【作者】 郭建侠；

【导师】 卞林根；

【作者基本信息】 南京信息工程大学 ， 气象学， 2006， 博士

【摘要】 本文围绕涡旋相关资料的处理方法和应用目标,利用中国气象局固城生态站2004年6月11日至10月20日玉米全生育期测量资料,详细分析了快速变化的农作物覆盖下垫面涡旋相关资料的预处理和质量控制方法,系统研究了湍流及湍流通量的统计特征,分析了地表能量平衡状况,估算了华北玉米生育期对大气的碳汇,并对比分析了模式中常用的几种通量参数化方案。主要结论如下:1、平均时间、坐标旋转与通量修正是资料预处理与质量控制中的重要问题,本文研究表明:（1）积分时间尺度在不同长度序列和不同变量之间变化较大,不适宜作为确定涡旋相关资料平均时间的依据。（2）平均时间低于30分钟有通量低估现象,高于l小时日变化型出现较大不稳定,30分钟到l小时之间通量无明显增加或减小。合理的平均时间应不低于30分钟,同时不高于1小时。（3）对于快速生长变化的下垫面,根据作物生长阶段,分时段平面拟合旋转坐标,能有效地订正资料中的倾斜和扰动。（4）经典的WPI,修正方法对水汽的作用估计过高,可能存在对潜热通量和CO2通量修正过量问题,Liu修正方法理论上较为合理,能够有效修正密度影响。（5）坐标旋转和通量修正使能量闭合率上升2.6%。2、湍流和湍流通量的基本统计特征分析表明:（1）水平风的平均湍流强度为0.2,垂直风的平均湍流强度为O.l。（2）风速和标量的方差、协方差均符合相似规律。方差相似函数与文献中平原地区的相似函数比较接近,但稳定条件下温度和水汽相似函数差.别较大。协方差相似函数表明动量输送在中性条件下效率最高,感热输送随稳定性绝对值增大而增大,潜热始终从地面向大气输送,输送效率随稳定度增大而减小。（3）三维速度谱与Kaimal et al.（1972）的经典谱具有大致相同的特征,但温度谱的惯性副区出现较宽收缩带。小湍涡对动量的输送比感热输送更为有效。3、地表能量平衡分析表明:（1）白天能量闭合率平均为78.3%,夜间平均仅有12.6%,整个生育期能量闭合率平均为77.9%。自天能量闭合率有明显的日变化和季节变化规律,下午能量闭合率普遍高于上午,随玉米生长和季节演变能量闭合率逐渐下降。（2）玉米整个生育期中,潜热一直是能量消耗的主要形式,占净辐射比例为32～54%,感热占净辐射比例14～37%,土壤热通量占净辐射比例11～l 8%,剩余项占净辐射比例4～28%,植被冠层的热储存是能量不闭合的主要原因。（3）讨论了能量不平衡的其他原因,认为源区差异对本文的能量不平衡有一定的贡献,能量各分量的位相差异可能是导致能量不平衡的原因之一。（4）地表反照率变化范围为0.07～O.18,出现概率最大为O.12。正午波文比在玉米播种和收获时期高达3.6,在玉米生长阶段维持在0.5左右。4、CO₂浓度和通量的分析表明:（1）人类播种和焚烧秸秆等活动对CO₂,浓度有更多还原

【Abstract】 For best understanding the turbulent exchange of energy and mass betweenatmosphere and land surface, eddy covariance measurements were conductedover a typical agriculture field on the North China Plain. In this paper,we focus on corn growing period （from 11 June to 20 October, 2004）, in whichthe characters of land cover vary rapidly with corn growing. Our study includethe data process and quality control technique, the statistical charactersof turbulent transfer and fluxes, the energy balance of surface, the netexchange of CO₂ between corn surface and atmosphere, and the fluxparameterization schemes near surface during corn growing. Main concludesare as following:1. Averaging, coordinate rotation and fluxes correction are importantissues in data processing procedure. Our study indicates that （1） Integraltime scale does not fit to determine the averaging time period because ofits variation with time series length and measurement variables. Fluxesincrease with increasing of average period from 2 min to 30 min, while ithas not visible increase or decrease from 30 min to 1 hour. When average periodlonger than 1 hour, fluxes appear much unstable daily pattern, and withoutsystematical increase or decrease with prolonging averaged period. So, wesuggest that the reasonable averaging period for our data should be neithershorter than 30 min, nor longer than 1 hour. （2） Planar fit coordinaterotation, which is recommended by the fluxes community recently, is performwell when applied to the divided periods according to corn growth rather thanto the whole raw data. （3） Density correction of latent heat flux and CO₂flux are necessary for our data, because they enhance the energy closure andremove the falsity of CO₂ uptake. We commend Liu method （Liu, 2005） insteadof WPL method （Webb et al.,1980） that was generally used in recent decades,because WPL method magnify the water vapor’s variation. （4） Both coordinaterotation and fluxes density correction can increase the energy closure, totalas 2.6%, indicate that data processing might become energy sink if withoutproper process procedure. Data process is also important to enhance theestimation accuracy of net exchange of CO₂ between ecosystem and atmosphere.2. The analyses of statistical characters of turbulent transfer showthat: （1） Averaged turbulent intensity are 0.2 and 0.1 for horizontal andvertical wind fluctuation, respectively. （2） All of the variance andcovariance of wind and scalars obey the Monin-Obukhov similarity theory. Thevariance similarity functions from our data are close to that of plain areafrom literature expect for temperature and water vapor in stable condition.Thecovariance similarity functions appear much difference from literature.They show that momentum transfer is more effectively in nature conditionrather than stable and unstable condition. Sensible heat transfer will bestronger with the increase of absolute value of stability. Latent heat fluxis always from surface to atmosphere, and transfer efficiency decrease withthe increase of stability. （3） Spectral of wind components are similar tothose in Kaimal et al （1972） if we exchange u and v direction. But temperaturespectral appears widedegradation zone in the inertial subrange. Thecospectrum of momentum and sensible heatare quite different from literature.They do not obey the -4/3 law of slope in inertial subrange, and predictedthat larger eddies are relatively more efficient in transporting heat thanmomentum in vertical direction.3.Examing of surface energy balance indicates that: （1） The averagedenergy closure is in the order of 78.3%, 12.6% and 77.9% for day time, nighttime and the whole data sets, respectively. Energy closure is alwaysincreased from morning to afternoon during day. While, it is usuallydecreased with corn growth, at the range of 93% to 67% from mid June to earlyOctober. （2） Latent heat is always the main energy consume form of radiationenergy over the whole experiment. It takes 32-54% of net radiation fromseeding period to harvesting period. Meanwhile, sensible heat takes 14-37%,soil heat takes 11-18% and residual energy takes 4-28%, respectively. It isevident that the heat storage of corn canopy contributes to the energyimbalance at a large degree. （3） Discussing of the other causes to energyimbalance, we agree that footprint difference may be a driver to energyimbalance in our experiment. In addition, we consider the lag of time phaseof energy component to net radiation might be one of the reasons for ill energyclosure. （4） Surface albedo varies at the range of 0.07-0.18 during ourexperiment, with the maximum probability of 0.12. Bowen ratio at noon is aslarge as 3.6 during the seeding and harvesting period, and keep in 0.5 duringmain growth period.4.Investigating of net CO2 exchange between corn surface and atmospheregets to following results:（1） Human’s field activities （e.g. tillage andfiring the residual） influence on CO2 concentration rather than on CO2 flux.（2） The concentration of CO2 holds diurnal cycle during the whole period ofmeasurement, but CO2 flux has not visible diurnal cycle until the corn canopydeveloped to some degree. The peak （vale） value of CO2 concentration usuallyappears at sunrise （sunset）, related to the altering direction of CO2 flux.（3） Corn surface serve as carbon sink to atmosphere from young canopy stageto mature stage, and carbon source before it. The maximum daily carbon takingof corn canopy occurs at the abloom stage, as -10.1 gC m-2day-1, followedby spin-ripe stage （-9 gC m-2day-1） and jointing stage （-7.1 gC m-2day-1）.Over the whole period, the net exchange of CO2 flux is -176 gC m-2 atmeasurement height. We noted that the different methods mentioned in chapter2 might bringas large as 160% bias to the total CO2 sink in our experiment.（4） The most proper net radiation condition for corn canopy assimilate CO2is 470 Wm-2. Both high temperature and high moisture is good for corn plantassimilation of CO2 from atmosphere, especially at noon, and more effectivelyfor mature canopy than young canopy.5. The research of flux parameterization shows that: （1） For estimatingwind speed at the height of 16 m during our experiment, it is feasible thataerodynamic roughness length z0 is fixed to 0.27 m through the whole period,and the zero displacement height d is fixed to 0 （1.8 m） before （after） midAugust. （2） By comparing measurement fluxes of momentum, sensible heat andlatent heat with that from parameterization schemes, we find that the schemessuch as Businger（1971）, Dyer（1974）, Beljaars （1991）, which employ z/L asstability variable, perform not better than those of using bulk Richardsonnumber RiB as stability variable, such as Louis（1982） and Uno（1995）. Amongthese schemes, Uno（1995） scheme is optimum for our experiment. Results stresson that differentiating ZT and zq from z0 is very important for the simulationof corn surface fluxes.更多还原