地表水热通量模拟尺度效应研究

发布时间：2018-05-16 14:51

本文选题：水热通量 + 尺度效应　；参考：《中国科学院大学(中国科学院遥感与数字地球研究所)》2017年博士论文

【摘要】：土壤和植被与大气间的水热通量交换是陆面大气动力和热力过程的基础,决定着全球能量平衡及水分循环,在气候变化适应策略、生态保护和农业水资源管理等领域的实际应用中也占有重要地位。目前,用于模拟地表与大气间水热通量交换的遥感模型和陆面过程模型在大尺度上应用时仍存在较大的不确定性。主要表现为:首先,基于均匀地表条件发展起来的模型参数及参数化方案在异质性地表条件下的适用性存在一定的不确定性;其次,在进行地表水热通量模拟时,中高空间分辨率遥感模型逐像元的模拟方式以及粗网格陆面过程模型通常采用的网格/次网格均质假设均不能充分表达地表的异质性,在模型非线性结构的影响下便会给地表水热通量真实空间异质性的表达以及像元尺度、网格\次网格尺度水热通量均值的模拟带来一定的不确定性;最后,地面水热通量观测源区空间范围与模型模拟单元空间范围之间往往存在较大差异,使得利用地面水热通量观测的直接模型精度验证存在一定的不确定性。这些不确定性形成了地表水热通量模拟中尺度效应的核心问题。本文分析了遥感高分辨率尺度(100 m)至陆面过程模型粗网格尺度(10km~50 km)地表水热通量模拟中的尺度效应,旨在提高对地表水热通量模拟中尺度依赖的认识,进而为陆面过程模型粗网格尺度下参数化方案及验证方法的改进提供理论和依据。主要研究内容及结论如下:1、从分析本论文所采用的遥感蒸散发模型ETMonitor出发,揭示不同地表土壤水分SSM和植被特征LAI条件下控制蒸散发过程的主要因子。该部分针对不同土壤水分和植被状态分布特征,发展了在同时考虑多种输入变量的高维空间中对地表蒸散发模型进行全局敏感性分析的方法。研究结果显示:蒸散发模拟对叶面积指数LAI和地表土壤水分SSM的敏感性在稀疏植被覆盖且土壤水分较低时更加明显;蒸散发模拟对空气温度和空气相对湿度的敏感性在不同的叶面积指数LAI和地表土壤水分SSM异质性分布条件下均较高;蒸散发模拟对风速的敏感性在高植被覆盖且地表土壤水分SSM较高时更加明显;蒸散发模拟对下行短波辐射的敏感性在叶面积指数lai和地表土壤水分ssm均较高时稍弱。2、中高空间分辨率的混合像元存在条件下,遥感模型模拟地表水热通量时逐像元的模拟往往不能充分表达地表的复杂异质性,在水热通量遥感模型非线性结构作用下便会导致水热通量模拟的尺度效应问题,本文对遥感模型水热通量模拟的尺度效应问题进行分析。该部分研究利用基于二维离散小波变换的图像多尺度分析方法分析了不同尺度的地表植被、土壤水分和水热通量交换异质性特征,并基于etmonitor模型框架,通过比较逐级空间尺度(90m、180m、360m、720m至1440m)下“参数聚合”与“通量聚合”的差异,分析了中高分辨率遥感逐像元蒸散发估算的“聚合误差”。结果显示:在所选研究区研究时段内,地表植被、土壤水分及蒸散发的异质性可存在于不同的空间尺度上,且各自的空间分布特征不尽相同,与蒸散发空间异质性分布高度相关的高空间分辨率地表特征数据的利用对蒸散发的高精度模拟至关重要;当前研究中,因与et空间分布特征高度相关的地表特征(ssm)及et自身的异质性信息较少出现在较小的空间尺度上,高空间分辨率(90m)输入数据逐级聚合到千米级空间分辨率的过程中,在研究区内多数像元的像元尺度聚合误差并不明显,仅在地表土壤水分和植被状态差异较大的黑河沿岸、绿洲内部作物与城镇交界处以及研究区东南部的滴灌区周围等区域出现了较大的聚合误差;因不同情景下蒸散发模拟与地表状态的非线性关系存在差异,像元尺度聚合误差的分布与地表状态异质性信息密度的分布不完全一致。3、陆面过程模型模拟地表水热通量时通常采用网格\次网格均质假设,对地表状态及气象条件异质性做简化处理,本文对该简化处理导致的水热通量模拟尺度效应问题进行了分析。该部分研究通过情景模拟和实例分析两种方式展开。情景模拟分析中,基于所模拟的模型次网格内多种ssm和lai的空间异质性分布情景(分别对应植被分布从稀到密,表层土壤水分从干到湿的情景)以及多种气象条件,分别分析了当忽略ssm和lai异质性时的单独及协同影响,并讨论了气象条件对蒸散发模型中蒸散发模拟与地表状态之间非线性及聚合误差的影响。情景模拟分析结果显示:在不同的地表特征异质性分布情景下,陆面过程模型次网格地表特征均质假设方案导致的蒸散发估算偏差表现不同,当ssm和lai为正偏态分布(干旱半干旱的稀疏植被覆盖区)时易出现较大的估算误差;多种气象条件下的统计结果显示,忽略ssm异质性比忽略lai异质性更易产生较大的聚合误差;忽略SSM和LAI的异质性在风速较高、空气温度较高、空气较干燥的气象背景下易产生低估,在风速较低、空气温度较低、空气较湿润的气象背景下易产生高估。在实例分析部分,以我国西北部黑河流域中游和上游为研究区,采用三种陆面过程模型模拟常用的网格尺度(10 km、25 km和50 km),以遥感模型ETMonitor的模拟结果为依托,分析了不同网格大小下陆面过程模型中采用的次网格内地表状态均质假设和网格内气象条件均质假设方案对模拟结果的影响。实例分析结果显示:对次网格及格网尺度水热通量估算,忽略同一次网格内地表特征异质性与忽略同一网格内气象条件异质性对子流域尺度平均值的估算影响不大,但可导致较大的次网格、网格尺度日均蒸散发估算误差,且较大的估算误差易出现在地表较干旱地区,因此基于地表覆盖类型次网格划分的陆面过程模型的应用效果依不同的应用目的及研究区而异;在研究区内忽略同一次网格内地表状态异质性和忽略同一网格内气象条件异质性导致的聚合误差多呈同向叠加形式,且二者交互作用在网格尺度较大时更加明显。
[Abstract]:The exchange of water and heat flux between the soil and the atmosphere is the basis of the dynamic and thermodynamic processes of the terrestrial atmosphere, which determines the global energy balance and water cycle. It also plays an important role in the practical applications of climate change adaptation strategies, ecological protection and agricultural water resources management. The exchange of remote sensing model and land surface process model still have great uncertainty when applying the land surface process model on large scale. The main manifestations are: first, the model parameters and parameterization schemes based on the uniform surface conditions have certain uncertainty in the applicability of the heterogeneous surface conditions; secondly, when the surface water heat flux is simulated, the surface water heat flux is simulated. The simulation of high spatial resolution remote sensing model and the assumption that the grid / subgrid homogenization assumption usually used in rough grid land surface process model can not fully express the heterogeneity of the surface. Under the influence of the model nonlinear structure, the expression of the real spatial heterogeneity of the surface water heat flux and the pixel scale, the grid subnet will be given. There is a certain uncertainty in the simulation of the mean of the mean of the thermal flux of the grid scale. Finally, there is a large difference between the spatial range of the surface water heat flux and the space range of the model simulation unit, which makes the accuracy of the direct model of the surface water heat flux observed. The core problem of mesoscale effect in the surface water heat flux is simulated. This paper analyzes the scale effect of the remote sensing high resolution scale (100 m) to the surface water heat flux simulation of the rough grid scale (10km~50 km) of the land surface process model, which aims to improve the understanding of the scale dependence of the surface water heat flux simulation, and then to the rough grid scale of the land surface process model. The main research contents and conclusions are as follows: 1, from the analysis of the remote sensing evapotranspiration model ETMonitor used in this paper, the main factors to control the evapotranspiration under different soil moisture SSM and vegetation feature LAI are revealed. This part is aimed at different soil moisture and soil moisture. The results show that the sensitivity of Evapotranspiration to leaf area index LAI and surface soil moisture SSM is more obvious when it is covered with sparse vegetation and the soil moisture is low; The sensitivity of the emission simulation to air temperature and air relative humidity was higher under the different leaf area index LAI and the surface soil moisture SSM heterogeneity. The sensitivity of the evapotranspiration simulation to the wind speed was more obvious when the high vegetation was covered and the surface soil moisture SSM was higher; the sensitivity of the Evapotranspiration to the downlink short wave radiation was high. The leaf area index (LAI) and the surface soil moisture SSM are both slightly weaker and slightly.2, and under the presence of mixed pixels with high spatial resolution, the simulation of the surface water heat flux can not fully express the complex heterogeneity of the surface when the remote sensing model simulates the surface water heat flux, and it will lead to the hydrothermal flux model under the action of the nonlinear structure of the water heat flux remote sensing model. In this paper, the scale effect problem of water heat flux simulation in remote sensing model is analyzed in this paper. In this part, the multi-scale analysis method based on two-dimensional discrete wavelet transform is used to analyze the surface vegetation of different scales, the characteristics of the exchange of soil moisture and water heat flux, and based on the etmonitor model framework. After comparing the difference between "parameter aggregation" and "flux aggregation" in 90m, 180m, 360m, and 720m to 1440m, the "aggregation error" of the estimation of the medium and high resolution pixel steaming is analyzed. The results show that the heterogeneity of soil vegetation, soil moisture and evapotranspiration can exist in different period of study area. On the spatial scale, the spatial distribution characteristics are different, and the high spatial resolution surface feature data, which is related to the spatial heterogeneity of the evapotranspiration, is very important for the high precision simulation of the evapotranspiration. In the current study, the surface feature (SSM) and the heterogeneity of the ET itself are highly correlated with the spatial distribution characteristics of the ET. In the process of small spatial scale, high spatial resolution (90m) input data converge to kilometer spatial resolution, the pixel scale aggregation error of most pixels in the study area is not obvious, only in Heihe along the surface soil moisture and vegetation state, and the border between crops and towns in oasis. There are large aggregation errors in the area around the drip irrigation area in the southeastern part of the study area, and the nonlinear relationship between the evapotranspiration simulation and the surface state under different scenarios, the distribution of the pixel scale aggregation error and the distribution of the information density of the surface state heterogeneity is not exactly the same as.3, and the land surface process model simulates the surface water heat flow of the surface. The grid subgrid homogenization assumption is usually used to simplify the surface state and the meteorological condition heterogeneity. This paper analyzes the scale effect of the hydrothermal flux simulation caused by the simplified treatment. This part of the study is carried out through two ways of scenario simulation and example analysis. In the scenario simulation analysis, the simulated model is based on the simulated model. The spatial heterogeneity of multiple SSM and Lai in the subgrid (corresponding vegetation distribution from sparse to dense, surface soil moisture from dry to wet) and a variety of meteorological conditions were analyzed, respectively, to analyze the individual and synergistic effects of ignoring SSM and Lai heterogeneity, and discussed the simulation of evapotranspiration and the surface of the evapotranspiration model in the evapotranspiration model. The results of the situation simulation analysis show that the estimation deviation of evapotranspiration in the land surface process model is different, when SSM and Lai are positive partial distribution (arid and semi-arid sparse vegetation cover) under different surface characteristic heterogeneity distribution scenarios. The statistical results under various weather conditions show that neglecting SSM heterogeneity is more likely to produce larger aggregation errors than neglecting Lai heterogeneity; ignoring the heterogeneity of SSM and LAI is higher in wind speed, higher air temperature and less air temperature, lower air temperature and air temperature, air temperature is lower, air temperature is lower, air temperature is lower, air temperature is lower, air temperature is lower, air temperature is lower, air temperature is lower, air temperature is low In an example analysis, three land surface process models are used to simulate the common grid scale (10 km, 25 km and 50 km). Based on the simulation results of remote sensing model ETMonitor, the process models for different grid size land surface are analyzed. The results show that the estimation of the water and heat flux of the grid and grid scale, neglecting the heterogeneity of the surface features in the same grid and neglecting the heterogeneity of the meteorological conditions in the same grid to the sub basin scale. The estimation of the average value has little influence, but it can lead to larger subgrid and the estimation error of the daily evapotranspiration on the grid scale, and the larger estimation error is easy to appear in the arid area. Therefore, the application effect of the land surface process model based on the subgrid submeshing of the surface cover type varies according to the different application purpose and the research area; in the study area, the application effect of the land surface process model is different. Neglecting the heterogeneity of the surface state in the same grid and ignoring the meteorological conditions heterogeneity in the same grid results in the same direction superposition, and the interaction between the two is more obvious when the grid scale is larger.

【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P422.4

【相似文献】