A novel change point approach for the, detection of gas emission sources using remotely contained concentration data

We consider a multivariate epidemic mean change model with dependent errors where the mean changes in each dimension at a time point t1 and returns back at a time point t2. These two change points can dier in each dimension but each change point has a functional connection depending on the dimension. To find t1 and t2 in each dimension we developed an asymptotic testing procedure. Therefore we use two different types of test statistics, the multivariate test statistic and the projection test statistic where we transform the multivariate data into univariate data.
For simulations we consider the situation that we search for a gas emission source in a big area. So we assume that we have data from an air plane which measures the gas concentration in the air.
Our testing procedure helps us to decide whether there exists a gas emission source in this area. If there is a source we want to estimate its coordinates as near as possible to the real location. Therefore we assume that outside of the gas plume the data have a constant mean and inside the plume the mean increases to a higher level. With the knowledge of the form of the gas plume and the gas concentration with the corresponding coordinates of the measurement points we can draw conclusions for the location of the gas emission source.
Additionally we use our method for real data to locate a landfill.