The simulation of remotely-sensed images of scenes containing gaseous plumes is of great interest to many researchers in both government and industry. Sensor and algorithm development efforts can be structured around simulated data only if the plume model is reliably accurate. Consequently, a gas plume model has been integrated into the DIRSIG environment for these studies. This model provides a simplified spatial distribution of gas concentration and temperature, while performing a full radiative transfer integration along the ray-path from the sensor through the plume. The plume model as implemented in DIRSIG will be described here.
A "plume" in the DIRSIG context is a release of gas into the atmosphere from a point source. For DIRSIG to successfully render the plume, as with all materials in the scene, the optical properties of the gas must be known. Physical properties such as gas density and temperature are also required. Full treatment of the transmission through and emission by the plume are incorporated and described below. Currently, DIRSIG does not support any scattering effects in the plume. Consequently, plumes with large water-droplet concentration, such as those from cooling tower stacks, are not able to be simulated within DIRSIG. This feature will be implemented in future versions of DIRSIG.
The following sections briefly describe the general theory of plume modeling, concentrating on different types of models and their general strengths and weaknesses. The plume model integrated into DIRSIG is then explained in some detail, followed by instructions on how to run the model and a detailed description of the absorption spectra data format and the configuration file parameters required to simulate a gas plume within the scene.
In general, the goal of all plume models is an accurate simulation of the hydrodynamic and thermodynamic expansion and cooling physics describing how the plume is spatially distributed and how the plume thermodynamic properties vary with time. Some plume models are designed to provided a snapshot of the gas, thus removing the temporal nature of the model. Others fully include the temporal effects of the gas inherently. Regardless of the temporal treatment of the model, a spatial distribution of the gas density and temperature is the required output for integration into a remote sensing simulation such as DIRSIG. Those plume models that only predict Ground Level Concentrations will not be sufficient as effects due to viewing geometry through the plume can not be correctly simulated given only a two-dimensional distribution of the gas density at ground level. Observed phenomenology such as self-absorption require the model to provide the temperature and concentration of the gas in three-dimensions. Utilization of a three-dimensional plume also allows the user to explore a parameter space including geometrical effects.
The DIRSIG plume model has been demonstrated to successfully reproduce spectra observed in gas releases from factory stacks for both up-looking and down-looking geometries. The applications can be extended to emissions from roof vents or underground facilities. Phenomenology such as signature dependence on temperature contrast with the background have been demonstrated in the model and can be used to test and train algorithms. In addition, minimum detectable quantities and sensor design issues can be studied with synthetic imagery of gas plumes.