Data from NASA's Orbiting Carbon Observatory-2 (OCO-2) satellite is essential to many carbon management strategies. A retrieval algorithm is used to estimate CO2 concentration using the radiance data measured by OCO-2. However, due to factors such as cloud cover and cosmic rays, the spatial coverage of the retrieval algorithm is limited in some areas of critical importance for carbon cycle science. Mixed land/water pixels along the coastline are also not used in the retrieval processing due to the lack of valid ancillary variables including land fraction. We propose an approach to model spatial spectral data to solve these two problems by radiance imputation and land fraction estimation. The spectral observations are modeled as spatially indexed functional data with footprint-specific parameters and are reduced to much lower dimensions by functional principal component analysis. The principal component scores are modeled as random fields to account for the spatial dependence, and the missing spectral observations are imputed by kriging the principal component scores. The proposed method is shown to impute spectral radiance with high accuracy for observations over the Pacific Ocean. An unmixing approach based on this model provides much more accurate land fraction estimates in our validation study along Greece coastlines.
翻译:美国航天局轨道碳观测站-2号(OCO-2)卫星的数据对许多碳管理战略至关重要,利用CO-2号测算的弧度数据估算CO2浓度时使用了一种检索算法,但由于云层覆盖和宇宙射线等因素,检索算法的空间范围有限,对碳循环科学至关重要的某些领域也很有限。由于缺少包括土地分数在内的有效辅助变量,海岸线上的混合土地/水象素也没有用于回收处理。我们建议采用一种方法,模型空间光谱数据来解决这两个问题。我们建议采用一种方法,通过光谱估计和土地分数来模拟空间光谱数据,以解决这两个问题。光谱观测以空间指数化功能数据为模型,用具体足迹参数进行计算,并通过功能性主要组成部分分析将其缩小到低得多。主要组成部分的分数以随机字段为模型,用以计算空间依赖性,而缺失的光谱观测结果则由主要组成部分分数的 krig 来估算。我们根据这一模型对太平洋观测结果进行估算的光谱亮度和高度精确度。根据这一模型,在希腊沿岸的校准研究中采用了一种不完全的方法。