HiQ-LAI: A High-Quality Reprocessed MODIS LAI Dataset with Better Spatio-temporal Consistency from 2000 to 2022

Abstract. Leaf Area Index (LAI) is a crucial parameter for characterizing vegetation canopy structure and energy absorption capacity. The Moderate Resolution Imaging Spectroradiometer (MODIS) LAI has played a significant role in landmark studies due to its clear theoretical basis, extensive historical time series, reliable validation results, and open accessibility. However, MODIS LAI retrievals are calculated independently for each pixel and a specific day, resulting in high noise levels in the time series and limiting its applications. Existing reprocessing studies for MODIS LAI predominantly rely on temporal information to achieve smoother LAI profiles, with little use of spatial information. This may not only easily ignore genuine LAI anomalies but also likely lead to certain overfitting issues. To address these problems, we designed the Spatio-Temporal Information Compositing Algorithm (STICA) for the reprocessing of MODIS LAI products. This method integrates information from multiple dimensions, including pixel quality information, spatio-temporal correlation, and original retrieval, and thus enables both "reprocessing" and "data value-added" of the existing MODIS LAI products, leading to the development of the High-Quality LAI (HiQ-LAI) dataset. Compared to ground measurements, HiQ-LAI shows better performance than the original MODIS product with Root-Mean-Square Error (RMSE) / Bias decreased from 0.87 / -0.17 to 0.78 / -0.06. This is due to the improvement of HiQ-LAI in capturing the seasonality of vegetation phenology and in reducing time-series abnormal fluctuations. The Time-series Stability (TSS) index which represents temporal stability, indicated that the area with smooth LAI time-series expanded from 31.8 % (MODIS) to 78.8 % (HiQ) globally, and this improvement is more obvious in equatorial regions where optical remote sensing usually cannot achieve good performance. We found that the HiQ-LAI demonstrates superior continuity and consistency compared to raw MODIS LAI from both spatial and temporal perspectives. We anticipate that the global HiQ-LAI time-series, generated by the STICA procedure on the Google Earth Engine (GEE) platform, will significantly enhance support for diverse global LAI time-series applications. The 500 m/5 km-8 days HiQ-LAI dataset from 2000 to 2022 is available at https://doi.org/10.5281/zenodo.8296768 (Yan et al., 2023).