不同季节强碱土土壤呼吸影响因子分析与模型预测

Impact factor analysis and model prediction of strong alkaline soil respiration in different seasons

利用LI-8100土壤碳通量测量仪测定了春夏秋三季晴朗天气下强碱土土壤呼吸速率、温度(气温和地温)、湿度(空气相对湿度和土壤湿度)数据,分析了它们之间的相关关系,获得不同季节对土壤呼吸影响较大的因子,并建立不同类别的多种回归模型;在精度检验及简单易行原则基础上,得到各季节土壤呼吸预测的最优模型。结果表明:(1)虽然温湿度均是影响不同季节强碱土土壤呼吸的主要因素,但均以温度的影响较大,其中气温是春秋两季土壤呼吸的最大直接影响因子,地温是夏季的最大直接影响因子,而土壤湿度为各季节最大的间接影响因子。(2)春秋季土壤呼吸的最佳预测模型均为10 cm处气温和土壤湿度所建的双因子方程,该方程具有较小的均方根误差(RMSE)(分别为0.159和0.259),且相对分析误差(RPD)>2(分别为2.9、2.094),具有非常好的预测能力。夏季土壤呼吸最佳预测模型则为包含10 cm处气温、地温、空气相对湿度和土壤湿度所建的4因子方程,RMSE为0.248,RPD>2(为2.406),可用于精确预测。(3)各季节土壤呼吸变化趋势与其影响因子的变化,因春季的完全同步,夏季基本一致,而秋季一致性较差,故春季土壤呼吸最佳预测模型的预测精度最高(92.67%),夏季次之(84.99%),秋季较差(77.23%)。

In this paper,strong alkaline soil respiration rate has been measured in spring,summer,autumn under sunny weather condition by using LI-8100 soil carbon flux meter,temperature（ air temperature and soil temperature）,humidity（ relative humidity and soil moisture） were also recorded simultaneously,and the correlation among above factors has been analyzed to identify the key factors influencing soil respiration in different season,and establish a variety of regression models. On the basis of the accuracy test and simple principle,the optimal model to predict the seasonal soil respiration has been selected. The results show that：（ 1） Although the temperature and humidity are the main factors affecting strong alkaline soil respiration in different seasons,but the temperature has relatively larger effect,where the air temperature is the biggest direct impact factor on soil respiration in spring and autumn,while in summer,the soil temperature is the greatest direct impact factor,and soil moisture is the largest indirect factor in all season.（ 2） The best prediction model of soil respiration in spring and autumn is the two-factor equation composed of air temperatures at 10 cm layer and soil moisture,which has a smaller RMSE（ 0. 159 and 0. 259,respectively）,and the RPD 〉2（ 2. 9,2. 094,respectively）,has a better predictive accuracy.The best soil respiration predictive model for summer season is composed of air temperature at 10 cm,surface temperature,relative humidity and soil moisture,RMSE is 0. 248,RPD 〉2（ of 2. 406）.（ 3） The trend of soil respiration variation and their impact factors was similar in spring,but different in summer and autumn. In spring the soil respiration model has the highest prediction accuracy（ 92. 67%）,followed by summer（ 84. 99%） and poor in fall（ 77. 23%）.