A Novel Method For Normalized Difference Vegetation Index Via Enhanced Whittaker Smoothing Framework

The Normalized Difference Vegetation Index (NDVI) time series data play a crucial role in vegetation dynamics monitoring and related fields. However, the continuity of these data is often compromised due to noise and data gaps caused by complicated atmospheric conditions and different sensor performances. To overcome existing issues and accurately describe NDVI characteristics, this study proposes a Gap Filling and L2 Regularized Whittaker Smoothing (GF-L2WS) method based on the Google Earth Engine (GEE) platform to reconstruct NDVI time series. GF-L2WS integrates MODIS and Landsat data, leveraging their temporal and spatial complementarity. Integrating MODIS NDVI time series data with cloud-free Landsat data to fill in missing values in the original Landsat NDVI time series, thereby enhancing data quality and resolution. Subsequently, Whittaker Smoothing with an added L2 regularization term is applied to the filled NDVI data to produce a higher-quality NDVI time series. Experimental results show that, compared to Whittaker and Savitzky-Golay (SG) methods, the proposed method achieves lower RMSE and higher R2, providing improved reconstruction performance for NDVI time series.