Empirical rather than simulated spectral band adjustment of MODIS to Sentinel-3 OLCI for European vegetation monitoring
Abstract
Spectral band adjustment derived from convolved spectra describes sensor differences under controlled conditions, but its applicability to operational products requires assessment. We compare this approach with adjustment estimated directly from MODIS and Sentinel-3 OLCI observations for European vegetation monitoring. Both streams undergo atmospheric correction and Ross-Thick/Li-Sparse Reciprocal BRDF normalization to nadir viewing and local solar noon. Using observations from 2019-2024, we derive annual OLCI/MODIS reflectance-ratio relationships with unadjusted MODIS NDVI. The operational correction uses quadratic red and linear near-infrared (NIR) functions. We compare these relationships with irradiance-weighted canopy simulations and assess the empirical correction at 460 locations. Simulated and empirical relationships differ, particularly for red reflectance at high NDVI. Background-dependent simulated ratios also demonstrate the limitations of a single NDVI-based function. The empirical correction reduces red and NIR root-mean-square error (RMSE) at 66.3% and 71.1% of locations, with median paired reductions of 5.98% and 13.67%. Median paired RMSE decreases by 0.86% for NDVI but increases by 0.95% for EVI2, 0.74% for NIRv and 7.07% for the Plant Phenology Index (PPI), calculated with stream-specific historical references. This study establishes an operational empirical adjustment and shows that the specified spectral simulation does not reproduce the relationship between the normalized products. It provides evidence for improved reflectance consistency while demonstrating that vegetation-index benefits require separate assessment.
Related
- Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds
- Spectral stability of empirical metric-measure Laplacians
- Empirical-Bayes spectral partial pooling across related tasks
- Regression adjustment in completely randomized experiments with many covariates
- Automated, Efficient, and Model-Free Inference for Randomized Clinical Trials via Data-Driven Covariate Adjustment
BibTeX
@misc{jin2026empirical,
title = {{Empirical rather than simulated spectral band adjustment of MODIS to Sentinel-3 OLCI for European vegetation monitoring}},
author = {Hongxiao Jin},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.02212},
eprint = {2611.02212},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.02212},
}
RIS
TY - UNPB
TI - Empirical rather than simulated spectral band adjustment of MODIS to Sentinel-3 OLCI for European vegetation monitoring
AU - Hongxiao Jin
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.02212
UR - https://arxiv.org/abs/2611.02212
AB - Spectral band adjustment derived from convolved spectra describes sensor differences under controlled conditions, but its applicability to operational products requires assessment. We compare this approach with adjustment estimated directly from MODIS and Sentinel-3 OLCI observations for European vegetation monitoring. Both streams undergo atmospheric correction and Ross-Thick/Li-Sparse Reciprocal BRDF normalization to nadir viewing and local solar noon. Using observations from 2019-2024, we derive annual OLCI/MODIS reflectance-ratio relationships with unadjusted MODIS NDVI. The operational correction uses quadratic red and linear near-infrared (NIR) functions. We compare these relationships with irradiance-weighted canopy simulations and assess the empirical correction at 460 locations. Simulated and empirical relationships differ, particularly for red reflectance at high NDVI. Background-dependent simulated ratios also demonstrate the limitations of a single NDVI-based function. The empirical correction reduces red and NIR root-mean-square error (RMSE) at 66.3% and 71.1% of locations, with median paired reductions of 5.98% and 13.67%. Median paired RMSE decreases by 0.86% for NDVI but increases by 0.95% for EVI2, 0.74% for NIRv and 7.07% for the Plant Phenology Index (PPI), calculated with stream-specific historical references. This study establishes an operational empirical adjustment and shows that the specified spectral simulation does not reproduce the relationship between the normalized products. It provides evidence for improved reflectance consistency while demonstrating that vegetation-index benefits require separate assessment.
ER -