采用双模态联合表征学习方法识别作物病害

基于深度卷积神经网络的视觉识别方法在病害诊断中表现出色,逐渐成为了研究热点。但是,基于深度卷积神经网络建立的视觉识别模型通常只利用了图像模态的数据,导致模型的识别准确率和鲁棒性,都依赖训练数据集的规模和标注的质量。构建开放环境下大规模的病害数据集并完成高质量的标注,通常需要付出巨大的经济和技术代价,限制了基于深度卷积神经网络的视觉识别方法在实际应用中的推广。该研究提出了一种基于图像与文本双模态联合表征学习的开放环境下作物病害识别模型(bimodalNet)。该模型在病害图像模态的基础上,进行了病害文本模态信息的嵌入,利用两种模态病害信息间的相关性和互补性,实现了病害特征的联合表征学习。最终bimodalNet在较小的数据集上取得了优于单纯的图像模态模型和文本模态模型的效果,最优模型组合在测试集的准确率、精确率、灵敏度、特异性和F1值分别为99.47%、98.51%、98.61%、99.68%和98.51%。该研究证明了利用病害图像和病害文本的双模态表征学习是解决开放环境下作物病害识别的有效方法。     

浙江遂昌毛竹鞭笋丰产培育技术

为了发挥毛竹林生产潜力，增加竹农经济收入，在多年实践研究的基础上总结出一套毛竹鞭笋丰产培育技术。研究认为：在毛竹笋材两用林中采挖鞭笋，不会影响整体的竹鞭鞭径和竹鞭总量生长；毛竹林地覆盖稻草和砻糠能显著提高第3年的鞭笋产量；培育高产鞭笋的最佳施肥时间为每年的5月；鞭笋采收时间为5-10月。毛竹鞭笋丰产培育技术可在毛竹笋材两用林经营中推广应用。     

北京市丰台区10kV配电线路故障分析

首先分析了北京市丰台区的电力负荷特点及设备情况,然后利用丰台区2019年的配电线路电力故障数据,分别以故障地点和故障原因两方面进行归类和分析,提出相应的防护措施,为提升丰台配电网的可靠性和保障配电网的安全运行提供了依据。     

Poorly crystalline iron and aluminium oxides contribute to the carbon saturation and sorption of dissolved organic carbon in the soil

Soil carbon (C) saturation implies an upper limit to a soil's capacity to store C depending on the contents of silt + clay and poorly crystalline Fe and Al oxides. We hypothesized that the poorly crystalline Fe and Al oxides in silt + clay fraction increased the C saturation and thus reduced the capacity of the soil to sorb additional C input. To test the hypothesis, we studied the sorption of dissolved organic carbon (DOC) on silt + clay fractions (<53 µm) of highly weathered oxic soils, collected from three different land uses (i.e., improved pasture, cropping and forest). Soils with high carbon saturation desorbed 38% more C than soils with low C saturation upon addition of DOC, whereas adsorption of DOC was only observed at higher concentration (>15 g kg^?1). While high Al oxide concentration significantly increased both the saturation and desorption of DOC, the high Fe oxide concentration significantly increased the desorption of DOC, supporting the proposition that both oxides have influence on the DOC sorption in soil. Our findings provide a new insight into the chemical control of stabilization and destabilization of DOC in soil.     

云南不同气候带农业气候资源变化特征研究

为分析全省平均以及各气候带光能、热量、水分资源的时空变化特征,使用云南115站及6个气候带代表站1961—2018年的气候要素计算各地农业气候资源统计量。结果表明,光能资源变化以减少趋势为主,出现显著突变,各气候带太阳辐射变化可能会引起云南太阳辐射高低值中心发生变化,2009年以来辐射明显增加可能会导致太阳辐射出现新的变化趋势和突变点;热量资源一致显著增加,喜凉及喜温作物的活动积温、积温持续时间及无霜期长度均显著增加,并且呈现初日提前、终日推后的趋势,对作物生长有利;水分资源总体呈现减少趋势,尤其21世纪以来下降趋势明显,亚热带地区暖干化现象突出,干旱风险等级较高。     

干旱处理对不同玉米品种叶片光合速率和抗氧化特性及产量的影响

玉米是对干旱比较敏感的粮食作物,不同品种应对干旱胁迫的能力存在差异。以6个普通玉米品种为材料,在北京市郊区大田条件下,采用裂区设计,以灌溉处理为对照,分析干旱处理对不同玉米品种叶片光合速率（P_n）和抗氧化特性及产量的影响。结果表明,与对照相比,干旱处理下不同品种玉米P_n降低幅度为京农科728（15.63%）相似文献   

Effect of farm management on topsoil organic carbon and aggregate stability in water: A case study from Southwest England,UK

There are few reliable data sets to inspire confidence in policymakers that soil organic carbon (SOC) can be measured on farms. We worked with farmers in the Tamar Valley region of southwest England to select sampling sites under similar conditions (soil type, aspect and slope) and management types. Topsoils (2–15 cm) were sampled in autumn 2015, and percentage soil organic matter (%SOM) was determined by loss on ignition and used to calculate %SOC. We also used the stability of macroaggregates in cold water (WSA) (‘soil slaking’) as a measure of ‘soil health’ and investigated its relationship with SOC in the clay‐rich soils. %SOM was significantly different between management types in the order woodland (11.1%) = permanent pasture (9.5%) > ley‐arable rotation (7.7%) = arable (7.3%). This related directly to SOC stocks that were larger in fields under permanent pasture and woodland compared with those under arable or ley‐arable rotation whether corrected for clay content (F = 8.500, p < .0001) or not (F = 8.516, p < .0001). WSA scores were strongly correlated with SOC content whether corrected for clay content (SOC_adj R² = .571, p < .0001) or not (SOC_unadj R² = 0.490, p = .002). Time since tillage controlled SOC stocks and WSA scores, accounting for 75.5% and 51.3% of the total variation, respectively. We conclude that (1) SOC can be reliably measured in farmed soils using accepted protocols and related to land management and (2) WSA scores can be rapidly measured in clay soils and related to SOC stocks and soil management.     

基于网络药理学分析黄芪防控猪繁殖与呼吸综合征的物质基础及靶点信息

为揭示黄芪防控猪繁殖与呼吸综合征(PRRS)的物质基础及靶点信息,采用网络药理学方法,从中药系统药理学数据库与分析平台(TCMSP)筛选黄芪的潜在活性成分、作用靶点,同时从比较毒理基因组学数据库(CTD)收集PRRS相关宿主基因,最后构建"活性成分-关键靶点"网络、蛋白互作网络,并进行GO功能和KEGG通路富集分析。结果显示,从黄芪筛选出的17个潜在活性成分中,有14个潜在活性成分共计65个基因靶点与PRRS相关宿主基因交叉,其中槲皮素、山奈酚、刺芒柄花素、异鼠李素等潜在活性成分可能对治疗PRRS有重要作用,TNF、IL8、TP53、IL6、IL10、IL1B、JUN等45个蛋白靶点间存在相互作用。此外,黄芪治疗PRRS的作用机制可能涉及细胞外间隙和胞浆等组分,并可能与炎症反应、细胞凋亡、T细胞受体信号通路等生物过程相关。本研究为临床应用黄芪及其提取物治疗PRRS奠定了理论基础。     

青冈栎林分空间结构对林下更新的影响

林分空间结构会显著影响其林下幼树更新。为明确其影响林下幼树更新的关键因子,以湖南省青羊湖国有林场青冈栎林天然次生林为研究对象,测量幼树株数密度、树高、地径和冠幅,计算混交度、大小比数、开敞度指数以及聚集指数代表林分空间结构,评价林下幼树更新情况,并用灰色关联度分析方法探究两者间的相关关系。结果表明,青羊湖国有林场13块青冈栎林样地林下更新幼树有6种,分属于4科5属,分别是青冈栎、苦槠、多脉青冈、南酸枣、黄檀、马尾松,青冈栎在更新幼树中占明显优势,是优势树种,幼树重要值为63.09%。青冈栎林各林分空间结构指数与林下更新指数均大于0.5,关联度较高,综合所有林分空间结构指数对林下更新指数关联度的均值来看,对林下更新的影响由大到小为大小比数>开敞度>聚集指数>混交度,大小比数对青冈栎林下更新影响最为显著。在促进青冈栎天然次生林林下更新时,应以调整林木大小分化程度为主,综合考虑林木空间分布格局的调整方案。