Abstract:The general problem of characterizing gas source parameters based on concentration measurements is known to be a difficult task. As many inverse problems, one of the main obstacles for accurate estimation is the non-uniqueness of solution, induced by the lack of sufficient information. As the number of detectors is lowered, which is more than a plausible scenario for many practical situations, the number of possible solutions that can characterize the source increases dramatically, leading to severe errors. In this paper, a Lagrangian stochastic based method for identifying these suspected points, which will be referred to as 'degenerate space', is formulated and analysed. Then, a new procedure for quantitative prediction of the effect of deploying a new detector in space is used to design an adaptive scheme for source term estimation. This scheme has been tested for several scenarios, differing by the location of the initial detectors, and is shown to reduce dramatically the degeneracy formed by insufficient information. The combined formulation of degenerate space with the new adaptive scheme is shown to give improved accuracy, and in particular for a relatively small number of detectors.
Computational Physics,Instrumentation and Methods for Astrophysics,Atmospheric and Oceanic Physics,Fluid Dynamics,Geophysics
What problem does this paper attempt to address?
### Problems the paper attempts to solve
The paper aims to solve the difficult problem of characterizing gas source parameters based on concentration measurements. This is a typical inverse problem, and the main obstacle lies in the non - uniqueness of solutions, which is caused by the lack of sufficient information. As the number of detectors decreases, this situation becomes more severe, the number of possible solutions increases sharply, resulting in serious errors.
Specifically, the paper proposes a method based on the Lagrangian Stochastic Model (LSM) to identify these suspicious points, called "degenerate space". Then, the paper proposes a new procedure for quantitatively predicting the effect of deploying new detectors in space and designs an adaptive scheme to estimate the source term. This scheme has been tested in multiple scenarios, and the results show that it can significantly reduce the degeneracy caused by insufficient information. Combining the degenerate space and the new adaptive scheme, this method can accurately estimate source parameters with a small number of detectors.
### Key points
1. **Source term estimation**: Determine the location and emission rate of gas sources through concentration measurements.
2. **Backward Lagrangian Stochastic Model**: Used to identify suspicious points in the degenerate space.
3. **Optimal design**: Optimize the deployment location of detectors through an adaptive scheme.
### Background
Solving the source term estimation problem (i.e., determining the location and emission rate of the source based on a given set of measurements) is very important for many practical applications. For example, in the process of evaluating the risks of biological or chemical releases, and identifying sources in accidental release events in industrial areas, accurate estimation of source parameters is required.
In the past two decades, new methods for the Source Term Estimation (STE) problem have been significantly developed and tested. The two common methods are optimization methods and probability methods. Optimization methods find the best combination of source parameters by minimizing the cost function (usually the sum of the squared differences between the predicted concentration and the observed concentration). Probability methods construct the probability density function of source parameters based on Bayesian inference theory.
Regardless of which method is chosen, a reliable diffusion model is required. However, the computational cost of the diffusion model is very large, which is a challenge in practical applications. To reduce the computational cost without losing accuracy, some researchers have adopted the adjoint source - receptor relationship instead of the standard forward diffusion model.
### Main difficulties
The main difficulty in solving this problem is that determining source parameters from measurements is an ill - posed problem, that is, the solution may not exist, be non - unique or be discontinuous. Increasing the number of detectors can gradually solve this problem, but it may be infeasible to deploy a dense detector network over a large area.
### Solutions
The paper proposes a new method to reduce the non - uniqueness of solutions in two steps:
1. **Identify degenerate space**: Based on a given set of measurements, a simple and efficient mechanism is proposed to identify suspicious points in the parameter space. The construction method of the degenerate space will be described in detail in section 2.2 of the paper.
2. **Select the location of new detectors**: Based on statistical evaluation, a criterion for selecting the location of new detectors is proposed to maximize the expected reduction in degeneracy. These two steps can be iteratively combined to generate an adaptive algorithm that uses existing measurement information to plan the next measurement.
### Theoretical basis
1. **Lagrangian Stochastic Model**: The paper uses the Lagrangian Stochastic Model (LSM) developed by the Israel Institute for Biological Research (IIBR), which can consistently describe complex gaseous diffusion phenomena.
2. **Degenerate space**: Define the degenerate space as the set of all equally probable hypotheses. Through a recursive method, the degenerate space can be updated after each new detector is added.
3. **Adaptive degeneracy reduction**: By predicting the effect of the location of new detectors, select the optimal location to reduce the degenerate space.
### Results
The paper tests the adaptive source term estimation algorithm in six different initial detector arrangements and source parameter cases. The results show that the algorithm can quickly converge to the correct source parameters, and can significantly reduce the degenerate space by gradually adding detectors even if the initial arrangement is not good.
### Summary and discussion
The paper proposes a new adaptive detector deployment method, aiming to reduce the existence of multiple hypotheses for describing source parameters. This method depends on two concepts: identifying suspicious points based on current measurement values