Evaluation of a multiple linear regression model for the prediction of panicle number in rice

Yusaku Yamauchi, Hirai Yasumaru, Keisuke Saruta, Takeo Yamakawa, Eiji Inoue, Takashi Okayasu, Mitsuoka Muneshi

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In Japan, rice production technology must be developed to meet stable yield and quality targets. This objective is important because of the demand for high quality rice by consumers, the necessity of enhancing Japan's international competitive power, and a declining trend in the production of first grade rice. To achieve stable yield and quality, it is important to control panicle number appropriately since it is a primary factor to determine rice yield and quality. In this study, a multiple linear regression model was built to predict panicle number, which can support the decision-making process of farmers in agricultural practice. Surveys of 33 paddy fields were conducted in the Fukuoka Prefecture in 2010-2011. Data regarding influential factors on panicle number, such as solar radiation, water temperature, and exchangeable ammonium (eNHJ content in the soil, were collected and used for building the model as candidates of the explanatory variable. As the results, eNH, content in the soil at the beginning of tillering stage and accumulated amount of daily solar radiation during the tillering stage were selected as explanatory variables. The adjusted coefficient of determination was 0.448 and RMSECV was 2.49. In the prediction model, eNH, content in the soil at the beginning of the tillering stage was selected as an explanatory variable to represent inorganic nitrogen supply. In comparison, this variable does not reflect the inorganic nitrogen supplied throughout the tillering stage. This difference resulted in a large error in the field with organic fertilizer application, as the manner of inorganic nitrogen release is different to that of typical fields in the survey area that are subject-ed to chemical fertilizer application.

Original languageEnglish
Pages (from-to)421-426
Number of pages6
JournalJournal of the Faculty of Agriculture, Kyushu University
Volume57
Issue number2
Publication statusPublished - Sep 1 2012

All Science Journal Classification (ASJC) codes

  • Biotechnology
  • Agronomy and Crop Science

Fingerprint Dive into the research topics of 'Evaluation of a multiple linear regression model for the prediction of panicle number in rice'. Together they form a unique fingerprint.

  • Cite this