Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper systematically cross validated the consistency of surface reflectance of two hyperspectral satellites (GF5-02 AHSI and EnMAP) from China and Germany under different surface types. Experimental results showed that GF5-02 and EnMAP satellites have good consistency on most surface types. The paper could be improved by addressing the following problems.

1. In the introduction, the author should summarize the current status of multi-source verification of hyperspectral satellites in China and Germany. At the same time, the challenges faced in the multi-source application of hyperspectral satellites in China and Germany are presented. The problems and gaps that need to be addressed in the introduction are not clear.
2. For the radiometric calibration, atmospheric correction and geometric correction, paco processor, the authors should briefly introduce its algorithm principles.
3. The details and assumptions of SRF convolution should be explained to avoid misunderstandings.
4. Although the data was collected from the same location, how should consider the impact of data collected at different times on the results?
5. In Figure 2, there are too few spectral curve comparisons for typical land objects.
6. In the result analysis, the authors should analyze the specific differences between FLAASH and PACO in atmospheric correction, such as the corresponding compensation parameters.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This work presents a comparison of the surface reflectance betwen GF5-02 and EnMAP missions across different land cover types. I personally find this kind of studies very interesting and helping to advance in the direction of inter-operability between missions and achieving consistent results with different missions. The work shows a good agreement betwen the two missions in terms of reflectance, and although I find that the methdology used for the inter-comparison is correct, I find some that some important details about the analysis are missing as well as bit more of depth on the analyusis. For that reason I would recommend to reconsider after a major revision where the following aspects shall be considered:

- Section 2.1: SSD (spectral sampling distance) of EnMAP and FWHM are not constant along the spectral range. I would suggest in Table 1 to include the actual ranges of FWHM because the "+-" notation could be interpreted as uncertainties. I know it is usually represented like that in some documents, but it can be misleading. If you want the details of the spectral calibration of EnMAP you can find them in the Mission Quarterly Reports available here: "https://www.enmap.org/mission/". The latest one is here: https://www.enmap.org/data/doc/EN-GS-RPT-1111_Mission_Quarterly_Report_11.pdf 

EnMAP provides in each product the spectral characterization (metadata) but also a summary in here: https://www.enmap.org/data/doc/EnMAP_Spectral_Bands_update.xlsx

- Section 2.1: It could be of interest other parameters for both sensors like the SNR. 

- Section 2.2: Equation 1 is the simplest model. Actually in EnMAP there is also a non-linearity correction applied as well. Is there a non-linearity correction in GF5-02? Non-linearity could be a contributor to differences at very low signal levels.

- Section 2.3: I believe the resampling is done using the individual band information (individual central wavelenghts and FWHMs that vary from one band to another). It would be good to mention that explicitly. In addition, the FWHM in the SWIR of the two sensors is very similar, that could limit the comparison. Have you tried other techniques to perform the spectral resampling, in particular other techniques like Burggraaff 2020 ("Biases from incorrect reflectance convolution", Optics Express Vol. 28, Issue 9, pp. 13801-13816 (2020)) could be better match. 

- The spectral resampling is necessary to compare the results, but also spatial resampling is needed. There are not much details on this point. Only that pixel-to-pixel comparison is possible. Is it really done pixel-pixel? It shall be done over homogeneous areas (it shall be mentoined) and I would suggest that even in that case to use 3x3 or 5x5 pixel areas. I do not know the deatils of GF5-02, but in the case of EnMAP a resampling between VNIR and SWIR sensors is necessary becasue after geometric calibration there is a typical offset of ~0.1 pixel  (requirement was 0.2 pixel). To minimize the influence of these effects it would be better to use several pixels in teh comparison. Perhaps this is how the comparison is done, but the details are missing in teh description of the methodology. 

- The 4 sites have different time intervals between the two sensors. For vegetation it is the same day and in the others can be very long, but the targets are expected to be more insensitive to that. Still there will be differences on view angles and Sun angles. It would be good to have more information on the local time passage of the two sensors and if effects like BRDF are considered (probably not in both cases). 

- Figure 2, I believe the image is coming from GF5-02, I would mention that in the caption

- Section 3.2: The comparison is done using the L2A EnMAP products processed with PACO software, but it shall be mentioned that in EnMAP dedicated water products exist that provide "remote sensing reflectance" 

- Section 3.2, in the sentence "This level of agreement provides confidence in using GF5-02 data for geological applications where accurate detection of subtle absorption features is required". The agreement in reflectance is an excellent starting point, but for detection of subtle absorption features other aspects are also important: spectral stability, radiometric stability, SNR, sensitivity to atmospheric parameters in the atmospheric correction.

- Figure 3 caption: it suggest the data corresponds to the complete scene, but most likely it is done on a small area. It shall be more precisely described where the comparison is performed

- Section 3.2 and 3.3: For all results and graphs in Figure 3, it is not mentioned for which bands the comparison is performed (all of them? excluding bands affected by strong atmospheric absorption?). It should be better described in the methodology how the comparison is done. In genreal it would make sense to exclude the bands that could be more affected by the atmospheric absorption to better separate the effect of the sensor and the atmospheric correction. 

- Section 3.3: It would be also good to see correlation coefficients for different individual bands or spectral ranges. I guess it would be possible to make a graph that summarizes those results. 

- Section 3.4: There are no details on the ground measurement (instrument, spectral response, measurement area, ...) and the methodology followed to be able to compare the measurements. I guess it would be similar to the one followed in the comparison with EnMAP, bue there are no details and they are needed to judge better the results. It would also be good to have a reference about the measurements or the insturment. 

- Conclussions: the larger disagreement over water could be due to the lower signal in these scenes or that the signal is present only on a limited spectral range. An analysis of the results on individual bands or spectral ranges for the 4 scenes could help to clarify this point. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors solved all the problems.