基于BP神经网络的巢湖市冬季浅层地温预报
|
摘要
利用2011—2016年冬季(12月至翌年2月)巢湖市国家基本站逐日气温、地温、日照、降水等气象资料,采用相关分析法筛选出影响地温的关键气象因子,运用MATLAB软件构建了基于BP神经网络浅层最低地温预报模型,并比较不同层模拟精度。结果表明,0~20 cm地温日变化均呈正弦曲线变化,越向深层地温变化幅度越小,位相逐层滞后。相关性分析表明,浅层最低地温与前一日的平均气温、最低气温、0~20 cm各层平均地温和最低地温成显著正相关,与前一日日照时数成显著负相关。模型模拟结果显示0、5、10、15、20 cm最低地温预报的标准误差和绝对误差逐层减小,20 cm层预报准确度明显优于0 cm层。
Based on the observation data of daily air temperature,ground temperature,sunshine duration and precipitation from Chaohu observation station in winter( from December to February) from 2011 to 2016, the main meteorological factors affecting ground temperature were selected by correlation analysis method,shallow layer minimum ground temperature prediction model was set up based on BP neural network by using MATLAB software,and the simulation precision of different layers was compared in this paper. Results showed that the diurnal variation of 0-20 cm ground temperature was presented sine curve,as the depth increased,the ground temperature fluctuant range decreased,and the phase was delayed.Correlation analysis showed that the shallow layer minimum ground temperature had a significant positive correlation with the average temperature,minimum temperature,average ground temperature and the lowest ground temperature of 0-20 cm of the previous day,and was negatively correlated with the sunshine duration of the previous day. The simulation results showed that the standard error and absolute error of 0,5,10,15,20 cm minimum ground temperature prediction decreased layer-by-layer,and the prediction accuracy of 20 cm layer was obviously better than that of 0 cm layer.
Based on the observation data of daily air temperature,ground temperature,sunshine duration and precipitation from Chaohu observation station in winter( from December to February) from 2011 to 2016, the main meteorological factors affecting ground temperature were selected by correlation analysis method,shallow layer minimum ground temperature prediction model was set up based on BP neural network by using MATLAB software,and the simulation precision of different layers was compared in this paper. Results showed that the diurnal variation of 0-20 cm ground temperature was presented sine curve,as the depth increased,the ground temperature fluctuant range decreased,and the phase was delayed.Correlation analysis showed that the shallow layer minimum ground temperature had a significant positive correlation with the average temperature,minimum temperature,average ground temperature and the lowest ground temperature of 0-20 cm of the previous day,and was negatively correlated with the sunshine duration of the previous day. The simulation results showed that the standard error and absolute error of 0,5,10,15,20 cm minimum ground temperature prediction decreased layer-by-layer,and the prediction accuracy of 20 cm layer was obviously better than that of 0 cm layer.
引文
[1]中国气象局.地面气象观测规范[M].北京:气象出版社,2003:85-89.
[2]贺欢,田长彦,王林霞.不同覆盖方式对新疆棉田土壤温度和水分的影响[J].干旱区研究,2009,6(6):826-831.
[3]米丽娜,王芳,李友宏,等.宁夏平原灌淤土土壤养分与作物产量的空间变异性[J].干旱区研究,2009,26(4):508-513.
[4]孙小妹,张涛,陈年来,等.土壤水分和氮素对春小麦叶片抗氧化系统的影响[J].干旱区研究,2011,28(2):206-214.
[5]WALKER J M.One-degree increment in soil temperature affects maizeseedling behavior[J].American Society Soil Science,1969,33:729-736.
[6]姜会飞,廖树华,叶尔克江,等.地面温度与气温关系的统计分析[J].中国农业气象,2004,25(3):1-4.
[7]杨丹,薛晓萍,李楠,等.日光温室地温预报技术研究[J].中国蔬菜,2013(20):54-60.
[8]李帅,王萍,陈莉,等.黑龙江省春季浅层(0~20 cm)地温变化特征及预报[J].冰川冻土,2014,36(1):55-62.
[9]罗喜平,周明飞,汪超.贵州省冬季地表(0 cm)温度预报探讨[J].贵州气象,2016,40(4):1-5.
[10]金志凤,符国槐,黄海静,等.基于BP神经网络的杨梅大棚内气温预测模型研究[J].中国农业气象,2011,32(3):362-367.
[11]杨再强,黄川容,费玉娟,等.基于BP神经网络的温室番茄气孔导度的模拟研究[J].东北农业大学学报,2011,42(11):70-77.
[12]倪玉红,孙擎,王学林,等.盱眙龙虾池塘夏季水温与溶解氧变化特征及预报模型研究[J].中国农学通报,2015,31(32):33-39.
[2]贺欢,田长彦,王林霞.不同覆盖方式对新疆棉田土壤温度和水分的影响[J].干旱区研究,2009,6(6):826-831.
[3]米丽娜,王芳,李友宏,等.宁夏平原灌淤土土壤养分与作物产量的空间变异性[J].干旱区研究,2009,26(4):508-513.
[4]孙小妹,张涛,陈年来,等.土壤水分和氮素对春小麦叶片抗氧化系统的影响[J].干旱区研究,2011,28(2):206-214.
[5]WALKER J M.One-degree increment in soil temperature affects maizeseedling behavior[J].American Society Soil Science,1969,33:729-736.
[6]姜会飞,廖树华,叶尔克江,等.地面温度与气温关系的统计分析[J].中国农业气象,2004,25(3):1-4.
[7]杨丹,薛晓萍,李楠,等.日光温室地温预报技术研究[J].中国蔬菜,2013(20):54-60.
[8]李帅,王萍,陈莉,等.黑龙江省春季浅层(0~20 cm)地温变化特征及预报[J].冰川冻土,2014,36(1):55-62.
[9]罗喜平,周明飞,汪超.贵州省冬季地表(0 cm)温度预报探讨[J].贵州气象,2016,40(4):1-5.
[10]金志凤,符国槐,黄海静,等.基于BP神经网络的杨梅大棚内气温预测模型研究[J].中国农业气象,2011,32(3):362-367.
[11]杨再强,黄川容,费玉娟,等.基于BP神经网络的温室番茄气孔导度的模拟研究[J].东北农业大学学报,2011,42(11):70-77.
[12]倪玉红,孙擎,王学林,等.盱眙龙虾池塘夏季水温与溶解氧变化特征及预报模型研究[J].中国农学通报,2015,31(32):33-39.