Influencing Factors of Seed Long-distance Dispersal on a Fragmented Forest Landscape on Changbai Mountains,China

Seed long-distance dispersal(LDD) events are typically rare, but are important in the population processes that determine large-scale forest changes and the persistence of species in fragmented landscapes. However, previous studies focused on species dispersed via animal-mediated LDD, and ignored those dispersed by wind. The aim of this study was to assess the effects of canopy openness, edge, seed source, and patch tree density on the LDD of seeds by wind in forest. We collected birch seeds, a typical wind-dispersed species, throughout a larch plantation. We then assessed the relationship between birch LDD and each factor that may influence LDD of seeds by wind including distance to edge, canopy openness size, distance to mature forest, and the tree density of the larch plantation. We used univariate linear regression analysis to assess the influence of those factors on birch LDD, and partial correlations to calculate the contribution of each factor to LDD. The results showed that both canopy openness and edge had significant influences on birch LDD. Specifically, a negative relationship was observed between distance to edge and birch LDD, whereas there was a positive correlation between canopy openness size and LDD. In contrast, the distance to the mature forest was not correlated with birch LDD. Our results suggest that patch tree density could potently affect the probability of LDD by wind vectors, which provides novel and revealing insights regarding the effect of fragmentation on wind dynamics. The data also provide compelling evidence for the previously undocumented effect of habitat fragmentation on wind-dispersed organisms. As such, these observations will facilitate reasonable conservation planning, which requires a detailed understanding of the mechanisms by which patch properties hamper the delivery of seeds of wind-dispersed plants to fragmented areas.

The Role of “Long Forest” Belief of the Dai Nationality in the Conservation of Forest Resources: A Case Study of Manjingchengzi Village in Xishuangbanna

The core of"Long Forest"belief of the Dai nationality is to protect natural forests and forests for conservation of water supply as the home of ancestral gods of the nation and to maintain the ecological balance by means of"Long Forest"worship,"Long Forest"taboo,the traditional customary law,and village regulation and non-governmental agreement.This paper takes Manjingchengzi Village in Xishuangbanna as an example to analyze the distribution of forest land in different periods over the past 60 years and explore the role of"Long Forest"belief in the conservation of local forest resources in a view to providing a frame of reference for regional ecological environment protection.

Long-range Dependence Characteristics of Forest Biological Disasters in China against the Background of Climate Change

Forest biological disasters(FBD) seriously impact energy flow and material cycling in forest ecosystems,while the underlying causes of FBD are complex. These disasters involve large areas and cause tremendous losses. As a result,the occurrence of FBDs in China( CFBD) threatens the country's ability to realize its strategic target of increasing both forested area(40 million ha) and forest volume(1.3 billion m^3) from 2005 to 2020. Collectively,China has officially named this effort to increase forest area and volume the "Two Increases" as national goals related to forestry. Based on Hurst index analysis from fractal theory,we analyzed the time series of the occurrence area and related data of FBDs from 1950 to 2007 to quantitatively determine the patterns of the macro occurrence of FBDs in China. Results indicates that,the time series of( CFBD) areas is fractal( self-affinity fractal dimension D = 1. 3548),the fluctuation of( CFBD) areas is positively correlated( auto-correlation coefficient C = 0. 2170),and the occurrence of the time series of( CFBD) is long-range dependent( Hurst index H =0. 6416),showing considerably strong trend of increases in FBDC area. Three different methods were further carried out on the original time series,and its two surrogate series generated by function surrogate in library t series,and function Surrogate Data in library in Wavelet software R,so as to analyze the reliability of Hurst indexes. The results showed that the Hurst indices calculated using different estimation methods were greater than 0. 5,ranging from 0. 64 to 0. 97,which indicated that the change of occurrence area data of FBDs was positively autocorrelated.The long-range dependence in forest biological disasters in China is obvious,and the spatial extent of FBDs tended to increase during this study period indicating this trend should be expected to persistent and worsen in the future.

面向喷染车间的挥发性有机物质量浓度预测方法及应用研究

以喷染车间挥发性有机物为研究对象,对喷染车间挥发性有机物(Volatile Organic Compounds, VOCs)质量浓度预测方法进行研究。首先,使用随机森林(Random Forest, RF)算法对影响喷染车间挥发性有机物质量浓度的特征变量进行权重分析。同时,构建基于长短期记忆神经网络(Long-Term and Short-Term Memory Neural Network, LSTM)的挥发性有机物质量浓度预测模型,并在此基础上引入麻雀搜索算法(Sparrow Search Algorithm, SSA)进行参数优化选择。最后,以浙江省杭州市某汽车喷染车间7月29日—10月28日的数据为样本,将温度、相对湿度、室内大气压、室外大气压作为模型输入变量,并与LSTM模型、随机森林-长短期记忆神经网络(Random Forest-Long Short-Term Memory neural network, RF-LSTM)模型、随机森林-反向传播神经网络(Random Forest-BP neural network, RF-BP)模型进行对比试验。结果显示,基于随机森林-麻雀搜索算法-长短期记忆神经网络(Random Forest-Sparrow Search Algorithm-Long Short-Term Memory neural network, RF-SSA-LSTM)模型的预测效果最佳,平均绝对误差、均方根误差和决定系数分别为2.812 2、3.457 4、0.988。同时,为验证RF-SSA-LSTM模型性能,通过不同时间步长实现对喷染车间VOCs质量浓度预测,结果显示预测误差较小,在可接受范围内。RF-SSA-LSTM预测模型提高了挥发性有机物质量浓度的预测精度,为减少挥发性有机物排放提供科学依据。

Effects of Aluminum Toxicity Induced by Acid Deposition on Pine Forest Ecosystem in Longli of Guizhou Province, Southwestern China

The effects of acid deposition on pine forest ecosystems in Longli of Guizhou Province, southwestern China are studied using indoor experiments and model simulations. Indoor experiments are designed to explore the aluminum toxicity on pine seedlings, and the long-term soil acidification model(LTSAM) and a terrestrial biogeochemistry model(CENTURY) are used to simulate the influences of acid deposition on pine forest ecosystems. The indoor experiment results of aluminum toxicity show that aluminum ions in solution limit plant growth and acid deposition enhances this effect by facilitating the release of aluminum ions from the soil. Pine seedling biomass and root elongation decrease as the aluminum concentration increases. The results of model simulations show that the soil chemistry varies significantly with different changes in acid deposition. When the acid deposition increases, the pH value in the soil solution decreases and the soil Al^(3+) concentration increases. The increased acid deposition also has negative impacts on the forest ecosystem, i.e., decreases plant biomass, net primary productivity(NPP) and net CO_2 uptake. As a result, the soil organic carbon(SOC) decreases because of the limited supply of decomposition material. Thus acid deposition need be reduced to help protect the forest ecosystems.

长时间序列生态系统服务权衡与协同驱动因素——以芜湖市生态系统服务功能极重要区为例

人类对生态系统的管理和干预促使生态系统服务权衡向协同转化,有助于生态系统服务综合效益提升。但当前缺乏对长时间序列分析背景下生态系统服务权衡协同驱动因素及其非线性影响的理解与把握。以芜湖市“生态系统服务功能极重要区”为研究案例,考虑生境质量、碳固定和土壤保持三种关键生态系统服务,基于多源数据,运用逐像元趋势叠加分析法评估1990—2020年间研究区生态系统服务变化趋势及其权衡协同空间分布规律,采用随机森林模型探索该区域生态系统服务权衡协同形成的关键驱动因素及其非线性影响效应。研究结果表明:(1)31年间芜湖市域生境质量显著下降区域面积约为提升面积的两倍,存在进一步衰退的风险。同时,城市扩张导致市郊边缘地带碳固定、土壤保持显著下降。(2)研究区生态系统服务供给能力受权衡协同影响显著,受影响区域总面积占比达64.48%。受权衡影响区域与协同影响区域存在显著空间差异。(3)土地利用强度是管理生态系统服务权衡向协同转化的主导因素。土地利用强度对权衡协同的影响表现出非线性特征并且存在影响阈值。当强度等级介于弱与中且偏向弱强度时对不同服务协同增益具有积极作用。

立竹密度对不同发笋期毛竹叶片功能性状的影响

【目的】探究长周期母竹留养模式下毛竹叶片功能性状对立竹密度的响应及在不同发笋期的变化,并筛选出适宜的经营密度。【方法】以漳平市长周期母竹留养经营模式下笋用毛竹林为研究对象,设置(1350±100)、(1950±100)、(2550±100)、(3150±100)株·hm^(-2)4个不同立竹密度处理组,选用传统经营模式下生产效益较优的笋用毛竹林作为对照组,分析立竹密度对不同发笋期毛竹叶片功能性状的影响。【结果】毛竹林的叶面积指数、叶平均倾角、全氮含量、全磷含量、全钾含量和叶绿素含量均在立竹密度为(2550±100)株·hm^(-2)时达到最高,且不同处理组间差异显著(P<0.05);各处理组的叶面积指数、叶平均倾角及全氮含量均在发笋前期达到最高,冠层开度和叶绿素含量在发笋盛期达到最高,全磷和全钾含量则在发笋后期达到最高;毛竹叶片各功能性状间存在显著相关性。【结论】在长周期母竹留养经营模式下,当立竹密度为(2550±100)株·hm^(-2)时,毛竹林的冠层结构得到显著改善,叶片的空间分布更加合理,且叶片养分及叶绿素含量显著提高。

随机森林优化的静动态耦合模型在滑坡位移预测中的应用

以重庆市奉节县生基包滑坡为例,首先采用静态的支持向量回归(SVR)机器学习算法和动态的长短期记忆神经网络(LSTM)机器学习算法对滑坡位移进行预测;其次引入随机森林(RF)算法,在输入因素筛选的基础上,对SVR模型和LSTM模型的预测结果进行更优解分类预测;最后通过RF模型输出概率值,对静动态耦合模型(SVR-LSTM)进行权重赋值,得到RF优化的SVR-LSTM滑坡位移预测模型。结果表明LSTM模型预测整体优于SVR模型,RF优化的SVR-LSTM滑坡位移预测模型集成了静态SVR与动态LSTM预测模型的优势,其预测性能与单一的SVR模型和LSTM模型相比更优。研究提供了一种滑坡位移预测模型集成的思路,为三峡库区的地质灾害预测预报提供借鉴和参考。

基于iForest-BiLSTM-Attention的数据库负载预测方法

针对数据库负载预测中物理资源的变化导致预测失效,模型易对异常数据敏感和未关注序列变化中潜在的加权隐层特征状态导致预测精度低等问题,在长短期记忆网络模型的基础上提出一种基于iForest-BiLSTM-Attention的数据库负载预测方法。首先,增加数据库基准规范内部指标,解决因物理资源改变而导致的传统指标预测失效问题;其次,建立多个孤立树,整合为孤立森林,评估样本异常分数并筛出异常数据进行热卡填充;最后,结合注意力机制与双向长短期记忆网络计算隐层状态以及注意力权值,并学习工作负载的形态、周期以及规律性。实验结果表明,所提方法在数据库工作负载预测精度上相比现有方法有显著提升,吞吐量和CPU利用率的R 2值分别达到0.93和0.95。

基于机器学习的盾构掘进参数预测

依托济南市济泺路穿黄隧道东线工程,选取1130组掘进数据,按照施工顺序划分数据集,采用粗细程度、软硬程度、密实程度和渗透能力4个维度描述土体的物理力学状态,分别建立基于长短期记忆模型(Long-Short Term Memory,LSTM)、随机森林模型(Random Forest)和BP神经网络的盾构隧道掘进参数预测模型,详细对比分析3种模型对总推力和掘进速度的预测效果。研究表明:(1)LSTM模型在按施工顺序预测盾构总推力和掘进速度时,平均相对误差仅为3.72%和7.41%,模型训练时间均在20 s以内,整体表现优于随机森林模型和BP神经网络;(2)在地形发生剧烈变化以及盾构掘进线路在直线与平曲线过渡时,总推力和掘进速度出现较大波动,LSTM模型预测结果相对误差偏大的组数仅占4%与10.2%,且总体误差满足施工要求;(3)随机森林模型预测结果的相对误差在总推力和掘进速度剧烈波动的环段处偏大,数量偏多,因此在按施工顺序预测时不是优选。

麻雀搜索算法优化的RF-BILSTM短期电力负荷预测

电力负荷的准确预测是电网安全运行以及社会正常生产的重要保障,但负荷数据因自身的非线性以及众多影响因素的不确定性大大降低预测精度的准确性,因此,为了提高短期电力负荷的预测精度,提出一种基于麻雀搜索算法(SSA)优化随机森林(RF)-双向长短期记忆神经网络(BILSTM)的预测模型。首先,针对电力负荷数据特征构造等问题,对数据进行量化以及标准化预处理操作,便于后续模型的输入。其次,利用RF算法对电力负荷的众多影响因素进行重要性排序,保留重要因素并将其与历史负荷数据进一步结合,从而构成神经网络的最终输入。最后,采用SSA算法对BILSTM模型的部分超参数进行优化选取,解决人工选取困难的问题。通过与其他模型对比,验证了该模型具有较高的预测精度。

基于RF-EMD-LSTM对PM_(2.5)浓度预测研究

实现PM_(2.5)浓度的精准预测对空气污染防治具有重要的指导作用.鉴于多数研究中PM_(2.5)浓度预测算法和特征都较为单一,不能精确实现PM_(2.5)浓度的短时预测,提出了一种新颖的混合预测框架(RF-EMD-LSTM),通过过去24 h数据实现下一小时PM_(2.5)浓度预测.首先利用随机森林算法(random forests, RF)对污染物数据特征进行重要性评估,实现特征选择;然后再将特征数据与通过经验模态分解(empirical mode decomposition, EMD)的PM_(2.5)特征分解后的序列联合作为长短期记忆网络(long short term memory network, LSTM)的输入数据,进行模型训练,训练结果相加获得最终PM_(2.5)浓度预测值.实验结果表明,提出的混合模型与其他单一模型相比,预测精度明显最佳.

考虑上游来水影响的中长期径流预报

雅砻江流域地面气象站点不足、分布不均,难以获得精确的流域面降雨资料,加之传统中长期径流预报模型泛化能力有限,中长期径流预报存在较大瓶颈。充分考虑流域水库间的物理联系,基于上下游水库流量变化在时空上的相似性,对1957年~2020年锦屏一级水库和二滩水库的历史月径流数据进行主成分分析,使用BP人工神经网络、随机森林和支持向量回归3种机器学习方法建立3种径流预报模型,通过决定系数R^(2),合格率Q R以及平均相对误差MRE三项指标构成的评价体系对预测结果进行评估。结果表明,上游水库对于下游水库的入库流量具有显著影响,且3种模型在二滩水库中长期径流预报上均具有较好的预报效果(R^(2)>0.8、Q R>0.7、MRE<0.2)。随机森林模型模拟效果整体优于BP人工神经网络和支持向量回归模型,3种模型均具有较好的实用性,能为流域水资源精细化调度及科学管理提供数据基础。

Integrating beneficiaries into assessment of ecosystem services from managed forests at the Hubbard Brook Experimental Forest,USA

Background：Forests contribute to human wellbeing through the provision of important ecosystem services.Methods：In this study,we investigated how the perceived importance of ecosystem services may impact the overall benefit provided by managed watersheds at the Hubbard Brook Experimental Forest over a 45-year period,using standardized measures of service capacity weighted by service importance weights derived from a survey of beneficiaries.Results：The capacity of watersheds to regulate water flow and quality was high in all watersheds throughout the study period,whereas cultural services such as scenic beauty declined after harvest.Impacts on greenhouse gas regulation depended on the efficiency with which harvested biomass was used.Surveys revealed that stakeholders placed high value on all ecosystem services,with regulating and cultural services seen as more important than provisioning services.When service metrics were weighted by survey responses and aggregated into a single measure,total service provision followed the same overall trend as greenhouse gas regulation.Where biomass use was less efficient in terms of greenhouse gas emissions,harvesting resulted in an overall ＂ecosystem service debt＂;where use was more efficient,this ＂ecosystem service debt＂ was reduced.Beneficiaries＇ educational backgrounds significantly affected overall assessment of service provision.Beneficiaries with college or university degrees incurred smaller ＂ecosystem service debts＂ and were less negatively affected by harvesting overall.Conclusions：This study highlights the importance of including empirical measures of beneficiary preference when attempting to quantify overall provision of ecosystem services to human beneficiaries over time.

基于改进BiLSTM-RF的短期负荷预测研究

电力负荷的准确预测能有效保持电网运行的稳定性,提高经济效益和社会效益。为了提高负荷预测的精准度,首先利用麻雀搜索算法(SSA)和变分模态分解(VMD)对输入的原始负荷进行模态分解,降低了电力负荷数据随机性与非平稳性;然后利用双向长短期记忆-随机森林(BiLSTM-RF)组合模型对分解后的子模态进行特征提取和预测。对某地区公开数据的性能验证与模型对比分析结果表明,改进的BiLSTM-RF(+BiLSTM-RF)组合模型在决定系数(R^(2))、均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分误差(MAPE)这四个预测精度指标方面分别达到了0.973、300.31、134.91、0.037。与传统的支持向量机(SVM)、长短期记忆(LSTM)网络、门控循环单元(GRU)等单一预测方法,以及未改进的BiLSTM-RF组合预测方法相比,+BiLSTM-RF组合模型有更好的预测表现。

基于特征变权的超短期风电功率预测

针对当前风电功率预测过程中历史信息利用不充分及多维输入权重值固定忽略了不同时间维度的特征重要性的问题,提出一种基于特征变权的风电功率预测模型。该方法利用随机森林(RF)分析不同高度处的风速、风向、温度等气象特征对风电输出功率的影响程度,并利用累积贡献率完成气象特征的提取。对提取的特征及历史功率信息利用奇异谱分析(SSA)去噪,以去噪后的数据作为输入建立级联式FA-CNN-LSTM多变量预测模型对超短期风电功率进行预测。通过在CNN-LSTM网络中增加特征注意力机制(FA)自适应挖掘不同时刻的特征关系,动态调整不同时间维度各输入特征的权重,加强预测时刻关键特征的注意力,从而提升预测性能。基于某风电场实测数据的算例分析表明,所提方法可有效提高超短期风电功率预测精度。

基于RF的并行CNN-TGLSTM热负荷预测模型

精准的热负荷预测不仅可以提高用户舒适度,还可以有效降低能源消耗。为了提升热负荷预测的准确性,本研究提出了一种基于随机森林的并行CNN和TGLSTM的短期热负荷预测模型。首先,采用随机森林算法对特征进行筛选;其次,利用并行网络CNN和改进的LSTM分别提取时空特征;最后,将提取的特征与多头注意力机制动态结合。实验结果表明,并行CNN-TGLSTM-MA相较于传统的串行CNN-TGLSTM模型,在MAE和MSE方面分别降低了16.9%、26.8%,同时提升了3.5%的R2值,证明了所提出的并行CNN-TGLSTM-MA模型在短期热负荷预测方面的有效性和优越性,为热力系统供热负荷的精准调控提供了参考。

Recalcitrant carbon controls the magnitude of soil organic matter mineralization in temperate forests of northern China

Background: The large potential of the soil organic carbon(SOC) pool to sequester CO2from the atmosphere could greatly ameliorate the effect of future climate change. However, the quantity of carbon stored in terrestrial soils largely depends upon the magnitude of SOC mineralization. SOC mineralization constitutes an important part of the carbon cycle, and is driven by many biophysical variables, such as temperature and moisture.Methods: Soil samples of a pine forest, an oak forest, and a pine and oak mixed forest were incubated for 387 days under conditions with six temperature settings(5 °C, 10 °C, 15 °C, 20 °C, 25 °C, 30 °C) and three levels of soil moisture content(SMC, 30%, 60%, 90%). The instantaneous rate of mineralized SOC was periodically and automatically measured using a Li-Cor CO2analyzer. Based on the measured amount of mineralized SOC,carbon fractions were estimated separately via first-order kinetic one-and two-compartment models.Results: During the 387 day incubation experiment, accumulative mineralized carbon ranged from 22.89 mg carbon(C) ·g-1SOC at 30 °C and 30% SMC for the mixed forest to 109.20 mg C·g-1SOC at 15 °C and 90% SMC for the oak forest. Mineralized recalcitrant carbon varied from 18.48 mg C·g-1SOC at 30 °C and 30% SMC for the mixed forest to 104.98 mg C·g-1SOC at 15 °C and 90% SMC for the oak forest, and contributed at least 80% to total mineralized carbon.Conclusions: Based on the results of this experiment, the soil organic matter of the pure broadleaved forest is more vulnerable to soil microbial degradation in northern China; most of the amount of the mineralized SOC derived from the recalcitrant carbon pool. Labile carbon fraction constitutes on average 0.4% of SOC across the three forest types and was rapidly digested by soil microbes in the early incubation stage. SOC mineralization markedly increased with soil moisture content, and correlated parabolically to temperature with the highest value at 15 °C. No significant interaction was detected among these variables in the present study.

基于RF-LSTM网络的风电机组状态参数预测

为实现风电机组的早期故障预警,提出一种基于随机森林(random forest, RF)结合长短期记忆(longshort-termmemory,LSTM)网络的风电机组状态参数预测方法。利用RF算法筛选出与目标变量有重要关联信息的状态参数,将其作为LSTM网络预测模型的输入变量,实现对目标变量的预测。采集风电机组监测的状态参数数据,将齿轮箱油温作为目标变量,对所提预测方法的有效性进行验证。结果表明,在不同的输入变量与预测方法下,RF-LSTM均具有较高的预测精度,不仅克服了众多状态参数存在的高维度、非线性等问题,还保证了输入变量的重要信息不会随着时间被遗忘。

基于GEE和多维特征集的锡林浩特露天矿区近30 a土地利用分类

随着矿产资源的不断开采,矿区地表土地利用会发生频繁的变化,快速获取矿区开采过程中长时间序列连续的土地利用分类结果对于矿区土地治理与生态重建具有重要意义。基于Google Earth Engine(GEE)遥感云平台,对Sentinel-2和Landsat数据从光谱特征、纹理特征、地形特征3个方面构建多维特征集,采用随机森林算法分别建立了不同特征模型并测试精度以筛选出适合矿区场景的最优分类模型。以锡林浩特市露天矿区为研究区,开展了近30 a(1991—2020年)长时序土地利用分类研究。结果表明:①基于GEE遥感云平台,能够高效、快速、准确地提取研究区1991—2020年近30 a的土地利用分类结果;②光谱特征对分类精度具有决定性作用,融入纹理特征和地形特征能够有效提高矿区土地利用分类模型精度;③Sentinel-2数据特有的红边波段对植被具有较高的敏感性,能有效提高分类精度。长时间连续的监测结果能够有效了解锡林浩特市露天矿区土地利用的变化情况及规律,为进一步分析人类生产生活和环境变化对土地利用的影响提供了坚实的数据支撑。