GEDICorrect: A Python Framework for GEDI Geolocation Correction Using Multiple Criteria and Parallel Processing Methods

Accurately monitoring terrestrial ecosystems is essential for addressing global environmental challenges, including deforestation, biodiversity loss, and climate change. NASA's Global Ecosystem Dynamics Investigation (GEDI) mission has revolutionized ecosystem monitoring by providing near-global, high-resolution data on vegetation structure and terrain elevation using spaceborne LiDAR. However, spaceborne LiDAR data often require correction due to various sources of error, such as instrument inaccuracies, atmospheric conditions (e.g., dense cloud cover), and spacecraft platform instability. A primary challenge in utilizing GEDI data is its horizontal geolocation error, which has an accuracy of approximately 10 meters for calibrated final products (Version 2). These errors, particularly in heterogeneous landscapes, can significantly compromise the accuracy of canopy height and terrain elevation estimates.

To address these challenges, the scientific community has developed methods to enhance GEDI’s geolocation accuracy. Notably, the GEDI Simulator tool, created by the GEDI Science Team, applies orbit-level systematic corrections using small-footprint ALS data. This approach assumes a uniform systematic error across the orbit and determines a single coordinate offset to correct horizontal deviations, which can often fail in complex and heterogeneous landscapes. Consequently, alternative methods, such as beam-level corrections (calculating an independent offset for each beam track) and footprint-level corrections (computing individual offsets for each footprint), have emerged. Despite their potential, these methods, including the GEDI Simulator, face practical limitations such as complexity, computational inefficiency, and a lack of user-friendly interfaces, restricting their broader adoption for remote sensing applications.

To overcome these limitations, we introduce GEDICorrect, an open-source Python framework for precise beam and/or footprint-level geolocation correction, designed with simplicity and accessibility in mind. GEDICorrect integrates multiple methods, criteria, and metrics, including waveform matching, terrain matching, and relative height (RH) profile matching, to achieve refined geolocation accuracy at the orbit, beam, or footprint levels. By leveraging advanced similarity metrics - such as Pearson and Spearman waveform correlations, Curve Root Sum Squared Differential Area (CRSSDA), and Kullback-Leibler divergence - GEDICorrect ensures precise alignment between GEDI measurements and simulated data.

Additionally, GEDICorrect incorporates parallel processing strategies using Python’s multiprocessing capabilities, enabling efficient handling of large-scale GEDI and ALS datasets. This scalability makes the framework practical for global-scale applications while maintaining accuracy and computational efficiency. By addressing critical barriers in geolocation correction with an open-source, user-friendly design, this framework enables a better assessment of canopy structure that can be applied to a wide range of fields, from advancing our understanding of carbon sequestration to supporting more informed planning and conservation efforts.