This manuscript presents a revised long-term GOME-2 solar-induced chlorophyll fluorescence (SIF) data record with substantial improvements in instrumental degradation correction, inter-sensor consistency, and temporal stability. The topic is highly relevant to the ESSD community, and the authors make a significant effort to diagnose and mitigate known issues in the GOME-2 SIF record. Overall, the manuscript is well written and technically sound, and the dataset is potentially valuable for long-term ecosystem productivity and carbon-cycle studies. However, several aspects of the data processing and correction methodology would benefit from additional clarification, validation, and uncertainty characterization.

In section 3.1, the instrumental degradation correction is derived by fitting long-term trends in globally averaged reflectance, under the assumption that the global mean reflectance should exhibit no secular trend and that any observed long-term change is attributable to sensor degradation. While this assumption is reasonable at first order, it would be helpful for users if the authors could further justify or validate it using the following ways:

Have the authors evaluated whether long-term changes in cloud fraction, aerosol loading, or land surface properties could contribute to non-instrumental trends in global mean reflectance?

Could the degradation polynomial be derived or cross-validated using more radiometrically stable reference targets (e.g. selected desert regions or invariant ocean areas), and if so, how consistent are the resulting correction factors with those derived from global averages?

In Section 3.2, the authors retrieve SIF using the 735–758 nm spectral window. However, Guanter et al. (2021) indicated that this window is more sensitive to atmospheric effects, particularly water vapor and cloud contamination, compared to narrower windows such as 743–758 nm. This raises the possibility that part of the observed inconsistency between GOME-2A and GOME-2B SIF over regions such as the Amazon and Eastern China, which were characterized by frequent cloud cover and high atmospheric humidity. This inconsistency may be driven by differences in atmospheric conditions rather than instrumental effects alone. Consequently, the robustness of the product in these regions may be partially limited by atmospheric influences inherent to the chosen fitting window. The authors are encouraged to discuss this potential limitation explicitly and, if possible, assess the sensitivity of the inter-sensor consistency to the choice of spectral window.

In Section 3.3, the authors assume that the systematic bias is primarily latitude-dependent and use the Pacific Ocean as a reference region for deriving the correction. However, ocean surfaces have intrinsically low radiance/reflectance, which may lead to systematically low retrieved SIF values and potentially different error characteristics compared to vegetated land. It is therefore unclear whether a bias estimated over water can be transferred to land vegetation in a physically consistent way. Please justify why an ocean-based reference is representative for correcting land SIF, and provide the post-correction seasonal SIF time series over the selected PICS regions to demonstrate that the correction preserves realistic seasonal dynamics. Moreover, previous studies have suggested that retrieval biases can be related to the level of reflected radiance (i.e., scene brightness) rather than latitude per se. Given this, the manuscript should clarify why radiance/reflectance-dependent effects were not explicitly evaluated (e.g., by stratifying bias as a function of reflectance or radiance, in addition to latitude). Including such an analysis would help disentangle whether the observed bias is genuinely latitude-driven or instead a manifestation of brightness- or scene-type-dependent retrieval errors.