Potential Predictability of Local Paddy Rice Yield Variation Using a Crop Model with Local Areal Information

• Iizumi Toshichika

  Agro-Meteorology Division, National Institute for Agro-Environmental Sciences

• Ishida Kenji

  Department of Rural Planning, National Institute for Rural Engineering

• Yokozawa Masayuki

  Agro-Meteorology Division, National Institute for Agro-Environmental Sciences

• Nishimori Motoki

  Agro-Meteorology Division, National Institute for Agro-Environmental Sciences

抄録

This study examined the potential predictability of paddy rice yield variation on a local scale (approximately 2 km × 2 km) by using a prefectural-scale dynamic paddy rice simulation model (the PRYSBI model) with the local observed weather data, taking the 160 local areas in Tochigi Prefecture, Japan, as the study area. From the comparison of the simulated and observed local yields during the 11-year period (1993, 1995, 1998-2006), the PRYSBI model showed high capability to simulate the interannual variation of area-mean local yield over the study area with the quite large root-mean-square error (RMSE) of 2.1 Mg ha^-1. However, the RMSE has the statistical-significant relationship with the local areal features on agriculture. We thereby incorporated the local areal features in the simulated yields in the manner of the multiplicative model approach. The parameter values of the multiplicative model were estimated by using the Bayesian approach, which can count the uncertainty of parameter value in a stochastic manner. By this means, the potential predictability in terms of the coefficient of determination (r²) and RMSE between the simulated and observed local yields improved from r²=0.430 to r²=0.527 and from RMSE=2.0 Mg ha^-1 to RMSE=0.4 Mg ha^-1 compared to the PRYSBI model alone. The potential predictability of local yield could improve by incorporating the local areal features in the output of crop model.<br>

収録刊行物

• 農業情報研究

  農業情報研究 19 (2), 36-42, 2010

  農業情報学会

キーワード

• Bayesian approach
• crop model
• crop yield variation
• local areal feature
• prediction

詳細情報

• CRID: 1390001204461378176
• NII論文ID: 130000299896
• DOI: 10.3173/air.19.36
• ISSN: 18815219
• ISSN: 09169482
• 本文言語コード: en
• データソース種別:
  • JaLC
  • Crossref
  • CiNii Articles
• 抄録ライセンスフラグ: 使用不可