Yield prediction in spring barley from spectral reflectance and weather data using machine learning

Accurate preharvest yield estimation is an important issue for agricultural planning purposes and precision farming. Machine learning (ML) based on readily obtained information on the cropping system, typically including spectral reflectance measurements, is an essential approach for achieving practical solutions. We tested in a 9-year soil compaction experiment the accuracy of ML-based yield predictions made up to 2 months before harvest from a Ratio Vegetation Index (RVI) and recordings of precipitation and reference evapotranspiration. The applied data set comprises 224 combinations of plots and years with measured grain yields in the range of 4.22–9.34 Mg/ha. The best ML model [i.e., with the smallest mean absolute error (MAE)] was selected automatically by the AutoML interface included in the R program package H₂O. Its cross-validated predictions made on June 30 more than 1 month before harvest showed an MAE of 0.38 Mg/ha when trained on all data from all years except the one under consideration. MAE increased to about 0.68 Mg/ha when determined 3 weeks earlier on June 10. MAE values in the range of 0.32–0.42 Mg/ha were obtained for predictions made on June 30 when based on data from at least six consecutive years; however, MAE showed no generally decreasing trend with the number of years. Yield estimations were robust towards a considerable soil variation observed within the experimental area due in part to the experimental treatments. The results show a potential of making yield predictions in barley 1–2 months before harvest, which, however, is not sufficiently early to support decisions on top-dress N fertilization.     

Study on forecasting models of grasshopper occurrence area in Qinghai Province

利用青海省1995-2006年蝗虫发生面积与相关大气环流特征量、气候资料,分析了不同气候因子与蝗虫发生面积之间的关系,建立了青海省蝗虫发生面积省、州、县级多元回归预测模型。研究表明,显著影响青海省蝗虫发生面积的环流指数特征量为副高强度指数和极涡面积、极涡强度指数。副高强度指数强,青海省蝗虫发生面积大;极涡面积大、极涡强度指数强,蝗虫发生面积小。在蝗虫孵化期,降水量的分布和数量对其发生面积产生显著的促进或抑制作用。温度对蝗虫发生面积的影响各地基本一致,对蝗虫产卵、越冬、孵化均呈显著正相关,表明温度偏低是制约蝗虫爆发的主导因素。蝗虫产卵、孵化期空气湿度低有利于蝗灾的爆发。用预测模型对2007年蝗虫发生面积进行逐级试报,效果较好,这为青海省蝗灾的防治工作提供了科学的决策依据。     

Applying ecological niche modelling to plan conservation actions for the Red-spectacled Amazon (Amazona pretrei)

The Red-spectacled Amazon (Amazona pretrei) is an endangered, small range and migratory parrot threatened by habitat destruction and pet trade. We modelled its geographical distribution with the objective of predicting its historical, current and future (in face of climatic changes) geographical distributions. We also tested the assumption that the species tracks similar ecological conditions while seasonally migrating. Finally, we estimated the overlap between its predicted distributions and the current protected areas network. We used eight modelling techniques in an ensemble-forecasting approach to reach consensus distributional scenarios. The modelled historical breeding and wintering ranges covered a larger area than today. Its current year-round distribution is located in the same region as before but decreased by 13%, whereas the breeding range decreased by 32%, and the wintering range by 43%. The future year-round distribution is predicted to decrease by 47% and shift to east by 2060. Even more drastic decreases in range are predicted for breeding (63%) and wintering (91%) ranges. The niches of the breeding and wintering distributions are good predictors of the records of the other season. The greatest variation (SDs) in the predictions comes from modelling tools and GCMs, and little from climatic scenarios. However, variation in predictions were small within high probability regions for modelling tools by higher everywhere for GCMs and climatic scenarios. The Red-spectacled Amazon is currently underrepresented (<1% of year-round range, and <4% of each seasonal range) in the current reserve system. A broad view over all estimated and predicted ranges revealed that a small portion of the range has a very important role in current and future conservation actions for this bird - in the northeast part of state of Rio Grande do Sul and southeast state of Santa Catarina, Brazil.     

基于近红外光谱和支持向量机的土壤参数预测

应用支持向量机算法对实时土壤光谱数据进行处理,获得了土壤全氮和有机质的回归模型并研究了模型随参数变化的规律。从中国农业大学试验田采集了150个土样,用光谱仪获取了原始土壤样本的近红外光谱,用实验室分析法获取了各样本的全氮和有机质含量。以近红外光谱数据为自变量对2个土壤参数进行了回归建模并评价了算法各参数对模型的影响。研究表明土壤参数适合于全谱支持向量回归。对于土壤全氮,基于小波降噪NIR光谱的SVM回归模型的标定R2为0.9224,验证R2为0.3667;对于土壤有机质,基于原始NIR光谱的SVM回归模型     

基于改进持续法的短期风电功率预测

为了有效减轻风能波动对电网的影响,提高风电在电力市场中的竞争力,风电功率预测研究具有重要意义。该文提出了基于小波变换的改进持续法,对短期风电功率预测进行研究。该方法首先利用小波变换将原始风速信号分解为高频部分和低频部分,针对高频信号相邻的两个数据之间相似度较低,波动较大的特点,采用滑动平均法进行预测,而低频信号仍然采用持续法预测,最后通过小波重构以及风电功率特性曲线转换得到风电功率预测值。与原持续法相比较,平均相对误差由17.10%降至11.81%,平均绝对误差由39.58 kW降至23.48 kW,有效地提高了短期风电功率预测的精度,具有一定的实际应用价值。     

基于神经网络的水产品价格预测模型研究

以水产品中鲫鱼为例,选择时间、地理环境和经济条件因素作为输入层变量,价格作为输出单元,输入样本进行训练和仿真,对训练好的网络输入预测样本,将预测结果与市场实际价格进行比较,其相对误差均小于1%。结果证明,所构建的水产品价格预测模型具有良好的精确性和准确性,将神经网络应用于水产品价格预测是可行的。     

Catch forecasting and relationship between water temperature and catch fluctuations in snow crab <Emphasis Type="Italic">Chionoecetes opilio</Emphasis> in the western Sea of Japan

Catch forecasting and the relationship between water temperature and catch in the snow crab Chionoecetes opilio in the western Sea of Japan were investigated. Catch was used as an index of abundance on the basis of high correlations between catch per unit effort for the period when the latter data were available. The pattern of fluctuations, in catches and index coincided well with each other. Therefore, catch data were regarded as an index of abundance and the correlation coefficient between catch and water temperature was calculated at several depths from 1964 to 1999. Catch forecasting models were composed using significantly correlated variables with the following results: (i) in April and September, the catch showed high positive correlation with water temperatures (depth 50, 100, and 200 m, time lag 4–6 years); and (ii) a model using water temperatures in April alone (depth 100 m, time lag 4–7 years) forecasted the catches with a coefficient of determination of 0.504, where models using more variables (water temperatures in the 2 months and catches) showed a coefficient of 0.587 at most. Environmental conditions during the early life stages of the snow crab are thought to deeply influence the fluctuations, in abundance.     

养殖锯缘青蟹黄水病流行病学及其预报模型

依据浙江三门县青蟹养殖基地青蟹黄水病发生及流行的定点观测资料和环境因子检测数据,对青蟹发病与养殖环境因子的关系进行了分析研究,找出3个显著影响青蟹黄水病发生与流行的环境因子,即盐度变化、降雨量和弧菌数量.在此基础上,用多元统计方法对黄水病的预报模型进行了研究,建立了三门县青蟹养殖基地的青蟹黄水病预报模型,(Y)=-0.088+0.00026TV+7.179SC.检验表明,该模型具较高拟合度,可以用来测报黄水病发病率.     

组合模型分析方法在我国粮食产量预测中的应用

尝试将组合预测法应用于我国未来粮食产量的预测,以提高预测精度。通过赋予合理权重,将C-D生产函数模型、多元回归模型和指数平滑模型加权组合。对各模型进行平均绝对百分误差(MAPE)、希尔不等系数(Theil IC)和均方根误差(RMSE)等指标的比较,证明单一模型经过组合能够提高预测精度。     

大中型拖拉机需求量混沌特征分析及预测时效研究

中国大中型拖拉机需求量是一个复杂的非线性系统。该文在分析大中型拖拉机需求特点的基础上,建立了基于混沌理论的农业装备需求特性分析及预测时效模型。采用中国1952至2004年大中型拖拉机需求量的实测数据,重点研究了其需求系统的相空间重构、关联维数的确定和柯尔莫哥洛夫熵的计算。结果表明：中国大中型拖拉机需求为混沌系统,影响需求量的因素有1～12个,最大可预测的时效为8a。