Supporting data for “The role of forest composition heterogeneity on temperate ecosystem carbon dynamic under climate change"

This dataset comprises four primary folders to support the research corresponding to the following chapters:Chapter 2 Reconstruction of 16-day, 30m, seamless satellite image time-series since 1995 through multi-sensor harmonization, fusion and gap-fillingIn this chapter, I developed a workflow that combines two advanced algorithms to produce a comprehensive data cube on Google Earth Engine (GEE). The process includes (1) harmonizing Landsat 5, 7, 8, and Sentinel-2 data using the HLS algorithm, and (2) filling temporal gaps with an optimized object-based STARFM fusion algorithm. This results in a seamless 30-meter spectral dataset spanning from 1995 to 2023, covering flux tower footprints within the northern Wisconsin mixed forest ecosystem. The dataset 'Decadal_NIRv_all_site.rar' is the derived product. 'Site_list.csv' is the metadata for study site in this thesis.<br>Chapter 3: Ecoregion-wise Fractional Mapping of Tree Functional Composition in Temperate Mixed Forests with Sentinel Data.In this chapter I generated the Tree Functional Type fraction maps are based on the Fisher-transformation-based Spectral and Radar Time-series Mixture Analysis (F-SRTMA) framework in temperate mixed forests across large landscapes using Sentinel-1 and -2 time-series imagery. The 3 groundtruth map in \Chapter3 foloder are used for evaluate the proposed algoritm. Please refer the published paper: https://doi.org/10.1016/j.rse.2024.114026<br>Chapter 4 Annual Forest Composition Change Mapping from 1995 to 2023 with Continuous Change Detection-based Unmixing AlgorithmIn this chapter, I generated maps of annual forest Composition change from 1995 to 2023 with Continuous Change Detection-based Unmixing Algorithm (CCDU). The ' SR_16day_endmembers_sample_conf80_HLS-STARFM_all_v2.csv' is the extracted endmembers with 16-day spectrum information and NIRv index for CCDU mapping. <br>Chapter 5 Spatio-temporal compositional heterogeneity effect on carbon cycleThe "Training_ALL_FLUX_FFPparam.csv" is post-processed eddy flux observation data with footprint-based compostional dynamics, and were used for training and testing the proposed footprint-based carbon mixture analysis (CMA) framework. For a more detailed description, please refer to the README.txt file.