A global long-term (2002–2022) C-band vegetation optical depth record retrieved after merging AMSR-E, AMSR2 and WindSat

• Proposed a calibration method for merging TB observations from three sensors. • Assessed the long-term C-VOD retrieved from the merged TB. • The long-term C-VOD was found to be highly consistent over three sensors. Vegetation optical depth (VOD) based on microwave remote sensing has become an essential indicator for monitoring large-scale vegetation conditions. The relatively long-term record of satellite-based passive C-band observations offers the potential for assessing dynamic vegetation changes. Due to the limited operational lifetime of individual microwave sensors, retrieving a long-term C-band VOD (C-VOD) dataset often requires combining multiple sensors, such as Advanced Microwave Scanning Radiometer for EOS Aqua (AMSR-E, 2002–2011), Advanced Microwave Scanning Radiometer 2 (AMSR2, 2012–present) and WindSat (2003–2020). Retrieving a long-term C-VOD from their C-band brightness temperature (TB) observations faces two major challenges: (i) non-negligible systematic biases exist between the TB observations among the three sensors; (ii) TB from different sensors may not be strongly associated, especially in tropical forests with small seasonal variations. To address these challenges, we employed a combined inter-calibration method, using linear regression in sparse vegetation and linear rescaling in dense vegetation, to merge TB from three sensors. Results showed that: (i) in undisturbed dense forests where minimal emissivity variation can be assumed, the merged TB (2002–2022) exhibited strong temporal correlations with skin temperature (H polarization: R = 0.90, V polarization: R = 0.86); (ii) the merged C-VOD retrieved from the merged TB exhibited substantially improved temporal consistency across three sensors, reducing global discrepancies between AMSR-E and AMSR2 from 6.20 % to 0.34 %. Furthermore, the merged C-VOD showed stable long-term consistency across sensors, with paired t-tests indicating no significant differences (P-value > 0.01) at the global and vegetation-type scales, confirming reliable cross-sensor continuity; (iii) the merged C-VOD showed stronger temporal correlations with normalized difference vegetation index, enhanced vegetation index, and site-level gross primary production across more vegetated areas globally, and exhibited higher and more stable spatial correlations with vegetation variables across sensor transitions compared to the existing merged C-VOD product.