CALPUFF - Sulphur dioxide ambient modelling

This case study summarises a project carried out for Carter Holt Harvey at Kinleith Pulp and Paper Mill (Tokoroa, New Zealand) which compared two different types of atmospheric dispersion model with actual ambient monitoring data for SO2.  One of these models is CALPUFF.

The first modelling study was carried out a the Kinleith Pulp and Piaper Mill in 2003 and used a dispersion model known as ISC-PRIME. That was the best applicable at the time but was limited in ability to handle the complex interaction of meteorology and terrain around the Mill site. The results of this modelling showed that SO2 concentrations in the receiving environment could be problematic, and that the Mill may need to restrict fuel options for the lime kilns it operates or reduce SO2 emissions from the recovery boilers to meet the required air quality standards.  However, before any such decisions were to be made, a period of ambient monitoring was carried out to confirm the modelling results.

Subsequent ambient monitoring over the period 2005-2007 showed that actual ambient concentrations of SO2 were much lower than the modelling results from 2003.  This raised the question of whether the modelling or the monitoring was more representative of potential worst case concentration of SO2.  Could the modelling have overpredicted concentrations, or were the monitors placed in the wrong position to be exposed to peak concentrations?

To answer this question, a second modelling study was carried out by Beca AMEC in 2009 using an advanced dispersion model known as CALPUFF to try and address this question.  The objective of the 2009 modelling study, which is the subject of this paper, was to determine whether the reason for the disparity between 2003 modelling results and subsequent monitoring data lay in the location of the monitoring stations, the emission data used in the model,  or overly conservative modelling methodology.


The objectives of the study were to provide answers to the following questions: 

  • Did the modelling (from 2003) use emissions data which reflected reality over the 2005-2007 monitoring period?
  • Were the modelling results too high or in the wrong place due to the type of model, meteorological data, and input data used at the time?
  • What does the best possible accuracy of input data and model tell us about expected SO2 concentrations in the ambient air?
  • Were the ambient monitors in the right place to capture peak concentrations?
  • Does the modelling (from 2003) or the monitoring  (from 2005-07) give a more accurate indication of maximum ambient SO2 concentrations downwind of the Kinleith Mill.



  • CALPUFF was run with a full three-dimensional wind field generated by its associated meteorological processor, CALMET.  
  • The same data period as the monitoring,  May 2005 to March 2007, was used in the model. 
  • The outputs from the CALPUFF model were also compared with the ISC-PRIME model produced in 2003 to illustrate the difference between the two dispersion models.


Outputs or findings

The 2003 modelling study concluded that if high-sulphur content fuels are burned in the lime kilns, there is a potential for some isolated areas to be impacted by SO2 concentrations approaching or exceeding the relevant ambient air quality criteria.  However, this conclusion was not supported by the ambient monitoring data collected from two locations near the Mill in 2005-2007.

It was found that generally, the SO2 emission rates used in the 2003 modelling study were rather conservative and did not reflect the present (2009) situation.  In addition, even with the same emission data the new modelling results showed about the same predicted SO2  concentration in Tokoroa as the ISC-PRIME model used in 2003, but much lower concentrations of SO2 near to the Mill and to the east and southeast.  The reason for the lower concentrations is likely to be the more advanced handling of plume buoyancy and plume movement around complex terrain provided by the CALPUFF model.

The CALPUFF model results matched well with the monitored data, and indicated that maximum expected ground level concentrations of SO2 were much lower than the National Environmental Standard for 1-hour averaging times, and were also much lower than both the NZAAQG and WHO Guidelines for 24-hour averaging times. 

The CALPUFF modelling results did not indicate any areas of local maximum ground level impacts or “hotspots” that should have been used as monitoring locations.  Therefore, it was concluded that the two monitoring sites were appropriate. 

With the updated emissions data and modelling results, there was no longer any great discrepancy between the modelled predictions and the ambient monitoring.  Both modelled and monitored SO2 concentration results showed similar concentration patterns.




The CALPUFF model results with the updated emissions data matched well with the monitored data, and indicated that maximum ground level concentrations of SO2  are much lower than all of the relevant air quality assessment criteria. This meant the Mill could continue with its plans to utilise a variety of fuels in the lime kilns, in order to maximise sustainable use of waste gas, wood, marine oil, and tall oil residues, and minimise the use of virgin fossil fuels. 

The study highlighted the useful application of dispersion modelling to determine whether fuel management options have an influence on adverse environmental effects, but also highlighted the importance of using the best possible accuracy in the model methodology and emissions assumptions.  The study also demonstrated the benefit of using ambient monitoring to support or confirm modelling results. 

Associated Models


CALPUFF is an advanced non-steady-state meteorological and air quality modeling system.


Freeman, Tracy : 2011 - Sulphur Dioxide Ambient Monitoring and Dispersion Modelling Comparison Study, Kinleith Mill

65th Appita Annual Conference and Exhibition, Rotorua New Zealand 10-13 April 2011: Conference Technical Papers