论文答辩气象驱动因素对法国地区年际
- 1. Impact of meteorological drivers on regional inter-annual crop yield variability in France 气象驱动因素对法国地区年际 作物产量变异性的影响 Andrej Ceglar*, Agricultural and Forest Meteorology , 2016
- 2. OUTLINE课题背景是指一项课题的由来、意义、环境、状态、前人的研究成果等，以及研究该课题目前所具有的条件等。撰写论文时，在论文的开头一般都要交代课题背景，以便让读者更好地了解课题的内容、研究方法、研究过程和研究成果。释 义背景ABSTRACT INTRODUCTION DATA METHODS RESULTS DISCUSSION CONCLUSIONS
- 3. 1ABSTRACT
- 4. ABSTRACT本文对92个法国地区季节内气候变化对冬小麦和谷物玉米产量年际变化的影响进行评估。将观测到的生长季节的温度、降水和太阳辐射每月的时间序列以及年度作物产量采用基于偏最小二乘回归的统计方法进行分析的。结果显示：主要气象驱动因素对作物产量变异性的贡献和最大影响时间存在显著的空间差异。总体来说，玉米：温度和全球太阳辐射 法国南部，东部和北部作物产量 降雨变率 中部和西北部的产量 冬小麦： 降雨变率 法国北部，西北部和东南部 温度 法国东部
- 5. 2INTRODUCTION
- 6. 主要的气候条件支撑了农业生产食物，饲料，燃料和纤维的适宜性。 季节内气候变化可以：通过温度、可用水、辐射拦截和碳固定直接影响作物生产；通过调节营养物利用度和疾病害虫的发生间接影响(Olesen et al., 2000). 区分确定作物产量背后的驱动因素的方法有两种：机械（动态）作物模型和统计模型（旨在将报告的作物产量变异性与一组解释变量相关联；e.g. Lobell and Burke, 2010)，尽管也已开发了混合方法(e.g. Lobell, 2013). 在过去十年中，随着观测数据可用性和质量的提高（例如,来自遥感和报告统计），表征产量和气象变量之间关系的统计方法的使用也在增加。INTRODUCTION
- 7. INTRODUCTION本文研究的主要目的是确定关键气象变量及其作物生长过程中玉米和冬小麦产量的年际变率的最大影响期随后，我们评估了统计方法的结果是否与主要气象驱动因素的农艺知识和敏感生长阶段的时间安排一致。
- 8. 3DATA
- 9. from October to July for winter wheatDATAFig. S.1: France and its subdivision into départements. Green areas correspond to agricultural areas (Bartholomé and Belward, 2005).Crop yields (1989-2014) AGRESTE Ministère de l’Agriculture Weather data: MARS Crop Yield Forecasting System (MCYFS) database winter wheat： from October to July Grain maize： from April to September
- 10. 4METHODS
- 11. De-trendingIn order to analyse the impact of climate variability on crop yield inter-annual variability, time series must be de-trended（去趋势）. The locally weighted polynomial regression（局部加权 多项式回归） (LOESS; Cleveland, 1979) is here applied to de-trend the crop yield time series.
- 12. Spatial clustering of crop yield time seriesA hierarchical clustering method （层次聚类方法）(Murtagh, 1985) is used to identify spatially homogenous areas in terms of inter-annual crop yield variability. This spatial classification can aid in the interpretation of the dominant climatic drivers and possibly prevalent agro-management techniques.
- 13. Inter-annual crop yield variabilityA Partial Least Squares Regression(PLSR; Garthwaite, 1994; Wold et al., 2001; Rosipal and Kramer, 2005) approach is used to estimate the relationship between meteorological variables and crop yield time series. In this study, the number of explanatory variables amounts to 18 (3 meteorological variables for 6 months of the growing season) and 30 (3 meteorological variables for 10 months of growing season) for grain maize and winter wheat, respectively. PLSR generalizes and combines features from principal component analysis (Jolliffe, 2002) and multiple-regression and can be interpreted as a form of Canonical Correlation Analysis (Rosipal and Kramer, 2005).
- 14. Inter-annual crop yield variabilityThe PLSR model is mainly based on the extraction of a sub-set of latent variables (i.e. inferred, not directly observed variables, to have the best predictive power) from the full set of predictors X Briefly, independent normalized variables X and Y are decomposed as: X= TP T + E P and Q: represent weight matrices（权重矩阵） Y= UQT + F T and U: the latent variable matrices（潜在变量矩阵） E and F: the matrices of residual terms（残差矩阵） Bootstrap（自助法） is used to determine the number of relevant latent variables as well as the importance of the explanatory meteorological variables on the prediction of crop yield anomalies.
- 15. 5RESULTS
- 16. Analysis of de-trended crop yield time series Fig. 1. Box-plots of grain maize (left) and winter wheat (right) yield time series over France for the départements where the inferred PLSR regression model has a prediction skill (see Fig. 6). 平稳
- 17. Analysis of de-trended crop yield time seriesFig. S.2. Inter-annual variability of de-trended crop yields for grain maize (left) and winter wheat (right). White denotes regions with identified inhomogeneities in yield time series. 69 Fig. 2. Homogeneous regions obtained by using hierarchial clustering of de-trended grain maize (left) and winter wheat (right) yield time series.
- 18. Assessment of PLSR regression predictive skillsFig. S.6: Ordinary bootstrap cross-validated mean square error of prediction (MSEPboot) for determining the optimal number of latent variables of derived PLSR models for grain maize (left) and winter wheat (right). Boxplots for each latent variable capture MSEP of all départements.
- 19. The importance of intra-seasonal climate variabilityGrain maizeFig. 3. Standardized regression coefficients of the explanatory meteorological variables for the identified homogeneous regions.南东中中北西北77888
- 20. The importance of intra-seasonal climate variabilityGrain maizeFig. 4. Cumulative importance of temperature, precipitation and global solar radiation for the explained variability of grain maize yields, expressed in relative terms.
- 21. The importance of intra-seasonal climate variabilityWinter wheatFig. 5. As Fig. 3 but for winter wheat.西南东南中西中东西北10 RG12 T4445
- 22. The importance of intra-seasonal climate variability winter wheatFig. 6. As Fig. 4 but for winter wheat
- 23. 6DISCUSSION
- 24. DISCUSSION为了理解并提高模式质量，特别是法国中北部和北部的部分地区（这对冬小麦生产很重要）需要进一步调查。应努力侧重于试图检测和归因农业管理实践变化引起的小麦产量变异性的变化。对于两种作物，在具有极端负观测产量异常的年份，模拟产量往往过高估计观测值。提出的统计模型没有明确考虑极端事件。事实上，极端天气作用于更短的时间尺度，并且有时可能导致作物减产。这项研究的结果有助于提高作物产量预报系统，例如联合研究中心使用的系统(MCYFS, 2015)。还有助于长期（如季节）作物产量预报的开发/改进。分析季节内气候变率对作物产量影响的类似方法也可以扩展到世界其它地区，只需提供足够长的作物产量和解释气象变量时间序列。
- 25. 7CONCLUSIONS
- 26. CONCLUSIONSFig. 7. Identified meteorological variables and their significant influence on inter-annual variability of winter wheat and grain maize yields.玉米：作物产量主要受7月和8月的天气影响。全球辐射和温度是影响法国西南和南部大量灌溉地区年际玉米产量变异性的主要气候变量。法国西南部，最西部和中部地区灌溉较少，对降雨和全球辐射变化更为敏感 冬小麦：生长季节气候变量对冬小麦的重要性在区域上比玉米变化更大，更分散。温度对法国西南部和东部的冬小麦产量有实质性影响，而降雨对法国北部和南部地区尤为重要
- 27. THANKS!
该用户的其他文档
相关PPT