Journals
  Publication Years
  Keywords
Search within results Open Search
Please wait a minute...
For Selected: Toggle Thumbnails
Grey correlation analysis of drought and yield at different growth stages of rice in Sichuan Province
DENG Guowei, SUN Jun, LAI Jiang, ZHANG Ling
Journal of Arid Meteorology    2022, 40 (5): 814-822.   DOI: 10.11755/j.issn.1006-7639(2022)-05-0814
Abstract281)   HTML5)    PDF(pc) (11897KB)(923)       Save

In order to overcome the shortage of disaster condition records and the difficulty of identification to drought effect at each growth stage, based on daily meteorological data at weather stations, growth period data at agro-meteorological stations and rice yield data in counties of Sichuan Province from 1981 to 2015, with improved water budget index as drought indicator, the relationship between drought and yield at each growth stage of rice was analyzed by using grey correlation analysis method. The results show that the drought frequency at growth stages of rice in Sichuan from high to low was transplanting-tillering stage, tillering-jointing stage, booting-heading stage, heading-maturity stage and jointing-booting stage in turn. The spatial characteristic of rice drought frequency at each growth stage was low consistent with the spatial distribution of grey correlation degree between water budget index and rice yield. The high frequency of drought occurred in the middle or northeast of Sichuan Basin, while the high grey correlation degree located in the mountainous areas around the basin edge and Panxi area. With the process of rice growth, the effect of drought on rice yield weakened, and at transplanting-tillering and tillering-jointing stage it was the most obvious. The growth period with the greatest influence of drought on rice yield was different in each county of different planting regions, so the policies of drought alleviation should be strengthened according to the spatial characteristics of drought influence at each growth stage of rice. We should focus on the drought at tillering-jointing stage of rice in planting regions of the Chengdu Plain and the central hill of Sichuan Basin and transplanting-tillering stage in other five rice planting regions.

Table and Figures | Reference | Related Articles | Metrics
Precipitation Forecast of Wudongde Hydropower Station Based on SVM Model Optimized by Multiple Algorithms
SUN Junkui, WANG Jiang, KANG Daojun, YAN Liping, ZHOU Xi
Journal of Arid Meteorology    2019, 37 (4): 670-675.  
Abstract244)      PDF(pc) (723KB)(1412)       Save
基于粒子群优化(PSO)算法和遗传算法(GA)对支持向量机(SVM)的核函数及主要参数进行训练优化,分别建立PSO算法、GA的支持向量机模型(PSO_SVM、GA_SVM)。选用ECMWF及T639数值预报产品资料和乌东德水电站降水资料,普查最优预报因子,构建包含各种类型降水过程的训练样本和测试样本。比较分析SVM模型RBF和Sigmoid核函数优劣。尝试先分段寻找局部最优,再选择全局最优的参数优化方法。通过增大训练样本集、降低交叉验证准确率、迭代次数截断和控制惩罚系数范围的方法,提高模型的稳定性和泛化能力,防止过拟合和收敛缓慢现象。利用测试样本对SVM、PSO_SVM和GA_SVM三种方案进行对比检验,优化的GA_SVM预报效果较好且稳定。经2018年试报表明,GA_SVM逐3 h累计降水量预报TS评分在50%以上,漏报率在15%以下,与ECMWF和T639比较,该模型TS评分提高1.4%。

Related Articles | Metrics