Model Change Bulletin MCB#3 3/18/93 CTSCREEN (dated 93076) This Model Change Bulletin is to notify users of changes made to correct problems in CTSCREEN (dated 91107). The specific changes, the problems addressed, and the consequences of the changes are summarized in the following. Reevaluation of the revised models demonstrated that these modifications did not affect design concentrations and, thus, should not have a significant regulatory impact. These revisions constitute CTSCREEN (dated 93076) and completely replace CTSCREEN (dated 91107). With this bulletin, the source codes and executables for CTSCREEN (dated 93076) are being made available for downloading from the Air Quality Models area of the File Transfer Section of SCRAM. CTSCREEN (dated 91107) is obsolete and, with this bulletin, will no longer be available on SCRAM. All copies of such should be discarded. Users of CTSCREEN are referred to a companion WordPerfect 5.1 document (CTPLSMOD.WPF) "Modifications to the CTDMPLUS and CTSCREEN Models" for more detailed discussion of these problems (see Documention below). This document addresses problems which are common to both CTDMPLUS (dated 91107) and CTSCREEN (dated 91107) and those unique to CTSCREEN (dated 91107). Sections 1 and 2 below follow the outline of the companion WordPerfect document. All of the problems described in the following have been corrected in CTDMPLUS (dated 93076) and CTSCREEN (dated 93076). 1. PROBLEMS COMMON TO BOTH CTDMPLUS (dated 91107) AND CTSCREEN (dated 91107) 1.1 Wrap Calculations Two problems are related to situations where a source is very close to the terrain feature: 1) model adjustment of WRAP ellipse size to prevent a source from being located inside the fitted ellipse (to prevent a mathematical error) and 2) a travel time problem for receptors located on actual terrain contours that 'jut out' in the upwind direction from the fitted ellipse. Both problems may inappropriately affect concentrations during stable/neutral conditions. The second problem, in particular, may result in significant overestimates. 1.2 Convergence Problem The algorithm in CTDMPLUS and CTSCREEN for computing the position of the streamline that passes through the source occasionally has problems converging on a solution. This results in the error message "ENDLESS LOOP IN PATH" and a null concentration calculation for the given simulation. Calculations for other simulations are not affected. This error occurs only in stable/neutral conditions. 1.3 Meteorological PROFILE file CTDMPLUS and CTSCREEN expect that the instrumentation levels for the meteorological tower are entered in increasing order in the PROFILE file. The check for this was coded incorrectly. In some cases this may result in an abnormal termination with the error message "PROFILE HEIGHT VALUE INCORRECT ..."; this may occur even when the data are entered correctly. This coding error does not affect concentration estimates. 1.4 Limit on Convective Scaling Velocity CTDMPLUS and CTSCREEN place a lower limit on the convective scaling velocity of 0.167 times the wind speed half-way between stack-top and plume height. In multiple stack situations this algorithm may result in an incorrect scaling velocity. This affects CTDMPLUS calculations in convective conditions only. It has no affect on CTSCREEN estimates because wind speed is assumed to be constant with height; thus, in CTSCREEN, the minimum convective scaling velocity will be the same for all stacks regardless of final plume rise. 1.5 Negative Emission Rates The models do not currently allow negative emission rates. This prevents the use of negative emission rates to estimate air quality improvement when an existing source is retired. 1.6 Source Contribution Table The CTDMPLUS and CTSCREEN option for source contribution tables potentially can result in an extremely large output file (CTDM.OUT). To avoid this, source contribution tables in CTDMPLUS (dated 93076) and CTSCTEEN (dated 93076) are written to a separate file (ASCII or Binary). 1.7 Special handling of receptors below stack top An additional switch (ISTKTP) has been added to the CTDM.IN file. Setting this switch to 1 will cause the contribution from a particular source to be set to 0 for all receptors with elevations below stack-top for that source. This switch will facilitate analysis of results for situations requiring both simple and complex terrain modeling. 2. PROBLEMS AFFECTING CTSCREEN (dated 91107) ONLY 2.1 Calculation of Convective Scaling Velocity CTSCREEN calculates an initial estimate of the convective scaling velocity (w*) using the default mixing height of 50 meters obtained from the surface input file. Later in the program when mixing heights are calculated as a function of hill height, CTSCREEN fails to recalculate w*. This affects concentration calculations in convective conditions. 2.2 Determination of Mixing Height In CTSCREEN the mixing height is determined as a function of hill height. This results in two problems. First, in multiple hill situations, modeling is performed using mixing height estimates for all hills regardless of where the receptors may be located. This is incorrect, modeling should only be performed using mixing height estimates for the hill on which the receptors reside. Secondly, there are occasions when the effective plume height exceeds the maximum mixing height estimate for a given hill. This results, erroneously, in a zero concentration estimate for the given simulation. 2.3 CTSCREEN Output Files As a result of our own experiences and those of many users, the output files from CTSCREEN (dated 91107) were determined to be inadequate for fully analyzing the model results. CTSCREEN (dated 93076) provides improvements, including a separate meteorology file (METDAT) which provides a better summary of the meteorological conditions and facilitates postprocessing of concentration estimates. In addition, the menu driver has been modified, as necessary, to accept this new (METDAT) file. 2.4 CTSCREEN Error Messages Diagnostic messages (in the CTDM.OUT file) are labeled with the wrong simulation number (one less than the correct number). This is a result of the error message being written before the counter is incremented. This error does not affect concentration estimates. 3. Documentation Modifications to correct these problems have been completed and are described in a companion WordPerfect 5.1 document (CTPLSMOD.WPF) "Modifications to the CTDMPLUS and CTSCREEN Models". This document provides detailed discussion of these problems, how they were resolved, the likely effect of the modifications on modeled concentrationss, and the results of a model reevaluation. It is available for downloading from SCRAM as file CTPLSMOD.ZIP. Additional material is provided in the WordPerfect 5.1 document (CTSCSUP.WPF) which describes modifications to the CTSCREEN user's guide. The latter document is available for downloading from SCRAM as CTSCSUP.ZIP. 4. CTSCREEN Design Value Comparisons CTSCREEN (dated 93076) was tested and reevaluated using the Lovett, Westvaco, and Widows Creek data sets from previous evaluations. For the Lovett data base, the overall maximum concentration decreased by 8.7 percent but, remained well above the observed maximum at Lovett; for the Westvaco data base, there was no change in the overall maximum concentration; for the Widows Creek data base, the overall maximum concentration decreased by 14.1 percent but, remained well above the observed maximum at Widows Creek. 5. Contacts for Questions Questions related to the use of CTDMPLUS in specific regulatory applications should be directed to the appropriate regulatory agency or Regional Modeling Contact (see State/Local Modeling Contacts in the Agency Communications Section of SCRAM). Generic questions related to the use and regulatory applicability of the model should be directed to Desmond Bailey via SCRAM EMAIL or phone (919) 541-5248. Questions or problems related to program execution should be reported to Donna Schwede via SCRAM EMAIL or phone (919) 541-3255 or to Steve Perry phone (919) 541-1341. Users who would like to share comments, suggestions, or helpful hints are encouraged to use the Model Conferencing area on the SCRAM Bulletin Board (see Model Conferencing area in the Public Communications Section of SCRAM). Technical correspondence should be directed to: Dr. Steven Perry or Ms. Donna Schwede Applied Modeling Research Branch ACMD/AREAL/USEPA MD-80 Research Triangle Park, NC 27711 Fax: (919) 541-1379