高分辨率FPAR遥感估算及时间扩展方法研究

Accurate estimation of instantaneous Fraction of Absorbed Photosynthetically Active Radiation (FPAR) at high spatial and high temporal resolutions from satellite data is of great significance to precise modelling of land surface processes at finer scales. However, contemporary studies provide insufficient methodological support for the estimation and temporal scaling-up of fine-resolution FPAR. Based on FPAR physical models and scaling theories, this study takes into account the spatial heterogeneity and complex changes of land surface and aims at developing a methodological framework for estimating instantaneous FPAR values at high spatial and high temporal resolutions. Firstly, the relations between coarse-resolution FPAR operational satellite products and fine-resolution land surface reflectance data will be investigated and simplified. Then a general, simple and efficient FPAR model for estimating FPAR from fine-resolution satellite data will be developed. Secondly, the spatio-temporal information from multi-source and multi-temporal satellite data will be mined in depth to build prior knowledge about complex changes of land surface. A new method for expanding temporally sparse FPAR to daily scales will be developed on the basis of constructing and updating prior knowledge. It is advantageous in overcoming the limits of subjective assumptions used in current spatio-temporal fusion methods. The methods will be applied to Guanzhong area in Shaanxi province for demonstration at regional scale. A time series dataset of daily instantaneous FPAR at 30 m resolution will be derived and validated by in-situ measurements. The study will provide scientific reference for further studies on the estimation of daily average FPAR values and the generation of operational FPAR satellite products at high spatial and high temporal resolutions.

高时空分辨率瞬时光合有效辐射吸收比(FPAR)的准确估算，对更精细尺度上的陆面过程建模具有重要意义，为瞬时FPAR向日均尺度扩展提供数据基础。目前，高分辨率FPAR估算及时间扩展方面的研究还很不充分，尚未形成有效的方法支撑。本研究顾及地表的空间异质性及复杂变化，围绕FPAR机理模型与尺度转换理论，研究低分辨率FPAR业务化遥感产品与高分辨率地表反射率间的转换途径，建立基于低分辨率FPAR实时降尺度的高分辨率FPAR估算模型，高效且具普适性；挖掘多源多时相数据中的时空信息及关联，以先验知识构建与更新为核心发展日间尺度扩展方法，克服现有时空融合方法中的强假设约束；以陕西省关中地区为例开展区域级示范应用与验证，形成30米分辨率每日瞬时FPAR数据集。将最终建立一套基于现有遥感数据的高时空分辨率瞬时FPAR估算方法，为未来研究日均FPAR估算及业务化生产高时空分辨率FPAR遥感产品提供科学依据。

高时空分辨率瞬时光合有效辐射吸收比(FPAR)的准确估算，为更精细尺度上的陆面过程建模提供数据基础。目前，高分辨率FPAR估算及时间扩展方面的研究还很不充分，在区域尺度应用时仍面临算法普适性、精度、计算效率等多重挑战。为此，本研究以建立一套基于现有遥感数据的、具有普适性的高时空分辨率瞬时FPAR估算方法为总体目标，通过合理、显式地表达时空因素对FPAR估算的影响，实现了低分辨率FPAR业务化遥感产品与高分辨率地表反射率间的转换途径，提出了一种基于低分辨率FPAR遥感数据实时降尺度的高分辨率FPAR估算方法；充分利用历史高分辨率数据中的时空信息及关联，结合作物生长过程模型、线性校正法、集合卡尔曼滤波方法，发展了一种以先验知识构建与更新为核心的高分辨率FPAR日间尺度时间扩展方法PK-STEM；以陕西省关中地区为例开展区域级示范应用与验证，形成30米分辨率每日瞬时FPAR数据集。与现有的典型算法相比，本项目提出的方法精度更高、可更好地重建FPAR时空变化细节、计算效率更高。项目成果为未来业务化生产高时空分辨率FPAR遥感产品和后续地学应用提供科学依据。