Estimating time to detect time trends in continuous cropping
作者:
Murari Singh,
Michael Jones,
期刊:
Journal of Applied Statistics
(Taylor Available online 1997)
卷期:
Volume 24,
issue 6
页码: 659-670
ISSN:0266-4763
年代: 1997
DOI:10.1080/02664769723404
出版商: Taylor & Francis Group
数据来源: Taylor
摘要:
In long-term field trials comparing different sequences of crops and husbandry practices, the identification and understanding of trends in productivity over time is an important issue of sustainable crop production. This paper presents a statistical technique for the estimation of time trends in yield variables of a seasonal annual crop under continuous cropping. The estimation procedure incorporates the correlation structure, which is assumed to follow first-order autocorrelation in the errors that arise over time on the same plot. Because large differences in annual rainfall have a major effect on crop performance, rainfall has been allowed for in the estimation of the time trends. Expressions for the number of years (time) required to detect statistically significant time trends have been obtained. Illustrations are based on a 7-year data set of grain and straw yields from a trial in northern Syria. Although agronomic interpretation is not intended in this paper, the barley yield data indicated that a significant time trend can apparently be detected even in a suboptimal data set of 7 years' duration.
点击下载:
PDF (155KB)
返 回