A new hybrid filter for NDVI time series reconstruction and data quality enhancement

Abstract

NDVI (Normalized Difference Vegetation Index) is an essential tool for climate and environmental monitoring, but it is often contaminated by clouds and unfavorable atmospheric conditions. In this study, we designed a new, simple yet effective method for reconstructing data, which we call the Hybrid Filter. This study is the first to reconstruct missing NDVI time series data in cloud-prone areas by combining several data reconstruction techniques with a forecasting technique based on Exponential Moving Average (EMA). The study was conducted in Cilegon City and Batu City using NDVI time series data from the Landsat 8 satellite for the period 2014-2022. Experimental results show that the hybrid filter significantly outperforms the Spatio-Temporal Savitzky-Golay (STSG) filter, Gap Filling Savitzky-Golay (GFSG) filter, Savitzky-Golay (SG) filter, and Whittaker filter. The hybrid filter is capable of recovering missing data with high accuracy, stability, noise reduction, and maintaining the temporal integrity of NDVI data even under conditions of large data gaps and high missing data rates, making it a reliable solution for NDVI analysis in cloud-prone areas. These findings affirm the superiority of the hybrid filter in producing accurate and reliable NDVI data for vegetation and environmental monitoring.

Keywords

• NDVI time series
• data reconstruction
• data gaps
• cloud
• Landsat-8 NDVI
• hybrid filter