OUTLINE :

1. Introduction 

2. Project Goals and Scope of work

3. Plume Rise Model Equations

4. Obtaining data

5. Using ArcMap to allocate location of fire 

6. Plume Rise Model Programming using VB v.6

7. Calculation Results

8. Conclusions

9. Future work

10. References

11. Acknowledgements

Introduction

“Fire plume model”, developed by David F. Brown and William E. Dunn, University of Illinois, is a prospective model for plume rise because of its simplicity. Inputs of this model are metrological data (wind speed, temperature), radius of fire and other physical parameters. This study will calculate the height of plume rise based on the 2000 TexAQS period within the Super-COAST (Coastal Oxidant Assessment for Southeast Texas). 

Click here for Super-COAST map

Back to TOP

Project Goals 

This study concentrates on  calculating and evaluating Fire plume model for Forest Fire using GIS as one of the modeling tool. Acreage burned, fuel consumption and emission factor have recently been identified by crews in CEER and stored in Excel database. To link this data with meteorological data, ArcMAP  is used to build grid information ,facilitate manipulation and display of data. To obtain  meteorological data,  Fortran codes were written to extract this data from MM5 in CAMx input files and MM5 output files itself, and reformatting it into Microsoft Access format. Visual Basic programming was used to assemble meteorological data into Fire plume calculation. 

This project accomplishes the following tasks:

• Incorporate grid information to fire locations using ArcMAP

• Extracting meteorological data into Microsoft Access Database

• Visual Basic Programming of Fire plume calculation 

Scope of work

The most detailed data was collected for the HG-BPA (Houston-Galveston, Beaumont-Port Arthur) sub-domain (red frame). I, therefore, limit this study only on fires occurred in HG-BPA domain. For Texaqs study, modeling episode is between August 22^nd- September 1^st. Thus, the metrological data is available only for these days. To observe the performance of plume rise model, only fire that burned more than 500 acres were chosen. 

Back to TOP

Fire Plume Model (Brown et al., 1999)

FIREPLUME model was developed to simulate atmospheric dispersion and air quality impacts from fires and other smoke source. The FIREPLUME model predicts the ground-level concentration field resulting from chemicals emitted from or within 1) instantaneously discharged thermals or explosive discharges, 2) fires that generate hot continuous plumes 3) smoldering or decaying fires. FIREPLUME model is an extension of a Monte Carlo Lagrangian particle model (MCLDM). The ability to treat buoyant plumes was added to MCLDM in support of an impact analysis for fire. 

In this model, plume rise of fires can be calculated according to the Brigg’s two-thirds law, which is applicable in cases where the buoyancy source has low initial momentum. Fires clearly fall into this category.   

(1)

where  Dh is the plume rise, x is the downward distance, r_o is the fire radius (m), K is the velocity ratio, b is the entrainment coefficient and F is the Froude number. The Froude number is:

(2)

where w_o is the initial  vertical velocity [m/s], r is the air density [kg/m³]and Dr is the initial density difference between ambient air and the fire plume.

Final rise

Due to the effects of entrainment into the plume and thermal stratification of the atmosphere, the rise of the fire plume is limited. It can be calculated depending on stability class (stable, neutral, unstable).

For stable conditions, plume rise is limited by thermal stratification. Based on a survey of field data Briggs (1984) suggests that the final rise in stably stratified air is

(3)

where Dh_f is the final rise in [m] and N is the Brunt Vasaila frequency. 

(Arya, 1999)

For neutral conditions, ambient turbulence limits the rise by breaking up the plume. The final rise in this case is the level at which the internal turbulent dissipation rate of the plume matches the ambient turbulent dissipation rate.

(4)

where u_* is the friction velocity [m/s].

 In unstable conditions, plume rise is also limited by turbulence. However, the turbulent dissipation of the downdrafts is equated with the plume dissipation rate since downdrafts are responsible for bringing elevated material to ground level.

(5)

where z_i is the inversion height [m] and w_* is the convective velocity scale [m/s].

(6)

In practice, the neutral plume rise relationship serves as a limiting case within our framework since both the stable and unstable limiting rise relationships go to infinity as neutral conditions are approached. Use the lesser of the stable or unstable final rise estimate (which ever is applicable) and the neutral estimate.

Therefore, in unstable conditions the final rise is given

whereas the stable conditions

The basic theory of plume rise (based on the so-called Brigg's formulas) is limited to assumptions of uniform wind field and constant stability in the atmospheric layers through which the plume rises.

Input data required in this model are

Click here to view CAMx vertical layers

Back to TOP

Obtaining Input Data

Emission Inventory Data

During Summer 2002, emission inventory associated with forest, grassland, and agricultural Burning during Texaqs period, namely August 1^st- September 30^th, have been studied and reported by CEER (Allen et al., 2002). The construction of this inventory involves estimated acreage burned, fuel consumption and emission factors. This study faced a problem of that individually, information on some of the fire events is well documented, but the magnitude and the spatial and temporal distribution are not well characterized. Because of lacking in knowledge of temporal distribution, each fire event is assumed to last for an entire day of the day the fire reported to occur. About 2000 fire events were estimated for the Texasqs period on a daily basis and were allocated to specific locations using either latitude and longitude or counties that fires occurred. All data are stored in Microsoft Excel database.  

From Excel database, fire locations were placed as points in a point shape file. To create shape point file, Excel file was converted to debase file (.dbf) and imported into ArcMap. Right Click on the imported table and click “Display XY Data”. Latitude and Longitude information were then used as X and Y to create point shape of fire locations. 

The attribute table of this shape file contains information on:

• Fire ID, which is a unique number for each fire occurred in Texaqs episode

• Acres burnt, which is to total area, in acres unit, burnt in each day for particular fire.

• Latitude and Longitude specifying a location of fire

• Fire date/ start date, representing the day that fire first occurred.

• Heat release (Btu)/day

          Table 1. Some records in an attribute table of point shape file 

Plume characteristics

Fire plume temperature

The plume temperature and other source characteristics are source-specific, and need to be determined for each application. Unfortunately, there is no observed information of plume temperature, for this study; therefore, typical plume temperature range of 900-1600 K is assumed here.

Heat release and Acres burnt

This data is the results from Emission Inventory study as stated above. Each fire event is assumed to last for the entire day of the day the fire reported to occur

Initial vertical velocity

There is a literature, (Chang et al., 1999), suggests using the following calculation :

(7)

where T_p and T_a are the temperatures of plume and ambient [K], respectively. r_p is a radius of fire [m]. Qh is the heat release rate [J/s]

Metrological data 

Meteorological Data can be obtained by extracting this data from CAMx input files and also from MM5 output. These hourly modeling data files and related documents for Texaqs 2000 period can be found on Texas Commission on Environmental Quality (TCEQ) website

- CAMx input file

CAMx is a Eulerian photochemical dispersion model that allows for integrated assessment of gaseous and particulate air pollution over many scales ranging from urban to super-regional. CAMx performs simulation based on the particular horizontal and vertical grid structures, configured to match the grid of Meteorological model that is used to provide environmental input fields.

Metrological data input for CAMx is in Fortran binary format as followings:

Wind File

Hour, idate  Loop from 1 to nlay layers:                  ((uwind(i,j,k), i=1,nx), j=1, ny)                  ((vwind(i,j,k), i=1,nx), j=1, ny)    ((dummy, i=1,nx), j=1, ny)

 

Temperature File

Hour, idate, ((temps(i,j), i=1,nx), j=1, ny) Loop from 1 to nlay layers:                 Hour, idate, ((temps(i,j), i=1,nx), j=1, ny)

For ease of binary file reading, Fortran codes were written to extract 24-hour wind and temperature data out of CAMx input file for particular grid cell (with 14 vertical layers) and to store this data in Microsoft Access database. 

- MM5 output file

MM5 output (The Fifth-Generation NCAR / Penn State Mesoscale Model) is a limited-area, nonhydrostatic or hydrostatic, terrain-following sigma-coordinate model designed to simulate or predict mesoscale and regional-scale atmospheric circulation. It has been developed at Penn State and NCAR as a community mesoscale model and is continuously being improved by contributions from users at several universities and government laboratories.

MM5 has Horizontal grid domain is different from the one used in CAMx as shown below. 

Click here to view MM5 domain

MM5 output reports meteorological data, including Mixing height and Surface heat flux, which are used in Fireplume model. This data has two dimensions. To extract this information in to Microsoft Access database, Existing Fortran code was modified and used for this task. The original Fortran code for extracting MM5 can be found at http://mailman.ucar.edu/pipermail/mm5-users/2002/000073.html. Note that Greenwich time is used in MM5 model; therefore, Visual Basic Programming code was created to revise time information to standard time (Greenwich time – 6 hours)   

Back to TOP

Using ArcMap to allocate location of fire 

To be able to link the meteorological data into the calculation, Arc Map was used to allocate locations of fire in X-Y coordinate according to  spatial domains that is used in CAMx and also in MM5. To do this, grid domains have to be created first. Horizontal grids information of both models are shown below:

Creating Grid Cell

4 km. Grid polygon shape file for these 2 domains ware created. Steps of doing this are as followings:

• Assign 6 digits integer value of x-y coordinate for each cell in Microsoft Excel. There should be totally 63x85 = 5355 cells for CAMx domain and 151x136 = 20536 cells for MM5. Example of cell value is 

  The starting cell, with x-y of 000000, is at the lower left of the domain.

• Save this Excel file as an ASCII text file.

• To recognize ASCII file, Arc GIS has its own heading format for ASCII file. Thus, we need to add heading to ASCII file created in Excel. The format is as shown

• Open Arc Tool Box; choose option “Import to Raster > ASCII to Grid”.  Raster file will be created with 5355 or 20536 values of 6 digits integer as x-y coordinate.

• Define projection system according to CAMx / MM5  horizontal grid structure to this raster file.

• Open Arc Map, use “spatial analysis > convert > raster to feature”. Polygon shape file of grid domain will be created. Attribute table of this grid shows the value of 6 digits integer.

Assemble of data

To choose the fire events for sample calculations, meteorological data from both CAMx and MM5 for those fire events have to be available in hand. Since CAMx domain is smaller than MM5 domain, CAMx grid was then used as a boundary of this study. 4km Grid of CAMx domain was firstly imported into Arc MAP and then followed by fire point shape file. State map and counties shape file were also imported into ArcMAP to aid visualization of locations of fire.

 Sources:

• statesp020.shp:  This polygon shapefile displays the state boundaries of the United States.  It was obtained from the National Atlas, a website maintained by the USGS. 

• countyp020.shp: This polygon shapefile displays the county boundaries of the United States and was also obtained from the National Atlas.

Layout of 4 km-grid cells of CAMx domain is shown below

It turns out that only 3 fire events shown above are considered to be in this scope of study, namely FIRE_ID 1085, 1094 and 1033. The criteria of choosing are that fire event occurred between August 22-September 1,2000, the location was in 4km grid CAMx domain and only big fire ( more than 500 acres) will be considered. From this layout,  identify tool is used to view which grid cell the fire is in. XY coordinate values of 4km grid for each fire location were than added to Forest Fire access database as shown in table 2 (X1-Y1). Wind and Temperature data are then extracted for each XY coordinate.

Table 2. Records in an attribute table of point shape file with XY coordinate 

Similar to meteorological data from CAMx, meteorological data from MM5 was extracted according to XY coordinate of each fire event on MM5 domain. These X-Y coordinates are shown as X2-Y2 in table 2

Back to TOP

Plume Rise Model Programming using Microsoft Visual Basic

Back to TOP

Calculation Results

From Sensitivity runs and model observation, plume temperature does not effect the plume rise calculations. Only heat release (J/s) plays an important role for plume rise height. Although radius of fires does not explicitly give a contribution to plume rise, it is implicitly applied in a value of heat release.  

As the first step of calculation, wind data of layer 1 was used to observe the characteristics of plume rise as shown in figure (a). Horizontal lines represent top height of layer 3, 4 and 5 of CAMx structure. Note that most of the plumes rise above layer 3. Thus, choosing layer 3 for wind velocity data may be a good approximation. All other calculations in this work were then based on wind data of layer 3. In the literature, since there is no suggestion of selecting wind data in plume rise equation, wind velocity of other layers might also be used. But, considering forest fire is a ground source, using wind of higher layer might result in losing the importance of ground wind data.   

Each grid in CAMx is assumed to be well mixed; therefore, plumes that have rise height within the same layer will be treated as having the same height with their value of the top of the layer. Figure (b) shows the plume rise with vertical CAMx structure. All three fire events show similar trend of plume rise with low height at night time and peak during the late afternoon. At night time up to about 10 am, atmospheric stabilities tend to be towards stable class, representing low turbulence in the atmosphere. In the other hand, unstable condition occurs during the day time with the maximum mixing height during late afternoon. This condition leads to a vertical convective transport, resulting in high plume rise in the afternoon. The maximum plume rise height is in range of 800-1300 m.

From figure (b), Fire ID 1033 does not have spike peak at any time of the day but it tends to have high plume rise during many hours in the afternoon. The gradual change in height of plume rise during the afternoon of this fire can be explained from the low wind velocity period in the afternoon on that day as shown in figure (c). Note also that low wind favors vertical displacement rather than horizontal dispersion, resulting in high plume rise.

Figure (c) Plume rise of fire ID 1033 and wind velocity on Aug 30th

Spike peaks occurred on Aug 31^st at 5 pm and on Sep 1^st at 6 pm. According to this, lapse rate calculation was observed. Lapse rates at these two times are 0.008915 and 0.009605 degree/m., respectively. These are the only values that are close to adiabatic lapse rate, which is 0.0098 degree/m., showing the atmospheric stability of stable near neutral condition. Sensitivity runs were performed to investigate the neutral condition calculation. By using neutral mode of calculation for all calculations, plume rise heights are very high compared with stable and unstable modes. Thus, even though, this model compares stable/unstable plume rise results with neutral results and chooses the lower one to prevent the blow off of the calculations for the case of near neutral condition, none of the neutral results was chosen. The high-rise values might be occurred as a consequence of the method of friction velocity estimation.

Back to TOP

Conclusions

CAMx, which is a current widely-used model for air quality assessment in Texas State, can not handle the emission from wild fire, which may provide significant pollutants. From the literature review, Fireplume model was selected as the prospective models for plume rise calculation for the wild fires. By using Arc GIS as a tool to incorporate grid information with emission inventory, also with Fortran and Visual Basic programming, plume rise can be calculated. The results suggest low plume rise at night time and high peak during late afternoon. To integrate the results into CAMx, simplification of the plume rise profile based on observation of this study and results of other models will be purposed. 

Back to TOP

Future Works

• Evaluate other prospective plume rise models , such as CALPUFF models, and compare their results with the result from this study.

• Purpose simple diurnal plume rise profiles and integrate the results into CAMx

• Do sensitivity runs on CAMx to evaluate the proposed plume rise profiles

Back to TOP

References

- Allen T.D., Katamreddy A., Junquera V., Pearson R., 2002, “ Report on Emissions Inventory associated with forest, grassland, and agricultural burning in the Texas Air Quality Study Domain for the period of August 1, 2000-September 30,2000”

- Arya S. P., 1999, “Air pollution meteorology and dispersion”

- Briggs, G. A., 1975, “Plume rise predictions” in Lectures on Air Pollution and Environmental Impact Analysis, D. A. Haugen, ed., American Meteorological Society, Boston (59-111).

- Briggs, G.A, 1969. Plume Rise. USAEC Critical Review Series, TID-25075, National Technical Information Service, Springfield, VA. 81pp.

- Brown D.F.,Dunn W. E., Lazaro M. A.and  Policastro A.J., 1999, Contributed paper from the modeling session from the joint Fire Science Conference and Workshop.

- Dennis A., 2000, Inventory of Air Pollutant Emissions Associated with Forest, Grassland and Agricultural Burning in Texas, MS Thesis, University of Texas at Austin.

- Chang J.C., Scire J.S., 1999, “ User’s Guide to the EPM2BAEM Interface Program”

- ENVIRON, 2000, “User’s Guide Comprehensive Air Quality Model with Extensions (CAMx) version 3.10”

- Scire J.S., Strimaitis D. G., Yamartino R. J., 2000, “ A User’s Guide for the CALPUFF Dispersion Model v. 5”

- Turner, D. B., T. Chico, and J. A. Catalano, 1986, TUPOSCA Multiple Source Gaussian Dispersion Algorithm Using On-Site Turbulence Data. U.S. Environmental Protection Agency,  Research Triangle Park, North Carolina (EPA-600/8-86/010).

-Turner, D.B., T. Chico and J.A.Catalano, 1986, TUPOS-P - A Program for Reducing Hourly and Partial Concentration Files Produced by TUPOS: User's Guide. EPA-600/8-86/012.U.S. Environmental Protection Agency, Research Triangle Park, NC

- Weil. J.C. (1988) Plume Rise, in Lectures on Air Pollution Modeling, A. Venkatram and J.C.Wyngaard (eds.), American Meteorological Society, Boston, 119-157

Back to TOP

Acknowledgements

I would like to thank these people for their gratefulness of helping me on this project

Dr. David Maidment, Center for Research in Water Resources, University of Texas at Austin

Dr. Dave T. Allen, Center of Energy and Environmental Resources, University of Texas at Austin

Dr. Yosuke Kimura, Center of Energy and Environmental Resources, University of Texas at Austin , who gave me a hand on Fortran code

Pithon Vithayasricharoen, Dept. of Civil Eng., who gave me some suggestions on Visual Basic Programming

 '~*~'~~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~'~*~'~~~'~*~'~~'~*~'

Back to TOP

View my project proposal

View my progress report 

Go back to Main