基于冠层归一化植被指数预测双氰胺配施后的芝麻产量  

Sesame yield forecast at different dicyandiamide rates based on canopy normalized difference vegetation index

作　　者：柳开楼[1] 胡惠文[1] 胡志华[1] 李大明[1] 周利军[1] 黄庆海[1] 余喜初[1] 叶会财[1] 徐小林[1] 王赛莲[1] 

机构地区：[1]江西省红壤研究所/国家红壤改良工程技术研究中心/农业部江西耕地保育科学观测实验站,江西南昌330046

出　　处：《中国油料作物学报》2016年第4期467-472,共6页Chinese Journal of Oil Crop Sciences

基　　金：江西现代农业科研协同创新专项(JXXTCX2015003-005);江西省科技支撑计划(20141BBF60050);公益性行业(农业)科研专项(201203030)

摘　　要：为在合适的生育期利用芝麻冠层的归一化植被指数（NDVI）预测芝麻产量,调查相同氮肥条件下不同双氰胺（DCD）施用量对关键生育期（苗期、初花期、盛花期和终花期）的芝麻冠层NDVI影响和产量变化,探讨不同生育期冠层NDVI与相同氮肥下DCD施用量之间的关系,构建NDVI预测产量的模型。结果表明：芝麻各生育期NDVI均表现出随DCD用量的增加而先增后降的趋势,苗期和盛花期以D3（占氮肥3%的DCD）、D5（占氮肥5%的DCD）和D7（占氮肥7%的DCD）较高,比D0（不添加DCD）增加6.2%-24.8%;终花期则以D5、D7和D10（占氮肥10%的DCD）较高,比D0增加了47.2%-103.1%。在添加DCD条件下,不同生育期芝麻冠层NDVI值与产量的拟合方程不一。结合方差分析发现,盛花期和终花期的冠层NDVI值与芝麻产量间为极显著正相关（p〈0.01）,对数方程y=ln（0.258 7＋6.968 7x）和y=ln（0.894 1＋3.998 8x）可以用来预测芝麻产量。验证试验发现,根据终花期冠层NDVI计算出的产量预测值和实测值之间的差异更小（RMSE〈10%,RRMSE〈25%）。因此,采用终花期冠层NDVI建立的模型对芝麻产量具有较好的预测性和普适性。To better forecast sesame yield,field experiment was carried out using different dicyandiamide（ DCD） rates. DCD amounts were marked as different percentages of DCD to urea. They were D0（ 0）,D3（ 3%）,D5（ 5%）,D7（ 7%）,D10（ 10%）,D15（ 15%） and D20（ 20%）. Trends of normalized difference vegetation index（ NDVI） at 4 key growth stages [seedling,and 3 flowering stages（ initial,full and late） ]were investigated. Relationship between yield and NDVI at each stages was established and validated. Result showed that with DCD increasing,NDVI values were improved. But when DCD was more than 3%- 10%,NDVI began to reduce. NDVI values at D3,D5 and D7 were higher in seedling and full flowering stages,and they were 6. 2%-24. 8% more than D0. During late flowering,the NDVI values of D5,D7 and D10 were the highest,and they were increased than D0 by 47. 2%- 103. 1%. In the condition of equivalent N fertilizers,the fitting equations between NDVI value and yield were not similar at different stages. However,it was found that the fitting equations at full and late flowering stages [y = ln（ 0. 258 7 ＋ 6. 968 7x） and y = ln（ 0. 894 1 ＋ 3. 998 8x） ] were significantly different（ p〈 0. 01）,which indicated that both equations could be used to predict sesame yield. In validation experiment,the error between the predicted and measure values in late flowering was small（ R2= 0. 611 5,RMSE 10% and RRMSE 25%）. Therefore,equation of late flowering was feasible to estimate sesame yield of differentamount of DCD based on NDVI（ DCD）.

关 键 词：芝麻 冠层归一化植被指数 产量预测 双氰胺(DCD) 

分 类 号：S156.6[农业科学—土壤学]