春秋季对近红外光谱模型预测奶牛场粪水氮磷含量结果的影响

为研究季节因素对规模化奶牛场粪水氮磷含量及其近红外光谱模型预测结果的影响,该研究采集了天津市春秋双季27家规模化奶牛场粪水处理全过程的250个粪水样品,解析了季节对粪水氮磷含量分布特征的影响,同时采集了所有样品的近红外光谱并进行主成分分析。采用偏最小二乘法(Partial Least Squares,PLS)建立了粪水氮磷季节内预测模型,包括春秋单季和双季融合模型以及季节间的相互预测模型。结果表明,粪水氮磷含量随季节变化呈现出差异性,季节内模型总体的预测效果较好,优于季节间模型;其中春季模型表现最佳,验证相关系数分别为0.98和0.90,剩余预测偏差(Residual Predictive Deviation,RPD)分别为4.67和2.03。研究表明,季节因素对粪水中氮磷含量的模型预测结果存在不同程度的影响,该研究可为建立全季节要素的综合模型提供依据。

Spatial modeling of the Ulmus pumila growing season in China's temperate zone

To reveal the ecological mechanism of spatial patterns of plant phenology and spatial sensitivity of plant phenology responses to climate change,we used Ulmus pumila leaf unfolding and leaf fall data at 46 stations of China's temperate zone during the period 1986-2005 to simulate 20-year mean and yearly spatial patterns of the beginning and end dates of the Ulmus pumila growing season by establishing air temperature-based spatial phenology models,and validate these models by extensive spatial extrapolation.Results show that the spatial patterns of 20-year mean and yearly February-April or September-November temperatures control the spatial patterns of 20-year mean and yearly beginning or end dates of the growing season.Spatial series of mean beginning dates shows a significantly negative correlation with spatial series of mean February-April temperatures at the 46 stations.The mean spring spatial phenology model explained 90% of beginning date variance(p<0.001) with a Root Mean Square Error(RMSE) of 4.7 days.In contrast,spatial series of mean end dates displays a significantly positive correlation with spatial series of mean September-November temperatures at the 46 stations.The mean autumn spatial phenology model explained 79% of end date variance(p<0.001) with a RMSE of 6 days.Similarly,spatial series of yearly beginning dates correlates negatively with spatial series of yearly February-April temperatures and the explained variances of yearly spring spatial phenology models to beginning date are between 72%-87%(p<0.001),whereas spatial series of yearly end dates correlates positively with spatial series of yearly September-November temperatures and the explained variances of yearly autumn spatial phenology models to end date are between 48%-76%(p<0.001).The overall RMSEs of yearly models in simulating beginning and end dates at all modeling stations are 7.3 days and 9 days,respectively.The spatial prediction accuracies of growing season's beginning and end dates based on both 20-year mean and yearly models are close to the spatial simulation accuracies of these models,indicating that the models have a strong spatial extrapolation capability.Further analysis displays that the negative spatial response rate of growing season's beginning date to air temperature was larger in warmer years with higher regional mean February-April temperatures than in colder years with lower regional mean February-April temperatures.This finding implies that climate warming in winter and spring may enhance sensitivity of the spatial response of growing season's beginning date to air temperature.