AAAR 31st Annual Conference
October 8-12, 2012
Hyatt Regency Minneapolis
Minneapolis, Minnesota, USA
Abstract View
Verification of Fire Weather Forecasts Using PM2.5 Sensitivity Analysis
SIVARAMAN BALACHANDRAN, Karsten Baumann, Jorge Pachon, James Mulholland, Armistead Russell, Georgia Institute of Technology
Abstract Number: 380 Working Group: Source Apportionment
Abstract Using records from a six year period of prescribed burning (PB) activity from Camp Lejeune, NC and ambient PM2.5 from neighboring Jacksonville, NC, we employed a principal components regression (PCR) technique to verify the validity and relative importance of weather forecast parameters used by land managers in determining optimum conditions to conduct PB. The PCR results help identify sensitivities of local PM2.5 burden to the different meteorological parameters relative to PB.
The approach involves running principal components analysis (PCA) on data sets containing daily average PM2.5 mass concentration, fire activity and two types of meteorological data: i) observed meteorological conditions with A.M.-reported meteorological forecast (PC-AM) and ii) observed meteorological conditions with P.M.-reported meteorological forecast (PC-PM). Both of these data sets contained 635 days of observed and modeled meteorological parameters. The first seven principal component scores explaining over 80% of the PM2.5 variance, were regressed against measured PM2.5 for a subset of days with simultaneous occurrence of PB activity. The importance of lag on sensitivities was examined by regressing principal component scores from day n against PM2.5 from days n (lag 0), n+1 (lag 1), and n+2 (lag 2).
This analysis shows that PB can have significant impact at lag 0 but this impact is not seen at greater lags; at lags 1 and 2, meteorology drives PM2.5 levels with PB smoldering emissions seemingly having negligible effect. For both PCA-AM and PCA-PM, at lag 0, i) PB contributes approximately 3 µg m-3 per 1000 acres burned, and ii) the same NWS forecast parameters characterizing atmospheric stability; e.g. Haines index, ventilation rate, wind speed and mixing height have similar importance in keeping ambient PM2.5 levels low. Thus, foresters can plan PB conduct the day before and still meet the objective of minimal impact on local PM2.5 pollution.