GAS: Generic Atmosphere Solutions for radar measurements

地表变形是一种主要的全球性灾害，可能由滑坡、地震和火山等自然过程或地下水、石油和煤炭开采等人为过程造成。在过去的二十年里，重复通过干涉合成孔径雷达（干涉合成孔径雷达）被广泛用于绘制地表变形图（Massonnet和Feigl，1998年）。然而，大气（特别是对流层水汽）的变化是使用干涉合成孔径雷达确定地表变形时不确定性的主要来源之一，而纠正这种误差是探测细微变形信号的关键。PI在干涉合成孔径雷达大气校正模型方面具有相当丰富的经验。事实上，PI一直处于干涉合成孔径雷达大气校正的前沿，开发了五种模型，以减少水汽对使用遥感水汽数据进行干涉合成孔径雷达测量的影响（例如全球定位系统、美国航天局中分辨率成像光谱仪和欧空局中分辨率成像光谱仪）：每个模型都能够将干涉合成孔径雷达导出的变形信号的精度从校正前的10 mm提高到校正后的5 mm（例如，Li等人，2009年，Li等人，2006 c，Li等人，2006 b，Li等人，2006 a，Li等人，2005年）。PI决心在这项研究中利用它。本研究的目的是：（i）开发基于大气模型的校正技术，使我们的校正模型在全球范围内适用;（ii）证明基于大气模型的校正技术在任何可能的情况下都易于应用：（a）使用单个干涉合成孔径雷达对，或（b）使用多个干涉合成孔径雷达对;开发干涉合成孔径雷达时间序列分析中的大气估计技术，以便在不对变形模型作任何假设或不要求大气数据的情况下提取变形信号。这对于无法访问任何大气模型的用户来说是可取的。方法学：所有上述校正模型都固有地受到数据可用性的限制，例如缺乏密集的GPS网络（对于基于GPS的模型）和对云的存在的敏感性（对于MODIS和MERIS）。由英国气象局开发的统一模型提供了对流层路径延迟的高空间分辨率（例如1公里）估计，全天候每天24小时覆盖全球，这对于干涉合成孔径雷达大气校正是非常理想的，本研究将研究该模型，以校正大气对单对和/或多对SAR干涉图的影响。现有的干涉合成孔径雷达时间序列分析技术通常采用先验形变模型或空时滤波技术来分离形变信号和大气效应，这显然不是最优的。基于大气信号的3个关键物理特征，提出了一种迭代方法，在不对形变模型做任何假设和对大气数据无任何要求的情况下，从多个干涉合成孔径雷达对中估计大气信号，然后提取形变信号。

Land surface deformation is a major worldwide hazard that can result from natural processes such as landslides, earthquakes, and volcanoes, or from anthropogenic processes including extraction of groundwater, oil and coal. Repeat-pass Interferometric Synthetic Aperture Radar (InSAR) has been widely used to map land surface deformation in the past two decades (Massonnet and Feigl, 1998). However, change in the atmosphere (especially tropospheric water vapour) is one of the major sources of uncertainty in determining surface deformation using InSAR, and correcting for such errors is key to the detection of subtle deformation signals. The PI has considerable experience in InSAR atmospheric correction models. Indeed, the PI has been at the forefront of InSAR atmospheric correction by developing five models to reduce water vapour effects on InSAR measurements using remotely-sensed water vapour data (e.g. Global Positioning System(GPS), NASA Moderate Resolution Imaging Spectroradiometer (MODIS) and ESA Medium Resolution Imaging Spectrometer (MERIS)): Each model is capable of improving the accuracy of InSAR derived deformation signals from 10 mm before correction to 5 mm after correction (e.g. Li et al., 2009, Li et al., 2006c, Li et al., 2006b, Li et al., 2006a, Li et al., 2005). The PI is determined to capitalize on it in this research. The objectives of this proposed research are as follows: (i) to develop atmospheric model based correction technique(s) to make our correction models globally applicable; (ii) to demonstrate the ease of application of the atmospheric model based correction technique(s) under any possible scenario: (a) with a single InSAR pair, or (b) with multiple InSAR pairs; (iii) to develop atmospheric estimation techniques in InSAR time series analysis so that deformation signals can be extracted without any assumption on deformation models or any requirement on atmosphere data. This is desirable for users who are not able to access any atmospheric model. Methodology: All the above-mentioned correction models are inherently limited by data availability, e.g. the lack of a dense GPS network (for GPS-based model) and the sensitivity to the presence of clouds (for MODIS and MERIS). The Unified Model developed by the UK MET Office provides high spatial resolution (e.g. 1 km) estimates of tropospheric path delays with a global coverage, 24 hours a day in all weather, which is highly desirable for InSAR atmospheric correction, and will be investigated for the correction of atmospheric effects on a single-pair and/or multi-pairs of SAR interferograms in this research. A prior deformation model and/or some spatial/temporal filtering techniques are commonly assumed in most of current InSAR time series techniques to separate deformation signals from atmospheric effects, which is evidently not optimal. Based on three key physical features of atmospheric signals identified by previous studies, an iteration process is proposed to estimate atmospheric signals from multiple InSAR pairs and then to extract deformation signals without any assumption of deformation models and any requirement on atmosphere data.