Outdoor Sound Propagation Modelling Across Scales with Uncertainties

The proposed project sets the following four objectives:
1. To study the uncertainties in the meteorological, ground and other environment
data which are used to model short, medium and long range outdoor sound
propagation.
2. To determine through modelling the influence of uncertainty in the input model
parameters on the accuracy of the model predications for sounds emitted by a
given military platform.
3. To develop and integrate several propagation models in one Universal Modelling
Tool for Outdoor Sound Propagation (UMTOSP) which is accurate over a range of
scales within the agreed bounds.
4. To validate the UMTOSP and demonstrate the accuracy and flexibility in terms of
calculating the sound resulting from a range of military platforms and in the
presence of realistic uncertainty in data.
The work will be built on the latest developments in acoustic propagation modelling. The
new universal model which will be developed as a result of the project will be user-friendly
and it will be able to calculate outdoor sound at one or more positions in range and height
from a broad range of sources, in a representative range of environments, at different
scales, on ground and at low and high altitudes.
This part of the project will include a detailed investigation into the capabilities of different
propagation models but, for the purposes of planning the project it is expected that the
UMTOSP is based on:
i. robust analytical models (e.g. [4,5]) to calculate short range sound propagation
when the ground impedance effect is dominant;
ii. energy-based boundary element methods to study short and middle range
sound propagation from complex sources and in the presence of complex
terrain and strong ground impedance effects (e.g. [6]);
iii. green's function, or time domain finite differences, parabolic equation methods
(e.g. [7,8]) when medium and long range propagation in the presence of
complex terrain is affected by the meteorological conditions.
In this way, the models should include effects observed at short range, 10s to 100s of
metres, such as the presence of local topography or buildings, medium range, 100s to
1000s of metres, such as ground impedance, wind conditions, and long range, 1000s to
10,000s (and more) of metres, such as temperature and wind profile.
The sensitivity of these models (and the universal model) to the uncertainties in the input
data will be studied using the Bayesian approach (e.g. [9]); this will enable us to
understand the errors in the calculations these models produce when inaccurate or timevariant
sets of source, meteorological, topographical and ground type input data are input
for a given source / receiver geometry.
When combined in one UMTOSP these models will enable us to calculate sound
propagation (for the first time) in an environment of any complexity and with a minimum
number of parameters within the set accuracy limits and propagation ranges.
The main output from the project would be software code for the UMTOSP that could, if
appropriate, be used as the basis for a tactical decision aid.