Abstract: Based on the observed data of sunlight greenhouse microclimate and nearby weather station during winter in 2012, the methods of correlation analysis and stepwise regression analysis are adopted to set up forecast models of microclimate in sunlight greenhouse under different weather types during winter, and a fitting test is carried on the forecast model of microclimate in sunlight greenhouse. Based on the research work above, the conclusions are as followsWhen the degree of correlation between the sunlight greenhouse microclimate and the outside weather conditions are different, the factors affecting the inside microclimate are also different. The methods of stepwise regression analysis and correlation analysis are adopted to set up forecast model of microclimate in sunlight greenhouse under different weather types during winter, and a fitting test is carried out. The fitting average absolute error of maximum daily temperature in sunlight greenhouse is 1.5~2.8 ℃;the fitting relative error of minimum daily temperature in sunlight greenhouse is 0.5~1.3 ℃; when the error range of the two is within ±2 ℃,the forecast effect is better in the forecast range. When the fitting average absolute error of minimum daily relative humidity in sunlight greenhouse is 1.2%~3.7%, the forecast effect is better; When the fitting average absolute error of daily minimum global radiation in sunlight greenhouse is 31.5~39.8 W/m²,the fitting average relative error of daily minimum global radiation in sunlight greenhouse is bigger, and the forecast effect is poor.

Key words: Panjin, sunlight greenhouse, microclimate, weather conditions, model