毛竹林总初级生产力年际变化及其驱动因素——以安吉县为例

森林生态系统碳通量的年际变化及其驱动因素分析是了解森林碳收支动态变化以及预测未来气候变化对森林碳收支影响的重要理论基础,对评估森林应对气候变化的贡献具有重要意义。结合MODIS叶面积指数(LAI)和归一化植被指数(NDVI)产品、MERRA气象数据和通量塔观测数据,采用光能利用率模型模拟2004—2011年安吉县毛竹林生态系统总初级生产力(GPP)空间分布,并分析GPP年际变化及其驱动因素。结果表明:(1)小年毛竹林GPP稍高于大年GPP;(2)2004—2011年安吉县毛竹林年日均GPP呈下降趋势,东部、西部和整个安吉县毛竹林年日均GPP变化速率分别为-0.064、-0.033和-0.045g C m-2W-1,年均温度持续下降是主要驱动因素;(3)LAI年际变化是GPP年际变化的主要驱动因素,主要原因是毛竹林大小年交替规律引起了有效LAI年际间差异;(4)西部GPP年际变化幅度大于东部,环境和生物因素对GPP年际变化的作用方向决定了毛竹林GPP年际变化的幅度。     

基于数据同化的哈佛森林地区水、碳通量模拟

模型模拟和站点观测是陆地生态系统水、碳循环研究最主要的两种手段,但各有优势和不足,若二者相互结合,则能更准确地反映生态系统水、碳通量的动态变化.数据同化为模型与观测结合提供了一条有效的途径.本文采用哈佛森林环境监测站相关数据,利用集合卡曼滤波同化算法,将实测叶面积指数(LAI)和遥感LAI同化进入Biome BGC模型中,对该地区水、碳通量进行模拟.结果表明：与未同化模拟相比,将1998、1999和2006年实测LAI数据同化后,模型模拟碳通量(NEE)与通量观测NEE的决定系数(R²)平均提升8.4%;蒸散发(ET)的R²平均提升10.6%;NEE的绝对误差和(SAE)和均方根误差(RMSE)平均下降17.7%和21.2%,ET的SAE和RMSE平均下降26.8%和28.3%.将2000-2004年MODIS LAI 产品与模型同化后,NEE、ET模拟值与观测值间的R²分别提升7.8%和4.7%;NEE的SAE和 RMSE分别下降21.9%和26.3%,ET的SAE和 RMSE分别下降24.5%和25.5%.无论实测LAI还是遥感观测LAI,同化进入模型都能不同程度地提高水碳通量的模拟精度.     

基于Biome-BGC模型和集合卡尔曼滤波方法的阔叶红松林生态系统水碳通量模拟

数据同化为模型与遥感观测结合提供了一条有效的途径,通过在模型运行过程中融入遥感观测数据,调整模型运行轨迹从而降低模型误差,提高模拟精度。本文利用集合卡尔曼滤波(En KF)算法同化生长季中分辨率成像光谱仪(MODIS)叶面积指数(LAI)与Biome-BGC模型模拟的LAI模拟长白山阔叶红松林的水碳通量。同时,通过改进模拟的雪面升华与土壤温度计算方法的参数,旨在降低冬季生态呼吸的模拟误差。结果表明,相对于原始模型,数据同化与模型改进后使得生态系统总初级生产力(GPP)的模拟值与观测值之间的相关系数提高0.06,中心化均方根误差(RMSE)降低0.48 g C·m~(-2)·d~(-1);生态系统呼吸(RE)的相关系数提高0.02,中心化均方根误差降低0.20 g C·m~(-2)·d~(-1);净生态系统碳交换量(NEE)相关系数提高0.35,中心化均方根误差降低0.50 g C·m~(-2)·d~(-1)。同时,数据同化对蒸散发(ET)的模拟精度没有显著影响,改进的模型提高了其相关系数。基于En KF算法的数据同化提高了长白山阔叶红松林碳通量模拟精度,对于精确估算区域碳通量有着重要的意义。     

基于集合卡尔曼滤波同化方法的农田土壤水分模拟

引入实际土壤水分观测数据,运用基于集合卡尔曼滤波的数据同化方法,利用改进的生态机理模型BEPS(boreal ecosystem productivity simulator)模拟了江苏省徐州农业试验站2000-2004年冬小麦生长季的根层土壤水分动态,结果表明:经过引入观测数据进行同化后由BEPS模型得到的土壤水分模拟结果与观测值之间的决定系数、均方根误差和平均绝对误差分别为0.626 ～O.943、0.018～0.042和0.021～0.041,模拟精度较同化前有显著提高,验证了利用数据同化算法提高模型对土壤水分模拟精度的可行性.单点试验结果表明,输入变量的误差、观测值的误差、引入观测数据的频率以及引入观测数据的深度等因素对利用集合卡尔曼滤波进行同化后得到的土壤水分模拟结果有较大影响.     

利用大气二氧化碳和羰基硫浓度评估陆地生态系统模型碳通量模拟的不确定性

生态系统模型是模拟全球陆地生态系统碳循环的重要工具,但是其在全球不同区域的模拟存在很大的不确定性。如何评估陆地生态系统模型的不确定性是一项重要的研究。以北美地区为例,利用8个高塔观测站点同步获取的大气CO_2和羰基硫(OCS)浓度数据,结合WRF-STILT大气粒子扩散模型,评估了CASA-GFED3、SiB3和SiBCASA三种陆地生态系统模型模拟总初级生产力(GPP)和净生态系统CO_2交换(NEE)通量的不确定性。结果表明,SiB3模型能很好地模拟北美陆地生态系统GPP和NEE的季节变化时相和幅度,在3种模型中具有最佳的模拟能力;CASA-GFED3模型模拟的NEE季节变化较为理想、但对生长季GPP的模拟存在较大的误差,SiBCASA模型在模拟冬季晚期和春季早期的NEE和GPP时表现较不理想。研究证明了大气CO_2和OCS在评估陆地生态系统模型碳通量模拟的不确定性中的作用,为利用大气CO_2和OCS观测数据优化计算陆地生态系统光合和呼吸碳通量提供了理论支撑。     

利用亚像元尺度信息改进区域冬小麦生长的模拟

为了探寻遥感观测面尺度与作物模型模拟点尺度不匹配问题的解决方案并改善区域作物生长模拟精度,以河南省鹤壁市为研究区,以冬小麦为研究对象,基于MODIS、Landsat 8遥感数据和Wheat SM作物生长模型,通过MODIS LAI过程线重建、亚像元尺度信息提取、集合卡尔曼滤波同化等方法,进行了冬小麦生长模拟的研究。结果表明:通过MODIS LAI过程线重建并提取亚像元尺度信息,冬小麦纯度在80%以上的遥感反演LAI与冬小麦两个关键生育期实测冠层LAI的均方根误差(RMSE)为0.69,以最近邻域法赋值到整个模拟区域,研究区2013—2017年模拟总产和实际总产相比的RMSE在未同化遥感反演的LAI信息时为6.73×108kg,同化未利用亚像元尺度信息调整的遥感估算LAI时,RMSE上升到8.24×108kg,同化利用亚像元尺度信息分区赋值的遥感LAI时,RMSE下降到3.48×108kg。利用亚像元尺度信息生成与作物模型时空尺度匹配的格点化LAI遥感产品,可提高作物生长模型区域化应用的精度。     

基于FLUXNET的CLM模型总初级生产力模拟评价与误差分析

总初级生产力(gross primary productivity,GPP)的准确估计是陆地生态系统碳循环研究以及未来气候变化预测的基础。本文利用全球通量网的通量观测数据评估了CLM4.5的GPP模拟效果,结合叶面积指数(leaf area index,LAI)观测数据分析了CLM4.5模型GPP模拟误差的主要原因,并对CLM4.5模型主要的光合参数进行敏感性分析,探讨改善光合作用模拟精度的可能途径。结果表明:CLM4.5对GPP的模拟效果优于CLM4,月尺度和年尺度GPP模拟值的平均绝对偏差分别降低15%和29%。但与观测值相比,CLM4.5模拟的GPP年总量仍然具有较大偏差,平均偏差为366.06 g C·m-2·a-1。不同植被功能型GPP模拟误差具有不同的季节变化特征,误差主要发生在春季和夏季。冠层顶部比叶面积、叶片碳氮比和叶片氮含量中Rubisco氮含量所占比例是GPP模拟的3个敏感参数。未来应主要从物候期及LAI模拟的改进、磷循环过程的影响、生态系统水平光合参数集的构建等方面实现对GPP模拟精度的提高。     

基于FORCCHN模型的中国典型森林生态系统碳通量模拟

森林生态系统是陆地碳循环的重要组成部分,其固碳能力显著高于其他陆地生态系统,研究森林生态系统碳通量是认识和理解全球变化对碳循环影响的关键。碳循环模型是研究森林生态系统碳通量有效工具。以长白山温带落叶阔叶林、千烟洲亚热带常绿针叶林、鼎湖山亚热带常绿阔叶林和西双版纳热带雨林等4种中国典型森林生态系统为研究对象,利用涡度相关2003-2012年观测数据,评估FORCCHN模型对生态系统呼吸（ER）,总初级生产力（GPP）,净生态系统生产力（NEP）的模型效果。结果表明：（1） FORCCHN模型能够较好的模拟中国4种典型森林生态系统不同时间尺度的碳通量。落叶阔叶林和常绿针叶林ER和GPP的逐日变化模拟效果较好（ER的相关系数分别为0.94和0.92,GPP的相关系数分别为0.86和0.74）;（2）4种森林生态系统碳通量季节动态模拟值和观测值显著相关（P<0.01）,ER、GPP、NEP的观测值和模拟值的R²分别为0.77-0.93、0.54-0.88和0.15-0.38;模型可以很好地模拟森林生态系统不同季节碳汇（NEP>0）,碳源（NEP<0）的变化规律;（3）4种森林生态系统碳通量模拟值与观测值的年际变化有很好的吻合度,但在数值大小上存在差异,模型高估了常绿阔叶林的ER和GPP,略微低估了其他3种森林生态系统ER和GPP。     

基于MODIS时间序列反射率数据的雷竹林LAI反演

本文以雷竹林为研究对象,基于MODIS地表反射率数据构建了归一化植被指数(NDVI)、比值植被指数(SR)、Gitelson绿色植被指数(GI)、增强型植被指数(EVI)和土壤调整植被指数(SAVI)5种植被指数,并将其与MODIS 7个波段原始反射率数据作为遥感变量,采用逐步回归和相关分析两种方法进行变量筛选,结合LAI实测数据构建了逐步回归和BP神经网络两种模型,对雷竹林生态系统观测站点2014年1月-2017年3月LAI时间系列数据进行反演,并将反演结果与同时期MOD15A2 LAI产品进行对比分析.结果表明: SR为唯一入选逐步回归模型的变量;b₁、b₂、b₃和b₇以及5种植被指数与LAI之间的相关性均达到显著水平,可作为BP神经网络模型的输入变量.使用BP神经网络反演得到的LAI与实测LAI之间的相关性显著,R²为0.71,RMSE为0.34,RMSE_r为13.6%,其R²比逐步回归模型提高了10.9%,RMSE降低了5.6%,RMSE_r降低了12.3%,与MODIS LAI相比,其R²提高了54.5%,RMSE降低了79.3%,RMSE_r降低了79.1%.结合MODIS时间序列反射率和BP神经网络模型能够精确地反演雷竹林LAI,为实现基于遥感技术快速监测区域雷竹林LAI提供可行的方法.     

PROSPECT5耦合4SAIL模型的亚热带典型森林冠层反射率时间序列模拟

冠层反射率在森林植被类型精确解译、森林碳同化关键参数如叶面积指数(LAI)、叶绿素等遥感反演等方面具有重要意义.本研究以亚热带毛竹林、雷竹林和常绿落叶阔叶混交林3种典型森林类型为研究对象,通过耦合PROSPECT5和4SAIL模型模拟其冠层反射率时间序列.首先,对PROSPECT5和4SAIL模型参数进行敏感性分析,探讨模型参数对冠层反射率的影响;其次,利用实测反射率对不敏感参数进行优化,并确定其参数值;最后,耦合PROSPECT5和4SAIL模型模拟3种亚热带森林冠层反射率,并与MODIS反射率进行对比.结果表明:LAI对第1、2、3、5、7波段最敏感,各波段的总敏感指数分别为0.80、0.83、0.94、0.66、0.47;叶绿素含量对第4波段最敏感,总敏感指数为0.59;叶片含水量对第6波段的敏感性最大,总敏感性指数为0.54;叶子结构参数、类胡萝卜素、热点参数、干物质含量和土壤干湿比等参数对各个波段都不敏感或敏感性较小.优化后的PROSPECT5和4SAIL模型模拟得到的冠层反射率能够真实反映3种典型森林的季节性变化规律,通过与MODIS反射率对比分析发现,模拟冠层反射率和MODIS反射率之间具有较高的决定系数,分别为0.86、0.90、0.93,均方根误差(RMSE)也较小,分别为0.09、0.07、0.05,且模拟反射率能在一定程度上解决MODIS反射率数据冬季易受雨雪、混合像元影响等问题.     

NEE观测误差分布类型对陆地生态系统机理模型参数估计的影响——以长白山温带阔叶红松林为例

基于观测数据的陆地生态系统模型参数估计有助于提高模型的模拟和预测能力,降低模拟不确定性.在已有参数估计研究中,涡度相关技术测定的净生态系统碳交换量(NEE)数据的随机误差通常被假设为服从零均值的正态分布.然而近年来已有研究表明NEE数据的随机误差更服从双指数分布.为探讨NEE观测误差分布类型的不同选择对陆地生态系统机理模型参数估计以及碳通量模拟结果造成的差异,以长白山温带阔叶红松林为研究区域,采用马尔可夫链-蒙特卡罗方法,利用2003～2005年测定的NEE数据对陆地生态系统机理模型CEVSA2的敏感参数进行估计,对比分析了两种误差分布类型(正态分布和双指数分布)的参数估计结果以及碳通量模拟的差异.结果表明,基于正态观测误差模拟的总初级生产力和生态系统呼吸的年总量分别比基于双指数观测误差的模拟结果高61～86 g C m-2 a-1和107～116 g C m-2 a-1,导致前者模拟的NEE年总量较后者低29～47 g C m-2 a-1,特别在生长旺季期间有明显低估.在参数估计研究中,不能忽略观测误差的分布类型以及相应的目标函数的选择,它们的不合理设置可能对参数估计以及模拟结果产生较大影响.     

Influences of error distributions of net ecosystem exchange on parameter estimation of a process-based terrestrial model: A case of broad-leaved Korean pine mixed forest in Changbaishan, China

Model predictions can be improved by parameter estimation from measurements. It was assumed that measurement errors of net ecosystem exchange (NEE) of CO₂ follow a normal distribution. However, recent studies have shown that errors in eddy covariance measurements closely follow a double exponential distribution. In this paper, we compared effects of different distributions of measurement errors of NEE data on parameter estimation. NEE measurements in the Changbaishan forest were assimilated into a process-based terrestrial ecosystem model. We used the Markov chain Monte Carlo method to derive probability density functions of estimated parameters. Our results showed that modeled annual total gross primary production (GPP) and ecosystem respiration (Re) using the normal error distribution were higher than those using the double exponential distribution by 61–86 gC m^?2 a^?1 and 107–116 gC m^?2 a^?1, respectively. As a result, modeled annual sum of NEE using the normal error distribution was lower by 29–47 gC m^?2 a^?1 than that using the double exponential error distribution. Especially, modeled daily NEE based on the normal distribution underestimated the strong carbon sink in the Changbaishan forest in the growing season. We concluded that types of measurement error distributions and corresponding cost functions can substantially influence the estimation of parameters and carbon fluxes.     

The Carbon Balance of Two Contrasting Mountain Forest Ecosystems in Switzerland: Similar Annual Trends,but Seasonal Differences

Net ecosystem exchange (NEE) of two contrasting mountain forest types in Switzerland was measured by eddy covariance (EC) measurements at a montane mixed forest, the Lägeren forest, over 5 years (2005–2009), and at a subalpine coniferous forest, the Seehornwald in Davos, over 12 years (1997–2009). NEE was validated against annual carbon (C) storage estimates, based on biometric and soil respiration measurements as well as soil C modeling. Three different approaches were used: (1) calculation of net ecosystem production by quantifying C pools and fluxes, (2) assessment of change in wood biomass and soil C storage (ΔC), and (3) application of biomass expansion factors. Although biometric estimates were sensitive to assumptions made for each method applied, they agreed well with measured NEE. Comparing 5 years of EC measurements available at both sites during 2005 and 2009 revealed that NEE, gross primary production (GPP), and total ecosystem respiration (TER) were larger at the Lägeren forest compared to the Davos forest, whereas soil respiration and soil C sequestration were of similar magnitudes. Both sites showed similar annual trends for NEE, GPP and TER, but different seasonal courses, due to different responses to environmental conditions (temperature, soil moisture, and radiation). Differences in the magnitude as well as in the seasonality of ecosystem CO₂ exchange could mainly be attributed to tree phenology, productivity, and carbon allocation patterns, which are combined effects of tree type (broad-leaved vs. coniferous trees) and site-specific climatic conditions. Flux differences between the two mountain sites highlight the importance of considering the role of altitude in ecological studies and modeling.     

Carbon dioxide exchange above a Mediterranean C3/C4 grassland during two climatologically contrasting years

Eddy‐covariance measurements of net ecosystem carbon exchange (NEE) were carried out above a grazed Mediterranean C3/C4 grassland in southern Portugal, during two hydrological years, 2004–2005 and 2005–2006, of contrasting rainfall. Here, we examine the seasonal and interannual variation in NEE and its major components, gross primary production (GPP) and ecosystem respiration (R_eco), in terms of the relevant biophysical controls. The first hydrological year was dry, with total precipitation 45% below the long‐term mean (669 mm) and the second was normal, with total precipitation only 12% above the long‐term mean. The drought conditions during the winter and early spring of the dry year limited grass production and the leaf area index (LAI) was very low. Hence, during the peak of the growth period, the maximum daily rate of NEE and the light‐use and water‐use efficiencies were approximately half of those observed in the normal year. In the summer of 2006, the warm‐season C4 grass, Cynodon dactylon L., exerted an evident positive effect on NEE by converting the ecosystem into a carbon sink after strong rain events and extending the carbon sequestration for several days, after the end of senescence of the C3 grasses. On an annual basis, the GPP and NEE were 524 and 49 g C m^?2, respectively, for the dry year, and 1261 and ?190 g C m^?2 for the normal year. Therefore, the grassland was a moderate net source of carbon to the atmosphere, in the dry year, and a considerable net carbon sink, in the normal year. In these 2 years of experiment the total amount of precipitation was the main factor determining the interannual variation in NEE. In terms of relevant controls, GPP and NEE were strongly related to incident photosynthetic photon flux density on short‐term time scales. Changes in LAI explained 84% and 77% of the variation found in GPP and NEE, respectively. Variations in R_eco were mainly controlled by canopy photosynthesis. After each grazing event, the reduction in LAI affected negatively the NEE.     

The annual carbon budget of a French pine forest (Pinus pinaster) following harvest

Eddy covariance measurements of net ecosystem exchange (NEE) of carbon dioxide and sensible and latent heat have operated since clear felling of a 50‐year old maritime pine stand in Les Landes, in Southwestern France. Turbulent fluxes from the closed‐path system are computed via different methodologies, including those recommended from EUROFLUX (Adv. Ecol. Res. 30 (2000) 113; Agric. Forest Meteorol. 107 (2001a, b) 43 and 71), and sensitivity analysis demonstrates the merit of post‐processing for accurate flux calculation. Footprint modeling, energy balance closure, and empirical modeling corroborate the eddy flux measurements, indicating best reliability in the daytime. The ecosystem, a net source of atmospheric CO₂, is capable of fixing carbon during fair weather during any season due to the abundance of re‐growing species (mostly grass), formerly from the understorey. Annual carbon loss of 200–340 g m^?2 depends on the period chosen, with inter‐annual variability evident during the 18‐month measurement period and apparently related to available light. Empirical models, with weekly photosynthetic parameters corresponding to seasonal vegetation and respiration depending on soil temperature, fit the data well and allow partitioning of annual NEE into GPP and TER components. Comparison with a similar nearby mature forest (Agric. Forest Meteorol. 108 (2001) 183) indicates that clear‐cutting reduces GPP by two thirds but TER by only one third, transforming a strong forest sink into a source of CO₂. Likewise, the loss of 50% of evapotranspiration (by the trees) leads to increased temperatures and thus reduced net radiation (by one third), and a 50% increase in sensible heat loss by the clear cut.     

Environmental and Vegetation Drivers of Seasonal CO2 Fluxes in a Sub-arctic Forest–Mire Ecotone

Unravelling the role of structural and environmental drivers of gross primary productivity (GPP) and ecosystem respiration (R _eco) in highly heterogeneous tundra is a major challenge for the upscaling of chamber-based CO₂ fluxes in Arctic landscapes. In a mountain birch woodland-mire ecotone, we investigated the role of LAI (and NDVI), environmental factors (microclimate, soil moisture), and microsite type across tundra shrub plots (wet hummocks, dry hummocks, dry hollows) and lichen hummocks, in controlling net ecosystem CO₂ exchange (NEE). During a growing season, we measured NEE fluxes continuously, with closed dynamic chambers, and performed multiple fits (one for each 3-day period) of a simple light and temperature response model to hourly NEE data. Tundra shrub plots were largely CO₂ sinks, as opposed to lichen plots, although fluxes were highly variable within microsite type. For tundra shrub plots, microsite type did not influence photosynthetic parameters but it affected basal (that is, temperature-normalized) ecosystem respiration (R ₀). PAR-normalized photosynthesis (P ₆₀₀) increased with air temperature and declined with increasing vapor pressure deficit. R ₀ declined with soil moisture and showed an apparent increase with temperature, which may underlie a tight link between GPP and R _eco. NDVI was a good proxy for LAI, maximum P ₆₀₀ and maximum R ₀ of shrub plots. Cumulative CO₂ fluxes were strongly correlated with LAI (NDVI) but we observed a comparatively low GPP/LAI in dry hummocks. Our results broadly agree with the reported functional convergence across tundra vegetation, but here we show that the role of decreased productivity in transition zones and the influence of temperature and water balance on seasonal CO₂ fluxes in sub-Arctic forest–mire ecotones cannot be overlooked.     

Paired comparisons of carbon exchange between undisturbed and regenerating stands in four managed forests in Europe

The effects of harvest on European forest net ecosystem exchange (NEE) of carbon and its photosynthetic and respiratory components (GPP (gross primary production) and TER (total ecosystem respiration)) were examined by comparing four pairs of mature/harvested sites in Europe via a combination of eddy covariance measurements and empirical modeling. Three of the comparisons represented high coniferous forestry (spruce in Britain, and pines in Finland and France), while a coppice‐with‐standard oak plantation was examined in Italy. While every comparison revealed that harvesting converted a mature forest carbon sink into a carbon source of similar magnitude, the mechanisms by which this occurred were very different according to species or management practice. In Britain, Finland, and France the annual sink (source) strength for mature (clear‐cut) stands was estimated at 496 (112), 138 (239), and 222 (225) g C m^?2, respectively, with 381 (427) g C m^?2 for the mature (coppiced) stand in Italy. In all three cases of high forestry in Britain, Finland, and France, clear‐cutting crippled the photosynthetic capacity of the ecosystem – with mature (clear‐cut) GPP of 1970 (988), 1010 (363), and 1600 (602) g C m^?2– and also reduced ecosystem respiration to a lesser degree – TER of 1385 (1100), 839 (603), and 1415 (878) g C m^?2, respectively. By contrast, harvesting of the coppice oak system provoked a burst in respiration – with mature (clear‐cut) TER estimated at 1160 (2220) gC m^?2– which endured for the 3 years sampled postharvest. The harvest disturbance also reduced GPP in the coppice system – with mature (clear‐cut) GPP of 1600 (1420) g C m^?2– but to a lesser extent than in the coniferous forests, and with near‐complete recovery within a few years. Understanding the effects of harvest on the carbon balance of European forest systems is a necessary step towards characterizing carbon exchange for timberlands on large scales.