Objective Because of statistical noise in Monte Carlo dose calculations,the effective point doses may not be accurately calculated.A user-defined sphere volume was adopted to substitute the effective point to take sphere sampling around the effective point,which minimize the random errors and improve the accuracy of statistical dose.Methods Direct dose measurements were performed at 0°and 90° using a 0.125 cm3 Semiflex ionization chamber (IC) 31010 isocentrically placed in the center of a homogeneous Cylindric sliced RW3 phantom (PTW,Germany).In the scanned CT phantom series,the sensitive volume length of the IC (6.5 mm) was delineated and the isocenter was defined as the simulated effective point.All beams were simulated in the treatment planning system (TPS) in accordance to the measured model.The grid spacing was calculated by 2 mm voxels and the relative standard deviation should be ≤ 0.5％.The statistical and measured doses were statistically compared among three IC models with different electron densities (ED;esophageal lumen ED =0.210 g/cm3 for model A,air ED =0.001 g/cm3 for model B and the default CT scanned ED for model C) at different sampling sphere radius (2.5,2.0,1.5 and 1.0 mm) to evaluate the effect of Monte Carlo.calculation uncertainty upon the dose accuracy.Results In the Monaco TPS,the statistical value was in the highest accordance with the measured value with an absolute average deviation of 0.49％ when the IC was set as esophageal lumen ED =0.210 g/cm3 and the sampling sphere radius was 1.5 mm.When the IC was set as air ED=0.001 g/cm3 and default CT scanned ED,and,the recommended statistical sampling sphere radius was 2.5 mm,the absolute average deviations were 0.61％ and 0.70％.Conclusion In the Monaco TPS,the calculation model with an ED of 0.210 g/cm3 and a sampling radius of 1.5 mm is recommended for the ionization chamber 31010 to substitute the effective point dose measurement to decrease the random stochastic errors of Monte Carlo.

Objective To construct a rapid prediction system to improve the accuracy and efficiency of evaluation of the consequences of nuclear accidents at a field scale. Methods Base on a diagnostic wind field model and Lagrangian particle diffusion, we established a rapid prediction method for wind field and pollutant dispersion around complex underlying surfaces within a field scale, in a way of visual discrimination of buildings and vegetation distribution. With data simulation and the use of a real urban field example, the simulated results were compared with wind tunnel test measurements and computational fluid dynamics results to study the influence of complex underlying surfaces on wind field and pollutant transport in the region. Results The rapid prediction system could clearly simulate the high-resolution wind field and pollutant concentration distribution of the region in about five minutes. It could interface with geographic information software and couple with a mesoscale weather prediction model. In terms of accuracy, the system performed well in wind field simulation, with the fractional deviations of wind speed and wind direction being 0.33 and −0.08, respectively. Concentration field simulation was greatly affected by the wind field, and the ratios of simulated concentrations to observed concentrations were between 0.05 and 3.4, except for a few low concentration points. Conclusion The rapid prediction system can effectively simulate the distribution characteristics of the flow field and improve calculation efficiency when ensuring calculation accuracy, which provides an important reference for emergency response to nuclear accidents.