Estimation of ecotype-specific cultivar parameters in a wheat phenology model and uncertainty analysis
Estimation of ecotype-specific cultivar parameters in a wheat phenology model and uncertainty analysis
复制标题
小麦物候模型中特定生态型品种参数的估计和不确定性分析
DOI：
10.1016/j.agrformet.2016.02.016
复制
发表时间：
2016-05
影响因子：
6.2
通讯作者：
Zhu, Yan
中科院分区：
文献类型：
--
作者：
Tang, Liang;Liu, Leilei;Cao, Weixing;Zhu, Yan
本研究旨在为小麦作物模拟模型的大规模应用提供一种有效的生态类型特异性品种参数预测方法。基于现有的模型，探讨了马尔可夫链蒙特卡罗（MCMC）技术的参数校准。利用1980 ~ 1995年小麦历史物候期和逐日气象资料，估计了淮安、郑州、潍坊和石家庄4个生态点的生态类型特异性品种参数的后验概率分布。从1996 - 2005年的后验概率分布中随机抽取1000组品种参数，对MCMC方法进行评价。从每个生态点的1000组参数中选择最优的50组参数，代表江苏、河南、山东和河北省的生态类型特异性品种参数，以在区域尺度上评估2005年基于MCMC的方法。结果表明，在立地尺度上，观测物候期与估算物候期的决定系数（R2）在0.6 1 ~ 0.72之间，均方根误差（RMSE）小于3.6 d，均方根偏差（RMSD <$）小于3.7 d。在四个生态点，使用后验概率分布获得的三个物候期的RMSE和RMSD <$值均显著低于基于先验概率分布的值。在区域尺度上，2005年观测物候期与估算物候期的R2大于0.86，RMSE小于3.4d，RMSD <$小于4.0d。结果表明，MCMC方法对小麦物候期多参数组合的估计具有较高的可靠性。将MCMC技术与物候模型相结合，可用于估算中国小麦主产区的生态类型特异性品种参数，并可用于区域尺度上的小麦发育进程预测。
The objective of this study was to develop an effective method for predicting ecotype-specific cultivar parameters for the large-scale application of wheat crop simulation models. The Markov Chain Monte Carlo (MCMC) technique was explored for parameter calibration based on an existing model. We estimated the posterior probability distribution of ecotype-specific cultivar parameters at four ecosites (Huai'an, Zhengzhou, Weifang, and Shijiazhuang in China) by using the historical phenological stages of wheat and daily weather data from 1980 to 1995. 1000 sets of cultivar parameters which were randomly sampled from the posterior probability distribution at each ecosite from 1996 to 2005 were used to evaluate the MCMC-based method. Optimal 50 sets of parameters were chosen from the 1000 sets of parameters at each ecosite to represent the ecotype-specific cultivar parameters of the Jiangsu, Henan, Shandong and Hebei provinces to evaluate the MCMC-based method for the year 2005 at the regional scale. The results showed that the coefficients of determination (R 2) between the observed and estimated phenological stages ranged from 0.61 to 0.72, with a root mean square error (RMSE) of less than 3.6 days and a root mean square deviation (RMSD¯) of less than 3.7 days at the site scale. All of the RMSE and RMSD¯ values for the three phenological stages obtained using the posterior probability distribution at the four ecosites were significantly lower than those based on the prior probability distribution. At the regional scale, R 2 between the observed and estimated phenological stages was greater than 0.86, with RMSE less than 3.4 days and RMSD¯ less than 4.0 days in most of the grids for year 2005. The estimated phenological stages agreed well with the observations, suggesting that the MCMC technique has high reliability of for estimating multiple parameter combinations in a wheat phenology model. The combination of the present MCMC technique and a phenology model could be used for estimating the ecotype-specific cultivar parameters for the main wheat growing regions of China, which can be used to predict progress of wheat development at the regional scale.
登录
查看更多内容
DOI：
10.1016/j.asoc.2008.02.002
发表时间：
2009
期刊：
Appl. Soft Comput.
影响因子：
--
作者：
Chunni Dai;Meng Yao;Zhujie Xie;Chunhong Chen;Jingao Liu
通讯作者：
Chunni Dai;Meng Yao;Zhujie Xie;Chunhong Chen;Jingao Liu
DOI：
10.1137/120891344
发表时间：
2013-10
期刊：
SIAM/ASA J. Uncertain. Quantification
影响因子：
--
作者：
F. Minunno;M. Oijen;D. Cameron;J. S. Pereira
通讯作者：
F. Minunno;M. Oijen;D. Cameron;J. S. Pereira
DOI：
--
发表时间：
2000
期刊：
--
影响因子：
--
作者：
Yang MeiChun;Cao Weixing;Li Cundong;Wang Zhaolong
通讯作者：
Yang MeiChun;Cao Weixing;Li Cundong;Wang Zhaolong
影响因子：
1.9
作者：
M. Herndl;J. White;S. Graeff;W. Claupein
通讯作者：
M. Herndl;J. White;S. Graeff;W. Claupein
影响因子：
6.6
作者：
Jagtap, SS;Jones, JW
通讯作者：
Jones, JW