基于日变化分析的卫星遥感海表温度重构研究

发布时间：2018-03-10 02:38

本文选题：日变化　切入点：海表温度　出处：《浙江大学》2016年博士论文　论文类型：学位论文

【摘要】：海表温度(Sea Surface Temperature,SST)是影响地球系统物理、化学和生物过程的一个重要物理量。当前,通过卫星遥感已能获得大量SST。但对于单一卫星遥感器,由于其轨道、平台、传感器波段设置及大气环境等的限制,往往只能瞬时观测有限的空间,即遥感产品存在观测时间的不一致,空间上有缺失等问题,给实际应用带来了困难。为此,本论文开展了基于日变化分析的卫星遥感SST重构研究。首先,基于卫星遥感和数值模型分析了研究区(115-130°E,22-42°N) SST的日变化特征,从物理机制上考虑海气热交换、海水混合等作用,选择一维海洋混合层模型(General Ocean Turbulence Model, GOTM)建立了不同气象条件下海温随深度和地方时变化的模型,将多源多时相的卫星遥感反演海表温度归一化到同一测量深度和时间基准；其次,在分析归一化SST空间覆盖分布特征的基础上,将每个时刻的研究区分为观测区和缺测区,观测区采用Markov估计进行多源数据融合,缺测区则根据日内尺度海表温度的时空相关模型,应用最优插值方法进行重构,最终生成时空基准统一的海表温度遥感分析产品。得到以下主要结论：(1)基于MODIS、AMSR-E和MTSAT的遥感观测,分析确定研究区域海温存在大幅度的日变化(最大振幅可达5。C以上)且主要发生在春夏季。SST日变化主要受风场和太阳短波辐射控制,风速越小,日照越强,日变化幅度越大。此外,海水的浑浊度也会促进SST日变化的幅度。(2)通过MTSAT观测的SST日变化比对验证两个当前常用的经验模型(CG03和ASM)和GOTM在本研究模拟的日变化能力,发现GOTM可以较好的模拟SST的日变化过程,平均偏差约为～-0.010C,均方根误差在0.3-0.50C。(3)多源海表温度的归一化还能有效填补卫星轨道缝隙,克服云雨等不利天气条件对卫星遥感的影响,月平均综合覆盖率能提高到75%左右。(4)通过对研究区内SST的时空相关模型分析,确定时间相关尺度为分析时刻前后2天、前后3小时,纬向的相关长度取～85 km,经向的相关长度取-100km。对2007年逐时的融合结果进行验证,总体偏差约-0.14-(2,均方根误差在0.570C,与卫星观测的精度相当。因此,本文的重构产品在提取快速变化的海洋特征(比如海洋内波)、目标信号,提高海-气通量的估算的精度等方面有应用价值,进一步也可输入到海洋模式作为初始场,发挥更大的作用。本文研究的主要创新点：(1)在卫星遥感SST日变化分析的基础上,从影响SST日变化的海气热交换、海洋混合等物理机制出发,利用GOTM模式建立了SST日变化模型,将多时相卫星遥感SST归一化到同一时刻基准。(2)利用Markov估计对归一化到同一时相的多源卫星共同覆盖区进行了融合,利用最优插值法对卫星未能覆盖区进行了重构,在空间上实现了对缺测区的填补。(3)基于(1)和(2)两个创新点,根据SST时空变化规律,可以生成时空基准统一的、任一时间与空间的卫星重构SST系列产品,为其他多源、多时相卫星资料的时间与空间的无缝重构提供了重要技术支撑。
[Abstract]:The sea surface temperature (Sea Surface Temperature, SST) is the influence of physics of the earth system, as an important physical quantity of chemical and biological processes. At present, the satellite remote sensing has been able to get a lot of SST. but for single satellite remote sensor, because of its orbit, platform, sensor band set and atmospheric environment and other constraints, often only instantaneous observation limited space, remote sensing products are inconsistent with the observation time, there are issues such as lack of space, bring difficulties to the practical application. Therefore, this paper carried out the research of satellite remote sensing SST reconstruction based on the analysis of the diurnal variation. Firstly, satellite remote sensing and numerical model based on analysis of the study area (115-130 degrees E, 22-42 degrees N) diurnal variation characteristics of SST, the physical mechanism of the air sea heat exchange, seawater mixing effect, selection of one-dimensional ocean mixed layer model (General Ocean Turbulence Model, GOTM) was established under different meteorological conditions The sea temperature variation with depth and time model of multi-source multi temporal satellite remote sensing sea surface temperature normalized to the same depth and time reference; secondly, based on the analysis of the normalized SST space coverage distribution feature, will study each time divided into observation area and missing area, observation area estimated by Markov the multi-source data fusion, data missing area according to the temporal and spatial correlation model on scale sea surface temperature, the application of optimal interpolation method for reconstruction of sea surface temperature, remote sensing analysis products generated space-time datum unification. The main conclusions are as follows: (1) based on MODIS, AMSR-E and MTSAT remote sensing observation, analysis on determination of regional SST there is a substantial change (above the maximum amplitude of up to 5.C) and mainly occurred in spring and summer.SST change is mainly affected by wind and solar radiation, wind speed, sunshine is strong, On the bigger the change. In addition, the turbidity will promote the diurnal changes of SST amplitude. (2) the diurnal variation of SST ratio on the MTSAT observation of the validation of two empirical models commonly used current (CG03 and ASM) on the changes of ability of simulation and GOTM in this study, found that the diurnal variation of SST GOTM can better simulate in the process, the average deviation is about -0.010C, the RMS error in 0.3-0.50C. (3) multi-source sea surface temperature normalization can effectively fill the gap of satellite orbit, clouds and other adverse weather conditions to overcome the influence of satellite remote sensing, the average monthly comprehensive coverage rate can be increased to about 75%. (4) through the space-time correlation model of the research area of SST analysis, to determine the time scale for the time related analysis before and after 2 days, after 3 hours, the correlation length of weft from - 85 km to verify the results of the 2007 fusion by related to the direction of the length of the -100km., the overall deviation is about -0.14- (2, root mean square error in the 0.570C, and the precision of satellite observations. Therefore, the reconstruction of marine products in the fast changing (such as feature extraction, target signal, internal waves) has application value to improve the estimation accuracy of the air sea flux, further can be input to the ocean model as the initial field, to play a greater role. The main innovations of this paper: (1) based on the analysis of satellite remote sensing SST diurnal variations of air sea heat exchange, the diurnal variation of SST effect, the physical mechanism of ocean mixing, established the diurnal changes of SST model with GOTM model, the multi temporal satellite remote sensing normalized to the same SST a moment of reference. (2) of the normalized to the same phase of multi satellite joint coverage area were fused using Markov estimation, the reconstruction of the satellite failed to cover the area using optimal interpolation method in space, realize the missing fill area. (3) based on (1) and (2) two innovation points, according to the temporal and spatial variation rule of SST, we can generate spatio-temporal datum unified, any time and space satellite reconstruction SST series products, providing an important technical support for seamless reconstruction of other multi-source and multi temporal satellite data.

【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P731.11;P714.1

【相似文献】