基于卷积神经网络与迁移学习的油茶病害图像识别

传统的植物病害图像识别准确率严重依赖于耗时费力的人工特征设计。该文利用深度卷积神经网络强大的特征学习和特征表达能力来自动学习油茶病害特征,并借助迁移学习方法将Alex Net模型在Image Net图像数据集上学习得到的知识迁移到油茶病害识别任务。对油茶叶片图像进行阈值分割、旋转对齐、尺度缩放等预处理后,按照病害特征由人工分为藻斑病、软腐病、煤污病、黄化病和健康叶5个类别。每个类别各选取750幅图像组成样本集,从样本集中随机选择80%的样本用作训练集,剩余20%用作测试集。利用随机裁剪、旋转变换和透视变换对训练集进行数据扩充,以模拟图像采集的不同视角和减少网络模型的过拟合。在Tensor Flow深度学习框架下,基于数据扩充前后的样本集,对Alex Net进行全新学习和迁移学习。试验结果表明,迁移学习能够明显提高模型的收敛速度和分类性能;数据扩充有助于增加数据的多样性,避免出现过拟合现象;在迁移学习和数据扩充方式下的分类准确率高达96.53%,对藻斑病、软腐病、煤污病、黄化病、健康叶5类病害的F1得分分别达到94.28%、94.67%、97.31%、98.34%和98.03%。该方法具有较高的识别准确率,对平移、旋转具有较强的鲁棒性,可为植物叶片病害智能诊断提供参考。     

黑龙港流域参考作物蒸散量的时序变化和分形特征

利用位于黑龙港流域的武强、深泽、饶阳、晋州、献县5站1957-2009年日最高气温、日平均气温、日最低气温、日平均相对湿度、日平均风速、日照时数资料,采用Penman-Monteith算法,计算各站不同时间尺度ET0,采用线性趋势分析法分析其趋势倾向,并应用滑动R/S分析方法研究该流域不同时间尺度ET0时间序列的分形特征。线性趋势分析显示,各站历史上自1957年以来的ET0年总值的气候倾向率在-33.81-10.79mm.10a^-1,即均呈下降趋势,但变化倾向率不同;各月ET0倾向率在-11.27-2.02mm.10a^-1,大多数为负值,其中5、6月份各站的下降趋势最大;春、夏、秋、冬季ET0倾向率为-15.87-1.30mm.10a^-1,且夏季各站之间的差异较大,尤以饶阳站与其他4站间的差异最大。气候要素倾向率的对比分析表明,5站参考作物蒸散量总体下降趋势的基本特征主要是由于风速下降、日照时数减少、日最高最低气温上升造成的;5、6月份下降趋势明显则主要是由于风速下降、日照时数减少、日平均气温上升减缓、相对湿度上升明显造成的;饶阳站与其他4站差异较大的主要原因在于2、3、11、12月风速下降和日照时数减少以及日最高最低气温上升趋势减缓、相对湿度和气压下降明显,以致这些月份的ET0倾向率大于0,形成秋季和冬季ET0倾向率大于0,年尺度ET0年际变化下降趋势不明显。R/S分析结果显示,5站全年和各季ET0时间序列的Hurst指数均大于0.5,相关系数均在0.98以上,分维数均小于1.5,说明各站全年和各季节ET0时间序列变化趋势在未来一段时间内具有持续性,即ET0在未来将呈较明显的下降趋势,这种趋势在除饶阳外的其他站月均有不同程度的表现。     

坡度和雨强对重庆市黄壤旱坡地产流产沙特征的影响

黄壤作为重庆市山地主要的土壤类型之一,对重庆市农业生产有重要影响。通过人工模拟降雨试验,研究不同坡度和降雨强度条件对黄壤旱坡地产流产沙特征影响及响应过程,结果表明：（1）输沙率随降雨强度的增大而增大,对于短历时强降雨来说,输沙率保持在一个相对平均的水平,比长历时中强度降雨的输沙率大。在坡度0°～25°以内,黄壤旱坡地产流量与坡度之间的关系可以用多相式进行拟合,得到产流临界坡度为22.83°;（2）土壤侵蚀量与坡度符合二次抛物线关系,随坡度的增加出现临界坡度,重庆市黄壤旱坡地土壤侵蚀的临界值坡度为24.7°。在降雨特征参数中,平均降雨强度是反映黄壤旱坡地泥沙量、径流系数和径流泥沙含量的最好特征参数,时段最大降雨强度I10最能反映黄壤旱坡地径流量的变化特征;（3）在10°,15°,20°3个不同坡度下,径流量和产沙量与雨强均表现为幂函数关系,相关系数均在大于0.05置信区间呈显著相关性。     

Considering sink strength to model crop production under elevated atmospheric CO2

Climatic changes and elevated atmospheric CO₂ concentrations will affect crop growth and production in the near future. Rising CO₂ concentration is a novel environmental aspect that should be considered when projections for future agricultural productivity are made. In addition to a reducing effect on stomatal conductance and crop transpiration, elevated CO₂ concentration can stimulate crop production. The magnitude of this stimulatory effect (‘CO₂ fertilization’) is subject of discussion. In this study, different calculation procedures of the generic crop model AquaCrop based on a foregoing theoretical framework and a meta-analysis of field responses, respectively, were evaluated against experimental data of free air CO₂ enrichment (FACE) environments. A flexible response of the water productivity parameter of the model to CO₂ concentration was introduced as the best option to consider crop sink strength and responsiveness to CO₂. By varying the response factor, differences in crop sink capacity and trends in breeding and management, which alter crop responsiveness, can be addressed. Projections of maize (Zea mays L.) and potato (Solanum tuberosum L.) production reflecting the differences in responsiveness were simulated for future time horizons when elevated CO₂ concentrations and climatic changes are expected. Variation in future yield potential associated with sink strength could be as high as 27% of the total production. Thus, taking into account crop sink strength and variation in responsiveness is equally relevant to considering climatic changes and elevated CO₂ concentration when assessing future crop production. Indicative values representing the crop responsiveness to elevated CO₂ concentration were proposed for all crops currently available in the database of AquaCrop as a first step in reducing part of the uncertainty involved in modeling future agricultural production.     

河北清苑县及周边农田土壤及农作物中重金属污染状况与分析评价

在河北省地质调查院1：250000多目标区域地球化学调查的研究基础上,对河北省清苑县及周边的农田土壤及玉米子实进行了采样分析,分析该地区农田土壤重金属的富集程度,并采用富集因子法和单因子土壤污染指数法进行污染状况评价。结果表明,该研究区土壤以重金属Zn、Cu、Cr、Pb、Cd污染最为严重,其中Cd污染已达显著富集程度;通过对子实重金属含量的描述统计显示,其农产品也受到一定程度的重金属污染,其中玉米子实中的重金属Ni和Pb存在超标现象。通过地统计学方法分析揭示了清苑县土壤重金属中污染严重的Zn、Cu、Cr、Pb的空间分布特征,并探索其主要成因。     

洱海流域北部农田养分平衡及残留特性研究

为减少农田氮磷流失,改善洱海水质、遏制富营养化发展,以洱海流域北部8个乡（镇）为研究区域,采用现场调查与分析测试相结合的方法研究了洱海流域北部10种典型轮作模式的施肥量、作物养分吸收量、农田养分盈余量以及土壤养分残留状况,对不同轮作模式农田养分平衡及残留特性进行了分析评价。结果表明,水稻-大蒜轮作模式下有机肥和氮肥的投入量显著高于其他模式,水稻-蔬菜作物轮作模式的磷肥投入量显著高于其他模式;不同轮作模式间土壤养分残留差异性以硝态氮最为突出：水稻-大蒜轮作条件下土壤硝态氮残留量达43.4mg.kg-1,明显高于其他轮作模式;10种轮作模式均处于养分盈余状态,以水稻-大蒜轮作模式养分盈余量最大,为1258.8kgN.hm-2和1472.7kgP2O5.hm-2;养分盈余量差异主要凸现在小春季作物上;土壤中易流失的硝态氮、铵态氮和速效磷与氮磷肥的投入量和养分盈余均呈显著相关。表明水稻-大蒜和水稻-蔬菜轮作是洱海流域农田环境污染风险较高的种植模式,需要重点防控。     

Effects of biochar compared to organic and inorganic fertilizers on soil quality and plant growth in a greenhouse experiment

Our contemporary society is struggling with soil degradation due to overuse and climate change. Pre‐Columbian people left behind sustainably fertile soils rich in organic matter and nutrients well known as terra preta (de Indio) by adding charred residues (biochar) together with organic and inorganic wastes such as excrements and household garbage being a model for sustainable agriculture today. This is the reason why new studies on biochar effects on ecosystem services rapidly emerge. Beneficial effects of biochar amendment on plant growth, soil nutrient content, and C storage were repeatedly observed although a number of negative effects were reported, too. In addition, there is no consensus on benefits of biochar when combined with fertilizers. Therefore, the objective of this study was to test whether biochar effects on soil quality and plant growth could be improved by addition of mineral and organic fertilizers. For this purpose, two growth periods of oat (Avena sativa L.) were studied under tropical conditions (26°C and 2600 mm annual rainfall) on an infertile sandy soil in the greenhouse in fivefold replication. Treatments comprised control (only water), mineral fertilizer (111.5 kg N ha^–1, 111.5 kg P ha^–1, and 82.9 kg K ha^–1), compost (5% by weight), biochar (5% by weight), and combinations of biochar (5% by weight) plus mineral fertilizer (111.5 kg N ha^–1, 111.5 kg P ha^–1, and 82.9 kg K ha^–1), and biochar (2.5% by weight) plus compost (2.5% by weight). Pure compost application showed highest yield during the two growth periods, followed by the biochar + compost mixture. biochar addition to mineral fertilizer significantly increased plant growth compared to mineral fertilizer alone. During the second growth period, plant yields were significantly smaller compared to the first growth period. biochar and compost additions significantly increased total organic C content during the two growth periods. Cation‐exchange capacity (CEC) could not be increased upon biochar addition while base saturation (BS) was significantly increased due to ash addition with biochar. On the other hand, compost addition significantly increased CEC. Biochar addition significantly increased soil pH but pH value was generally lower during the second growth period probably due to leaching of base cations. Biochar addition did not reduce ammonium, nitrate, and phosphate leaching during the experiment but it reduced nitrification. The overall plant growth and soil fertility decreased in the order compost > biochar + compost > mineral fertilizer + biochar > mineral fertilizer > control. Further experiments should optimize biochar–organic fertilizer systems.     

基于作物生育期灾害敏感指数的旱灾粮食减产量研究

我国是自然灾害频繁发生的国家,自然灾害造成不同程度的粮食减产,是粮食安全的潜在风险源。该文以旱灾为主,将全国各地区作物生育期、不同旱灾强度、发生时间及发生次数进行耦合分析,研究其耦合规律及主要粮食作物在各生育期对灾害的敏感性,并据此计算我国不同地区旱灾减产量并进行风险评价分析。研究表明,冬小麦的返青、乳熟和成熟期,春小麦的分蘖、乳熟和成熟期,玉米的三叶、七叶和成熟期以及水稻的返青、孕穗、乳熟和成熟期均是防灾减灾的关键期,应该加强农田管理,减少灾害损失。旱灾对我国粮食主产区影响较大,粮食减产风险防范关键区主要分布在黄淮海区、长江中下游区。     

秸秆还田对宁南旱作农田土壤水分及作物生产力的影响

在宁南旱区通过研究秸秆还田对土壤水分及作物生产力的影响,为该区土壤扩蓄增容及作物水分利用效率的提高提供理论依据。在3年秸秆还田定位试验中,设置了不同秸秆还田量处理（谷子秸秆按3000、6000、9000kg·hm-2粉碎还田;玉米秸秆按4500、9000、13500kg·hm-2粉碎还田,对照为秸秆不还田）,对不同处理条件下的土壤含水量、作物水分利用效率和作物产量等指标进行了分析。结果表明,随秸秆还田量由高到低,在试验第3年（2009年）玉米播种期0～200cm土层的土壤贮水量分别较CK提高8.8%、9.9%和6.8%;成熟期0～200cm土层的土壤贮水量分别较CK提高14.8%、13.9%和12.8%;产量分别较CK显著提高30.7%、29.2%和12.5%（P〈0.05）;作物水分利用效率分别较CK显著提高41.1%、35.9%和21.3%（P〈0.01）。在宁南半干旱区采用秸秆还田能较好地保蓄土壤水分,利于土壤水库的扩蓄增容,且对提高作物产量和作物水分利用效率有显著效果。     

Relation between soil organic matter and yield levels of nonlegume crops in organic and conventional farming systems

The aim of this study was to evaluate the interaction between yield levels of nonleguminous crops and soil organic matter (SOM) under the specific conditions of organic and conventional farming, respectively, and to identify implications for SOM management in arable farming considering the farming system (organic vs. conventional). For that purpose, correlations between yield levels of nonlegume crops and actual SOM level (C_org, N_t, C_hwe, N_hwe) as well as SOM‐level development were examined including primary data from selected treatments of seven long‐term field experiments in Germany and Switzerland. Yield levels of nonlegume crops were positively correlated with SOM levels, but the correlation was significant only under conditions of organic farming, and not with conventional farming treatments. While absolute SOM levels had a positive impact on yield levels of nonlegumes, the yield levels of nonlegumes and SOM‐level development over time correlated negatively. Due to an increased demand of N from SOM mineralization, higher yield levels of nonlegumes obviously indicate an increased demand for OM supply to maintain SOM levels. Since this observation is highly significant for farming without mineral‐N fertilization but not for farming with such fertilization, we conclude that the demand of SOM‐level maintenance or enhancement and thus adequate SOM management is highly relevant for crop production in organic farming both from an agronomical and ecological point of view. Under conventional management, the agronomic relevance of SOM with regard to nutrient supply is much lower than under organic management. However, it has to be considered that we excluded other possible benefits of SOM in our survey that may be highly relevant for conventional farming as well.