贵德县温室大棚内温度变化规律及预报模型研究

Study on Temperature Change Rule in Greenhouse in Guide County Forecast Model

以青海省贵德县日光温室大棚为研究对象，收集2014年9月至2015年9月大棚内外气温的连续观测资料，采用不同的天气条件分析温室大棚内部温度日变化和季节变化特征。分析室内外气温最高、最低的相关关系。结果表明，大棚内气温与棚外气温呈现显著的相关关系，温室大棚的气温是随着外界气象条件和季节的变化而相应变化的，并具有显著日变化规律和季节变化规律。四季。中，晴天、多云、阴天条件下温度均存在单波峰变化情况，日光温室内温度的增高受天气条件影响较大，晴天的增温效果明显高于多云和阴天。在此基础上，采用逐步回归法构建大棚内最高及最低气温预报模型，并对预报模型进行实况检验，结果表明，建立的预报模型对温室内最低气温的预报误差2℃以下能达到83％～98％，最高气温的预报误差2℃以下在65％～76％。按天况分型，晴天预测效果最好，多云次之，阴天由于样本值少模型效果相对较差。

Taking solar greenhouse in Guide County of Qinghai Province as the research object, the variation characteristics of greenhouse temperature diurnal and seasonal with different weather condition were analyzed, by continuous observation data collected from September 2014 to September 2015 in the temperature inside and outside, and the correlation of the highest temperature and the lowest temperature indoor and outdoor were analyzed. The results showed that the temperature in the greenhouse and temperature outside the greenhouse showed significant correlation. The greenhouse temperature changed with the external weather conditions and seasonal changes, and had significant variation of diurnal variation and seasonal. In the four seasons, sunny, cloudy and cloudy conditions, there was a single peak change of temperature. The increase of the temperature in the solar greenhouse was affected by the weather conditions, and the effect of increasing temperature was significantly higher than that of cloudy and cloudy. On this basis, the construction of greenhouses in the high- est and lowest temperature forecast mod- el of stepwise regression method, and the live test of the prediction model results showed that the prediction errors of the model of the indoor temperature minimum temperature below 2 degrees could reach 83%-98%, the prediction error of maximum temperature below 2 degrees between 65%-76%. According to the weather condition classification, the weather forecast effect was the best, cloudy and less cloudy was the second, cloudy day, due to the small sample value of the model is relatively poor.