Indian Institute of Technology Kanpur Indian Institute of Technology Kanpur 1 Susham Biswas & Bharat Lohani Geoinformatics Laboratory Indian Institute of Technology Kanpur Kanpur, 208016 INDIA 20 th January , 2011 Sound Propagation Modeling at Sound Propagation Modeling at High Resolution Using LiDAR Data High Resolution Using LiDAR Data and Aerial Photograph for and Aerial Photograph for Outdoor environments Outdoor environments
73
Embed
Indian Institute of Technology Kanpur Susham Biswas Susham Biswas 1 Susham Biswas & Bharat Lohani Geoinformatics Laboratory Indian Institute of Technology.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
11
Susham Biswas & Bharat Lohani
Geoinformatics LaboratoryIndian Institute of Technology KanpurKanpur, 208016 INDIA20th January , 2011
Sound Propagation Modeling at High Sound Propagation Modeling at High Resolution Using LiDAR Data and Aerial Resolution Using LiDAR Data and Aerial Photograph for Outdoor environmentsPhotograph for Outdoor environments
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
77
Sound Modeling (Semi_emperical)
How sound can reach receiver R from source PS
Direct Transmission
Direct transmission of sound from source to receiver
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
88
Ground Reflected
Sound transmission after ground reflection
How sound can reach receiver R from source PS
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
99
Diffracted over and around sides of building
How sound can reach receiver R from source PS
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1010
Wall reflected
Transmission through reflection from wall
How sound can reach receiver R from source PS
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1111
Tree absorbedHow sound can reach receiver R from source PS
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1212
In outdoor environment sound from a source follows one or many of the above paths before reaching to a receiver location thus the outdoor Sound propagation involves following spatial parameters
1. Distance between Source* (PS) and Receiver (R)
2. Path_difference for Diffraction3. Path for Ground_reflection with ground type4. Possibility of Wall_reflection5. Extent of path length involved in
tree_absorption
* When there exists objects between Source and receiver the intermediate diffracting, reflecting points involved in transmission is termed as secondary source (SS) and original source as primary source (PS)
Spatial information required in sound modeling !!
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Spatial Data• Approximate estimation of terrain heights• Low resolution satellite image/ aerial photo• Use of Total Station/GPS in limited scale
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1616
Objectives of research
1. Can Lidar data be used for sound modeling- how?
2. Is there any Advantage of using high resolution spatial data?
3. Whether Better Data and/ or Model can lead to better prediction?
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1717
Observed Sound
at outdoor
Prediction of Sound at Outdoor with Poor Data
and Poor Model
Prediction of Sound at Outdoor with Good Data
and Poor Model
Prediction of Sound at Outdoor with Good Data
and Good Model
Comparison
Sound Prediction Schemes
Field Measurement
Is there any Advantage of using high resolution spatial data? Whether Better Data and/ or Model can lead to better prediction? Can better data suggest ways of improvement in sound modeling?
Can LiDAR data & Aerial Photograph be used for
Sound Modeling?
Methodology
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1818
Methodology
Field Measurement
Sound MeasurementSpatial data MeasurementNon_Spatial data Measurement
Sound Prediction Schemes
Poor Data and Poor ModelGood Data and Poor ModelGood Data and Good Model
Comparative Analysis
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
1919
Sound Model
Spatial Parameters
Non-Spatial Parameters
Spatial Survey
Measured Field Data
Design of Model
Design of Algorithms
PredictedSound for outdoor
location
Sound Prediction Scheme
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
2020
Design of Algorithm to extract Spatial ParametersData Preparation
For Good data: Accurate Data e.g. from Total Station, or LiDAR and Aerial Photographic Survey etc
For Poor data: Errors added to good data
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4242
A transact (130mX 30 m) at IIT Kanpur Air strip containing building, different ground types, tree, is used to monitor SPL and validate that with developed model
Validation experiment
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4343
Measured Sound at Receiver
Propagated Sound
Spatial Survey
Measured Sound at
SourceTotal Station
Reflector for TS
Experiment-Measurement of Sound and Spatial data
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4444
Computer
(Generation of Tonal
Sound)
AmplifierSpeaker(output device)
Experimental Detail- Sound Generation
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4545
Experi-mental Site
Parameter studied
Frequency (Hz) of Study
No. of Position
No. of Ht/position
Total No. Observation
Duration of measurement at each observ-ation
Metero-logical Condition
130mX 30m, At Air Strip, having building and different ground types
Sound Measured simultaneously at a fixed position at source as well
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4646
Measured SPL in dB
dBFrequency=250Hz
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4747
dBFrequency=250Hz
Predicted SPL in dB (for Good data and Good Model1)
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4848
Good data – Good Model1 -250 Hz
Deviation
in dB
Deviation between measured & predicted
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
4949
Good data – Good Model1 500 Hz
Deviation
in dB
Deviation between measured & predicted
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5050
Good data – Good Model1 1000 Hz
Deviation
in dB
Deviation between measured & predicted
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5151
Good data – Good Model1 4000 Hz
Deviation
in dB
Deviation between measured & predicted
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5252
Interdata analysis for three different schemes of prediction
•Mean and SD•ANOVA•Paired t Test•Tukey Test
Error Propagation
Intradata analysis for the best prediction scheme of the above three
Data Processing and Data Analysis
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5353
Mean STD Max MinFreq
250 7.12 6.55 39.87 0.01Good Data & Good Model 1
500 8.46 7.11 34.02 0.03Good Data & Good Model 1
1000 8.35 7.35 41.99 0.04Good Data & Good Model 1
4000 9.59 8.75 42.00 0.03Good Data & Good Model 1
250 7.75 7.46 42.54 0.01Good Data & Good Model 2
500 9.22 8.32 39.55 0.00Good Data & Good Model 2
1000 9.09 8.46 44.94 0.07Good Data & Good Model 2
4000 10.86 9.39 44.94 0.03Good Data & Good Model 2
250 11.99 9.45 52.73 0.15Good Data & Poor Model 500 16.64 12.42 51.62 0.02Good Data & Poor Model 1000 10.59 9.09 47.49 0.11Good Data & Poor Model 4000 10.16 9.49 41.99 0.10Good Data & Poor Model
250 12.86 9.65 52.24 0.15Poor Data & Poor Model 500 16.85 12.80 51.29 0.04Poor Data & Poor Model 1000 11.14 9.53 47.01 0.06Poor Data & Poor Model 4000 10.42 9.33 41.51 0.07Poor Data & Poor Model
Comparison of Different prediction SchemesStatistical Analysis
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5454
Frequ-ency
Probability of H0 to be true at 0.05 significance level
250 Hz 1.11*10-16
500 Hz 0
1000 Hz
0.0018
4000 Hz
0.5354
250 Hz
500 Hz 1000 Hz 4000 Hz
ANOVA statistical test for four prediction schemes
At least one of the four prediction schemes giving different results
Statistical Analysis
Abbreviations usedM1=Good data & Model1M2=Good Data & Model2GP=Good Data & Poor ModelPP= Poor Data & Poor Model
Comparison of Deviation
µ(D)M1=H0: µ(D)M2=µ(D)GP= µ(D)PP
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Abbreviations usedM1=Good data & Model1M2=Good Data & Model2GP=Good Data & Poor ModelPP= Poor Data & Poor Model
Statistical Analysis
Comparison of different pairs of prediction schemes in terms of deviation
t(0.05,120)=1.645
H0: µ(D)1= µ(D)2
Generally the pairs of prediction schemes are not giving similar results
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5656
250 8.31500 8.311000 9.674000 10.98
Error in DeviationFreq
.
Propagation of Error
Data analysis
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5757
Hence, generally the prediction scheme involving Good Data & Good Model1(Coherent) seems to performed the best
Analysis
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5858
Study area: part of academic area of IIT Kanpur. Buildings are shown with red and ground with blue
SPL (in dB) at different locations due to distance attenuations
SPL (in dB) at different locations due to ground attenuations
Stages of sound prediction over LiDAR data points
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
5959
SPL (in dB) at different locations due to barrier attenuations
SPL (in dB) at different locations due to distance + barrier attenuations
Binary plot of probable reflecting and non-reflecting points
Stages of sound prediction over LiDAR data points
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6060
Sound propagating from a single source to nearby points in 3D
Sound map developed by incorporating LiDAR data/Google Earth Image of IIT Kanpur inside sound model
Representation of Sound
250 Hz sound of 90 dB propagated
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6161
Perception of a street noise at different spatial location
Audio Realization
G T Road
IIT Gate
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6262
Discussion and Conclusion
• LiDAR/Aerial Photo data can be use to incorporate detail terrain information for outdoor
sound propagation modeling • In general Good data and Good Modeling (complex coherent) scheme is giving the best results which answers the research question whether better data and model can lead to better results.• Present study indicates the technique to determine principal paths of propagation even for
complex terrain
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6363
• In indicates techniques to incorporate accurate spatial parameters such as path-difference, ground type, angle of reflection, barrier shape etc which were not been possible previously for real outdoor sound modeling.
• It can be used for 3D sound mapping rather than conventional 2D mapping
• It can generate higher resolution sound map/sound contour
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6464
Thank you !!!
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6565
D. A. Bies and C.H. Hansen. Engineering Noise Control., Theory and practices Unwin Hyman (2003)
"The Propagation of Noise from Petroleum and Petrochemical Complexes to Neighbouring Communities". CONCAWE Report 4/18, (1981).
ISO 9613-2, 1996(E), ‘Acoustics-Attenuation of sound during propagation outdoor-Part 2: General method of calculation’, p.1-18.
Maekawa, Z 1968, ‘Noise reduction by screens’, Applied Acoustics, 1, p. 157–173.
RTA group (n.d.),ENM-Environmental Noise Model-Program Specification, viewed 7 December, 2007<http://www.rtagroup.com.au/enm/environmental_noise_model.html>.
Renzo Tonin 2004, Modeling and Predicting Environmental Noise, viewed 12 November 2006, <http://www.rtagroup.com.au/pdfs/22.pdf>.
Soundscape, further reading, viewed 7 september 2007, http://en.wikipedia.org/wiki/Soundscape#Further_reading
Important References
Noise Mapping, Assesment of data sources and available modeling techniques- are they good enough for comprehensive coverage by computer noise mapping? (2002), http://www.cerc.co.uk/services/Noise%20Mapping%20CERC%20IofA%20Feb2002.pdf, viewed 17 January 2007
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6666
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6767
Coherent Addition of two 40dB Tonal Sounds at Diff. Phase Difference
Simulations to assist theoretical understanding and research findings
Phase Difference in deg.
SPL in dB
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6868
In-Coherent Addition of two 40dB Tonal Sounds
Simulations to assist theoretical understanding and research findings
SPLcombined_40_40=43.01 dB1
Effect of Background Noise
In-Coherent Addition of two Tonal Sounds 80dB, 40dB
SPLcombined_80_40=80.0004 dB
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
6969
SourceReceiver
Path_diff=1 mBarrier Attenuation for a
tone of 250 Hz=13.16 dB
tone of 4000 Hz=25 dB
Simulations to assist theoretical understanding and research findings
How different frequency sounds are affected by same geometry
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
7070
High Frequency wave (shorter wavelength) 2
Low Frequency wave (higher wavelength)Spatial distribution
of Interference maxima
Wavelengths of 4 tonal sounds used
250Hz1.36 m500Hz0.68 m1000Hz0.34 m4000Hz0.085 m
Interference and Frequency
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
7171
10 m 10 m
5 m
Path_difference=2.36 m 5 m
50 m 50 m
Path_difference=0.49m
Why at shorter distance prediction is more dependent on accurate data
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
7272
Determination of Principal Path for over top of buildings –after points of intersection and vertical lines of intersection are determined ( It tries to determine path from PS to R via secondary source(s) (SS))
1. Straight line is drawn between PS and R , if PS-R line is not been intersected by any line of intersection Direct transmission else
2. All the intersecting point/line below PS-R are eliminated (if any)
3. Straight line is made between PS and tallest intersecting point, if this line is not been intersected by any line of intersection then tallest intersecting point becomes a secondary source (SS). And iteration continues from SS as above till sound reaches the receiver.
4. If this line is being intersected then, tallest amongst them becomes the SS .And iteration continues from SS as above till sound reaches the receiver.
Indian Institute of Technology KanpurIndian Institute of Technology Kanpur
Susham BiswasSusham Biswas
7373
Determination of Principal Paths around sides of the buildings –after points of intersection and lines of intersection at side walls are determined ( It tries to determine path from PS to R via secondary source(s) (SS))
1. Line is drawn between PS to R , all lines of intersection not intersected by this line will be deleted for the current iteration. Rest of the intersecting points along with lines of intersection will be tested for finding SS
2. Among the available lines of intersections nearest one from PS is chosen first. Two intersecting points attached to it becomes the first pair of SS
3. From each of the above SS iteration continues seperately4. Straight line is drawn between SS to S and checked for finding
intersecting ‘lines of intersection’. When there is no such line or only one line belonging to same building then there is no further SS in the principal path. When there are two or more such line but all belonging to same building from which the iteration is being tested then the next SS belongs to same building. When there are two or more such line but belonging to same or different building then, it tries to select intersecting point(s) attached to nearest ‘line of intersection’ of either building maintaining criteria of shortest route to reach R.
5. Step 4 repeats iteratively till sound reaches to R