levels in field study sites 53 3.1.8 Sound sources in the case study sites· 55 3.2 Laboratory experiments 56 3.2.1 Experimental procedures 57 3.2.2 Sound samples recording 58 3.2.3 Studied sounds 59 3.2.4 Subjects 61 3.3 Statistical analyses 61 3.3.1 The variables and data issues 62 3.3.2 The sample distributions 65 3.3.3 Use of statistical techniques in this study 67 3.4 ANN model system and application software 69 3.4.1 A modelling framework 69 3.4.2 Qnet model 71 3.4.3 NeuroSolutions model 74 IV 3.5 Conclusions 76 Chapter 4 - Sound Level and Acoustic Comfort Evaluation 77 4.1 Relationships amongst social/demographical factors 77 4.1.1 Age, occupation and education 78 4.1.2 Gender, occupation, education and residential status 79 4.1.3 Occupation and education 80 4.1.4 Summary 81 4.2 Social/demo graphical factors 81 4.2.1 Age, occupation and education reo the sound level evaluations 81 4.2.2 Gender and residential status reo the sound level evaluations 83 4.2.3 Long-term sound experience reo the sound level evaluations 84 4.2.4 Social/demographical factors reo the acoustic comfort evaluations 87 4.3 Acoustic/physical factors 89 4.3.1 Acoustic/physical factors reo the sound level evaluations 89 4.3.2 Acoustic/physical factors reo the acoustic comfort evaluations 91 4.4 Behavioural factors 93 4.4.1 Earphones, reading/writing reo the sound level evaluations 93 4.4.2 Watching, movement statues reo the sound level evaluations 95 4.4.3 Frequency, reason and grouping reo the sound level evaluations 99 4.4.4 Behavioural factors reo the acoustic comfort evaluations 100 4.5 Psychological factors 101 4.5.1 Site preference and view assessment reo the sound level evaluations 101 4.5.2 Thermal conditions, brightness and overall physical comfort evaluations 103 reo the sound level evaluations 4.5.3 Site preference and view assessment reo the acoustic comfort evaluations 105 4.5.4 Thermal conditions, brightness and overall physical comfort evaluations 106 reo the acoustic comfort evaluations 4.5.5 The sound level evaluation reo the acoustic comfort evaluation 107 4.5.6 The sound preference evaluations reo the acoustic comfort evaluations 107 4.6 Social/demographical factors and physical, behavioural, psychological factors 108 4.7 Conclusions 109 Chapter 5 -Sound Preference Evaluations 112 5.1 Effect of types of sound sources 112 5.1.1 Effect of sound category 113 5.1.2 Effect of sound subcategory 114 5.1.3 Effect of sound meaning on subjective evaluations 116 5.1.4 Differences between the EU and China 117 5.1.5 Aural and visual interactions 118 5.2 Psychoacoustic parameters: loudness and sharpness 120 v 5.2.1 Loudness of the studied sounds 120 5.2.2 Loudness reo the subjective evaluations of the studied sound 121 5.2.3 Loudness reo the subjective evaluations of different sound levels 124 5.2.4 Sharpness of the studied sounds 125 5.2.5 Sharpness reo the subjective evaluations of the studied sound 126 5.3 Socialldemographical factors 128 5.3.1 Age 128 5.3.2 Education 131 5.3.3 Gender 132 5.3.4 Occupation 134 5.3.5 Residential status 136 5.4 Physical, behaviourallpsychological factors and home sound experience 138 5.4.1 Physical conditions 138 5.4.2 Behavioural factors 141 5.4.3 Site preferences 144 5.4.4 Home sound experience 145 5.5 Conclusions 146 Chapter 6 - ANN Models for the Sound Level/Acoustic Comfort 149 Evaluations 6.1 Test of the model performance 149 6.1.1 The model of sound level evaluations for the Makedonomahon 149 6.1.2 Results and discussions 150 6.2 Qnet models for the sound level evaluations 151 6.2.1 General models 151 6.2.2 Individual models 153 6.2.3 Group models ' ISS 6.2.3.1 Models for city centres 155 6.2.3.2 Models for residential areas 157 6.2.3.3 Models for tourist spots 158 6.2.3.4 A model for railway stations 159 6.2.3.5 Summary and discussions 160 6.3 Qnet models for the acoustic comfort evaluations 160 6.3.1 General models 161 6.3.2 Individual models 163 6.3.3 Group models 164 6.3.4 Summary and discussions 165 6.4 NeuroSolutions models for the sound level evaluations 165 6.4.1 A general model 166 6.4.2 An individual model 167 6.4.3 A model for tourist spots 168 vi 6.4.4 Summary and discussions 168 6.5 Ordinal logistic regression (OLR) models 169 6.5.1 OLR models for the sound level evaluations 169 6.5.2 OLR models for the acoustic comfort evaluations 170 6.5.3 Summary and discussions 171 6.6 Maps of the sound level/acoustic comfort evaluations 171 6.6.1 Maps for the age groups 172 6.6.2 Maps for the education groups 173 6.6.3 Maps for the groups of different sound level evaluations at home 174 6.6.4 Summary and discussions 175 6.7 Discussions 175 6.8 Conclusions 176 Chapter 7 • ANN Models for the Sound Preference Evaluations 178 7.1 Modelling the subjective evaluations for multiple single sounds 179 7.1.1 Data issues 179 7.1.2 Network construction 180 7.1.3 Model performance 181 7.2 Modelling the subjective evaluations for birdsong 181 7.2.1 A general model 182 7.2.2 An individual model 183 7.2.3 Group models 184 7.2.4 A lab model 185 7.2.5 Summary 186 7.3 Modelling the subjective evaluations for children shouting 186 7.3.1 A general model 187 7.3.2 An individual 188 7.3.3 Group models 188 7.3.4 Summary 189 7.4 Modelling the subjective evaluations for cars passing 189 7.4.1 A general model 189 7.4.2 An individual model 191 7.4.3 Group models 191 7.4.4 A lab model 192 7.4.5 Summary 193 7.5 Qnet models for predicting the subjective evaluations of bird song 193 7.5.1 A general model 194 7.5.2 An individual model 195 7.5.3 A group model 195 7.5.4 A lab model 196 7.5.5 Summary 197 VII 7.6 Development of sound preference maps 197 7.7 Conclusions 200 Chapter 8 - Conclusions and Future Works 203 8.1 Conclusions 203 8.1.1 Application of ANNs in soundscape study 203 8.1.2 Factors related to the sound level/acoustic comfort evaluations 204 8.1.3 Factors related to the sound preference evaluations 205 8.1.4 ANN models for predicting the sound level/acoustic comfort evaluations 206 8.1.5 ANN models for predicting the subjective evaluations of sound 206 8.2 Future works 207 References 209 Appendix I: Questionnaire of RUROS project 223 Appendix II: Questionnaire of Chinese social surveys 225 Appendix III: Questionnaire of laboratory experiment 231 Appendix IV: Relationships between social/demographical, physical, behavioural 234 and psychological factors viii List of tables Table 3.1 Summary of case study sites in the EU 42 Table 3.2 Plan and basic information of 14 case study sites in the EU (Kang, 2006a) 45 Table 3.3 The average SPL of 19 case study sites 53 Table 3.4 Noticed sounds (marked by ...J) in the case study sites 56 Table 3.5 Studied sounds in the laboratory experiments 60 Table 3.6 Description & categorization of variables 63 Table 3.7 Populations of social/demographical and behavioural variables 66 Table 4.1 Relationships amongst age and occupation, education, gender, and 78 residential status Table 4.2 Relationships amongst gender and occupations, education, residential 79 status Table 4.3 Relationships between occupation and education as well as between 80 residential status and occupation and education Table 4.4 Percentage (number) of the case study sites where significant correlations 81 or differences exist between pairs of social/demographical factors Table 4.5 Relationships between age, occupation, education, gender, residential 82 status and the sound level evaluations Table 4.6 Relationships between the sound level evaluations at home and the sound 85 evaluations on-site Table 4.7 Relationships between the sound level evaluations at home and age, 87 occupation, education, gender, residential status Table4.8 Relationships between social/demographical factors and the acoustic 88 comfort evaluations Table 4.9 Relationships between acoustic/physical factors and the sound level 89 evaluations Table 4.10 Relationships between acoustic/physical factors and the acoustic 93 ix