A Method for Reconstructing Long-Term Daily Resolution EVIs Based on MODIS Daily Surface Reflectance Products
WANG Ning,TIAN Jia,TIAN Qingjiu
DOI: https://doi.org/10.11834/jrs.20243141
2024-01-01
Abstract:The Enhanced Vegetation Index(EVI)combines factors such as atmospheric,soil,and saturation conditions and effectively correlates these data with vegetation biomass,leaf area index,and photosynthetically active radiation.Although the performance of the EVI is better than that of the Normalized Difference Vegetation Index(NDVI),the low temporal resolution of EVI products and the presence of cloud cover often result in a large number of missing pixels.In this study,we propose a daily resolution EVI reconstruction method that combines the Maximum Value Composite(MVC)and harmonic analysis of time series(HANTS)algorithms based on MODIS daily surface reflectance products. Given the spectral response differences of varying sensors carried by different satellites,the comparability of the EVIs calculated based on the Terra and Aqua satellites was analyzed prior to conducting the MVC operation.The analysis revealed a strong spatial linear correlation between the two variables,with R2 and RMSE values ranging from 0.9796-0.9935 and 0.0116-0.0297,respectively.The annual mean R2 and RMSE values were 0.9883 and 0.0196,respectively.The fitted parameters a and b had value ranges of 0.9447 to 1.0420 and-0.0065 to-0.0072,respectively,with annual mean values of 0.9910 and 0.0012.Despite spectral differences,the calculated EVIs based on the two satellite datasets exhibit minimal differences and thus are suitable for further processing via the MVC algorithm. This method was applied to reconstruct daily resolution EVI time series data for the North China Plain in 2021.The proposed EVI reconstruction algorithm is effective for large-scale and long-term reconstructions of daily resolution EVI time series data.The reconstructed EVI yields a rich texture,fills in the missing pixels,removes noise from the original EVI data,and follows the changing patterns of various land cover types.The HANTS method offers three advantages over the S-G filtering algorithm.First,compared with the original EVI,the HANTS method better preserved the spatial distribution patterns of the original EVI during reconstruction;by contrast,the S-G algorithm exhibited larger changes in spatial distribution in the reconstructed EVI.Second,the EVI curves reconstructed using the HANTS algorithm are smoother with minimal noise for typical land cover types;by contrast,the EVI curves reconstructed using the S-G algorithm have more local noise and nondifferentiable points,which hinders the extraction of vegetation phenological characteristics.Third,in terms of fidelity evaluation against high-quality reference EVI pixels,the HANTS algorithm demonstrated a strong linear correlation with the reference EVI pixels.The R2 and RMSE values ranged from 0.91 to 0.97 and from 0.017 to 0.032 across the months,with the strongest and weakest correlations occurring in September and June,respectively.By contrast,the S-G algorithm showed a weaker linear correlation with the reference EVI pixels.The R2 and RMSE values ranged from 0.38 to 0.91 and from 0.055 to 0.206 across the months,with the strongest and weakest correlations occurring in May and August,respectively.Overall,the HANTS method consistently outperformed the S-G method in terms of fidelity,with higher R2 values and lower RMSE values across all months.The proposed daily resolution EVI reconstruction method offers new guidelines and technical approaches for generating high-temporal resolution EVI data.