Despite policy attempts to limit or prevent agricultural burning, its use to remove crop residues either immediately after harvest (e.g. field burning of wheat stubble) or after subsequent crop processing (e.g. "bonfires" of rice straw and rapeseed residues) appears to remain widespread across parts of China. Emission factors for these types of small but highly numerous fire are therefore required to fully assess their impact on atmospheric composition and air pollution. Here we describe the design and deployment of a new smoke measurement system for the close-range sampling of key gases and particles within smoke from crop residue fires, using it to assess instantaneous mixing ratios of CO and CO2 and mass concentrations of black carbon (BC) and PM2.5 from wheat stubble, rice straw, and rapeseed residue fires. Using data of our new smoke sampling system, we find a strong linear correlation between the PM2.5 mass and BC, with very high PM2.5 to BC emission ratios found in the smouldering phase (up to 80.7 mg m(-3).(mg m(-3))(-1)) compared to the flaming phase (2.0 mg m(-3).(mg m(-3))(-1)). We conclude that the contribution of BC to PM2.5 mass was as high as 50% in the flaming phase of some burns, whilst during smouldering it sometimes decreased to little over one percent. A linear mixing model is used to quantify the relative contribution of each combustion phase to the overall measured smoke composition, and we find that flaming combustion dominated the total emission of most species assessed. Using time series of trace gas concentrations from different fire cases, we calculated 'fire integrated' trace gas emission factors (EFs) for wheat, rice and rapeseed residue burns as 1739 +/- 19 g kg(-1),1761 +/- 30 g kg(-1) and 1704 +/- 27 g kg(-1) respectively for CO2, and 60 +/- 12 g kg(-1), 47 +/- 19 g kg(-1) and 82 +/- 17 g kg(-1) respectively for CO. Where comparisons were possible, our EFs agreed well with those derived via a simultaneously-deployed open path Fourier transform infrared (OP-FTIR) spectrometer. These EFs, and the linear best fit relationships between both PM2.5 and BC mass and the CO2 and CO measurements, were used to generate particulate EFs, which varied over the 5.8 -20.3 g kg(-1) and 0.25-2.89 g kg(-1) range respectively. We note a particularly high 2.89 g kg(-1) BC emission factor for the rapeseed bonfires, reflective of intense flaming combustion that gave off visible clouds of soot. These field-measured EFs offer a different perspective than is obtained when burning in laboratory combustion chambers, and are suitable for combining with landscape-scale fuel consumption estimates to provide atmospheric modelling inputs of emissions from these types of crop residue fires. (C) 2015 The Authors. Published by Elsevier Ltd.
