摘要
为了提高设施农业气象服务水平,促进中国设施农业发展,通过对日光温室内外气象条件对比观测,分别采用BP神经网络、逐步回归和能量平衡原理构建了温室内气温预报模型,探讨能适用于业务服务的小气候预报技术方法。结果表明,基于BP神经网络的预报模型虽预报精度高,由于种植作物生长特性差异较大,缺乏服务的广适性;通过能量平衡原理构建的预报模型机理性强,但相关参数难以获得,预报精度差,服务时效短;采用逐步回归方法构建的温室温度预报模型较前两者具有比较优势,且预报时效可以为未来的1～7天,比较适合于目前气象部门开展设施农业气象服务。
In order to improve the service level and promote the protected agricultural development,it was critical for establish microclimate forecast models through observation inside and outside meteorological elements,using of BP nerve network,stepwise regression and energy balance theory.Results showed that：forecasting model based on BP nerve network was of high accuracy,but it was not appropriate for application.The forecasting model set up on energy balance theory was low accuracy and long computing time.Compared with the two above models,greenhouse temperature forecasting model adopting stepwise regression method was of comprehensive advantage and is properly applied to routine operation service.
出处
《中国农学通报》
CSCD
2012年第29期195-202,共8页
Chinese Agricultural Science Bulletin
基金
公益性行业(气象)科研专项"设施农业气象灾害预警及防御关键技术"[GYHY(QX)201006028]
公益性行业(气象)科研专项"设施农业及特色农产品气象保障关键技术研究"[GYHY(QX)2010200906023]
关键词
日光温室
小气候
模型
预报
sunlight greenhouse
microclimate
model
forecast
