四类求解地表湍流热通量方法在中国西部地区的应用及其误差分析

发布时间：2018-01-20 16:51

本文关键词： 地表湍流热通量误差中国西部涡动相关法变分法 Noah-MP陆面模式集合卡尔曼滤波　出处：《兰州大学》2017年博士论文　论文类型：学位论文

【摘要】：地表湍流热通量是陆面过程中关键物理量之一,对它准确计算是大气、水文和农业等学科必需的。伴随着天气和气候模式物理过程的不断改进以及时空分辨率的快速提高,为模式提供高精度的地表湍流热通量下边界显得尤为重要。根据观测资料和下垫面的不同,目前存在多种计算地表湍流热通量的方法,影响计算精度的不确定因素也不尽同。中国西部地区下垫面种类复杂,地形起伏较大,各类方法在该地区应用时的具体表现可能不尽相同。因此,本文从方法论的角度出发,评估四类代表性估算地表湍流热通量方案在西部地区的性能,它们是涡动相关法(EC)、变分法、陆面模式Noah-MP的模拟方法和集合卡尔曼滤波(EnKF)数据同化方法;重点分析可能存在的误差源并针对部分方案提出改进措施,以期提升不同方法估算地表湍流热通量的精度。主要结果如下:(1)针对西北地区一类特殊的非均匀绿洲下垫面上单站点EC法求解地表湍流热通量存在较大的不确定性问题,论文进行了定量误差源分析。首先,利用单点观测资料计算EC湍流热通量并探究低估原因。单站点EC湍流热通量的误差部分来源于绿洲上空较大的平均垂直速度(VA)能量输送以及下沉气流,尤其在中午前后,其中VA感热输送(潜热输送)对感热通量(潜热通量)日均方根平均值贡献约为16.5%(7.4%)。其次,为了弥补上述单一站点资料分析结果的代表性不足问题,借助WRF-LES大涡模拟手段研究理想带状非均匀绿洲沙漠下垫面上不同位置处EC热通量的空间代表性。绿洲不同位置EC观测对区域平均热通量的代表性空间差异较大,原因各点受下垫面上空湍流有组织结构(TOS)和次级环流(SC)的影响不同;由TOS与SC引起的热通量与区域平均热通量的比值在边界层中上层较大,从而单站点EC热通量观测很难代表区域平均值,部分位置处甚至出现严重低估。(2)针对西部地区部分下垫面(绿洲和高原冻土)存在较大能量存储进而影响变分法求解精度的问题,论文在变分法中提出两种能量存储方案:第一种方法为利用观测资料计算,第二种方法为分段线性拟合能量存储。最后,利用地处甘肃省张掖绿洲上大满站(有植被能量存储和灌溉造成的较大土壤热存储)和青藏高原上安多站(土壤冻融会伴随大量热量吸收/排放)的资料,对所提两类方案进行验证,并与EC观测的热通量进行比较,结果表明:两种方法都提高了热通量估计精度;相对而言,第一种方法结果较好,尤其在中午或边界层强不稳定条件下。此外,在没有足够观测资料计算这些能量存储时,还可以把已有能量存储项“移植”到同一地区其他观测时段用于热通量的估算,从而增大了方法的适用范围。(3)利用甘肃张掖绿洲大满站观测资料,利用陆面模式Noah-MP求解地表热通量,并仔细分析误差源。若只采用默认参数化方案,模式能够较好地模拟晴天净辐射、感热通量和潜热通量,但在晴天中午前后模拟的峰值明显低于观测值,并且高估土壤蒸发比和低估植物蒸腾。其次,通过相关物理过程分析和不同的参数化方案的替换试验探寻误差源。采用Cai11地表反照率方案和/或Jarvis气孔阻抗方案替代默认方案可以部分解决模拟的净辐射通量和潜热通量偏低的问题,同时也使土壤蒸发和植物蒸腾的分配比例更为合理,地表辐射温度更接近观测值;采用Alpha方案可解决原有SZ09方案土壤蒸发过大问题,减少土壤蒸发的比例。(4)利用地处甘肃省榆中县原生植被下垫面上的SACOL站观测资料,开展EnKF同化试验来改善地表热通量的计算。相对于没有进行同化的陆面模式模拟结果,同化表层土壤湿度观测可以有效地克服感热通量模拟偏高、潜热通量模拟偏低的问题,其中在9:00-12:00 LT改进最大。最后,采用敏感试验方法,探究观测误差和背景场误差对EnKF估算的影响:当观测误差减小(增大),感热通量和潜热通量估算值均方根误差也减小(增大);EnKF估算值随着背景场方差放大系数的增大而增大。因此,高质量的观测资料和能真实反映背景场误差协方差的样本是提升地表热通量估算精度的关键。总之,上述四类求解地表湍流热通量的方法各有优缺点,虽然在西部地区能较好地估算地表湍流热通量,但是由于所用观测资料来源和性质不同,难于对它们估算结果进行优劣比较,在实际中可根据观测资料性质和不同下垫面特点进行针对性地选用。
[Abstract]:The surface turbulent heat flux is one of the key physical quantity in the land surface process, it is necessary to accurately calculate the atmospheric, hydrological and agricultural science. With the rapid increase of continuous improvement of weather and climate model and spatial resolution, it is especially important to provide high precision surface turbulent heat flux boundary appears as a model. According to the observation data and the different underlying surfaces, there many methods to calculate surface turbulent heat fluxes at present, influence the precision of calculating uncertainties are not the same. The western region China surface species complex, undulating terrain, the specific performance of various methods in the area of the application may not be the same. Therefore, this article from the methods on the basis of evaluation of four representative types of performance estimation of surface turbulent heat flux in the western region, which is the eddy correlation method (EC), variational method, land surface model Noah-MP model Quasi Calman method and set filter (EnKF) data assimilation method; focus on the analysis of error sources may exist and put forward the improvement measures for the part of the program, in order to improve the different methods of estimation of surface turbulent heat flux precision. The main results are as follows: (1) in the Northwest region of a special kind of non uniform underlying surface on a single site in Oasis EC method for solving the surface turbulent heat flux is a big uncertainty problem, the quantitative analysis of error sources. Firstly, calculate EC turbulent heat flux and explore the reasons of underestimation by the single point observation data. The error sources of single site EC turbulent heat fluxes over the oasis in larger average vertical velocity (VA) of energy transfer and sinking air, especially at noon, which VA sensible heat fluxes (latent heat transfer) of sensible heat fluxes (latent heat flux) the average root mean square average contribution is about 16.5% (7.4%). Secondly, in order to make up for the single A lack of representation of the site data analysis results, with the help of WRF-LES large eddy simulation method to study the non ideal zonal spatial representation of different positions of EC heat flux uniformity of oasis desert surfaces. The representative of the regional differences in the average heat flux space oasis EC different position observation points by reason of large surface over turbulent structure (TOS) and secondary circulation (SC) the effect of different heat flux ratio; the average heat flux and area caused by TOS and SC in the boundary layer in the upper is larger, so the single site EC heat flux is difficult to represent the Qu Yuping mean, part of position or even seriously underestimated. (2) according to the the western part of the surface area (oasis and permafrost) there are large energy storage and variational method to solve the accuracy problem, the two energy storage scheme is proposed by the first method for using observation The data calculation, second kinds of methods for piecewise linear fitting energy storage. Finally, the Gansu province is located in Zhangye oasis on the full station (with larger energy storage and heat storage of soil vegetation caused by irrigation) and Tibetan Plateau (Amdo station soil freezing and thawing with lots of heat absorption / emission) data to validate the two kind of scheme, and heat flux and EC observation were compared, the results show that the two methods can improve the estimation accuracy of heat flux; relatively good results, the first method, especially at noon or in the boundary layer instability conditions. In addition, there is not enough data for calculating the energy storage, but also the existing energy storage "transplant" to the same area of other observation period for estimating heat flux, thereby increasing the scope of the method. (3) by Gansu Zhangye oasis over the station observation data of land use planning Type Noah-MP to solve the surface heat flux, and careful analysis of error sources. If only using the default parameter scheme model can better simulate the net radiation, sensible heat flux and latent heat flux, but the peak at noon before and after the simulation was significantly lower than the observed value, and overestimate the soil evaporation and plant transpiration than underestimated. Secondly, explore through the test error source replacement parameterization scheme analysis of relevant physical processes and different surface albedo. The Cai11 scheme and / or Jarvis stomatal resistance alternative default scheme can partly solve the simulation of net radiation flux and the latent heat flux is low, but also the proportion of soil evaporation and plant transpiration is more reasonable. The surface radiation temperature is closer to the observation; using Alpha scheme can solve the original SZ09 scheme of soil evaporation is too large, reduce soil evaporation ratio. (4) the Gansu province is located in Yu The observation data of earth's surface SACOL station in the county of native vegetation, carry out calculation EnKF Assimilation Experiment to improve the surface heat flux. Compared with the land surface model without assimilation of simulation results, assimilation of surface soil moisture observations can effectively overcome the simulated sensible heat flux is high, the problem of low heat flux simulation, the biggest improvement in 9:00-12:00 LT. Finally, the sensitive test method, explore the observation error and background field error on EnKF estimation when the observation error decreases (increases), sensible heat flux and latent heat flux estimates of root mean square error also reduced (increased); EnKF value increases with the increase of background estimation field variance amplification factor. Therefore, the observation data of high quality and can reflect the background error covariance of the sample is to enhance the surface heat flux estimation accuracy of the key. In short, the above four kinds of solving methods of surface turbulent heat flux Each has advantages and disadvantages, although better estimation of surface turbulent heat fluxes in the western region, but due to the source of the observed data and the different nature of them difficult to estimate the results of comparative advantages and disadvantages in practice, according to the nature of observations and different surface features are targeted to choose.

【学位授予单位】：兰州大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P425.2

【参考文献】