12.5 Source-Receptor Matrix Approach




In the previous section, to determine the location or magnitude of an unknown source, the dispersion model was integrated backwards in time from a measurement location to determine the geographic area of potential sources. To quantify the emission rate in these calculations, we assumed that the dilution factors are about the same in the forward or backward integration of the model.

  1. Start by retrieving the previously saved captex_control.txt and captex_setup.txt settings into the GUI menu. Then as before, set a 19 hour run duration which limits the results to the 3-hr measurement data. Following the same approach as was used in the trajectory matrix, open the source location menu for three locations and enter the lower left (38,-85), upper right (45,-78), and first internal grid point (38.5,-84.5). This will create a regular grid of emission points at 0.5 degree resolution.

  2. Then open the concentration grid menu and change the sample starting time to 83 09 26 09 00 to simplify the subsequent analysis, which will result in only one output time with one 03 hour average output at the end of the 3-hr measurements. Save the change and open the pollutant menu and change the emission rate from its actual value to a unit value, in this case 0.333333 so that one unit of mass is released after a duration of 3.0 hours. Save the change and close the Setup Run menu.

  3. Now open the Advanced / Concentration / Conversion Modules menu #10 and check the radio-button to restructure the concentration grid to the source-receptor format. The effect of this switch is that instead of all sources contributing mass to a single concentration grid, each source has its own concentration grid, which can be treated as a matrix with each row representing a source location and the columns representing the receptors (the concentration grid points). Plotting the values on a row gives a concentration map for that source. Plotting the values in a column gives the contribution of each source to that receptor.

  4. Save the changes to close the menu and then run the model from the Special Runs / Matrix menu tab. The startup message that some points are off the grid is because the NW corner area of the source domain is outside of the meteorological domain. This does not affect the calculations at the remaining points.

  5. When the run completes, press the Display / Source-Receptor / View menu tab, to open the display menu. This menu calls a special program to extract the source or receptor information from the matrix output file. It writes out a single binary file that can be displayed with the normal concentration plotting program or any other concentration utility. Rename the binary output file to give it a unique name, such as SRM_fwrd.bin. The extraction method radio-button should be set to source, which means we are going to extract the row computed from the tracer source location entry box values: 39.90 -84.22. Then execute to view the plume dilution factors for the 26th from 0900-1200. The source location is rounded to the nearest half degree point.

  6. Go back and check the receptor radio-button and enter the location of sampler 316 at 41.3 -82.22 and give the output file a unique name, such as SRM_back.bin The resulting source display map shows the dilution factors for sources contained within the contoured region. The dilution factors can then be divided into the measurement to estimate the emission rate from that source location.

  7. Naturally the forward and backward dilution factors must be identical because they are all derived from the same forward calculation. This can be checked by using the Grid to Station analysis as was done in the forward-backward analysis tutorial.

  8. For future reference save the CONTROL file to matrix_control.txt and the SETUP.CFG file to matrix_setup.txt.

The results shown here are another approach to source location from the source perspective, relying only on a forward calculation from all potential source locations. Simply selecting a receptor location will show the contributions of all sources contributing to that location. The computation can be very resource demanding, in terms of time and memory, because a unique concentration grid is required for each source.