the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Jun 2021
Fusing MODIS and AVHRR products to generate a global 1-km continuous NDVI time series covering four decades
Abstract. Satellite normalized difference vegetation index (NDVI) time-series data are an essential data source for numerous ecological and environmental applications. Although various long-term global NDVI products have been produced with different characteristics over the past decades, there is still an apparent trade-off between the spatiotemporal resolution and time coverage. The Advanced Very High-Resolution Radiometer (AVHRR) instrument can provide the only continuous time series with the longest time coverage since the early 1980s, but with the drawback of a coarse spatial resolution and poor data quality compared to the observations of later instruments. To address this issue, a spatio-temporal fusion-based long-term NDVI product (STFLNDVI) since 1982 was generated in this study, with a 1-km spatial resolution and a monthly temporal resolution. A multi-step processing fusion framework was employed to combine the superior characteristics of Moderate Resolution Imaging Spectroradiometer (MODIS) and AVHRR products, respectively. Simulated and real-data assessments both confirm the ideal accuracy of the fusion result with regard to the spatial distribution and temporal variation. Only a few relatively unsatisfactory results are found due to the poor relationship between the original AVHRR and MODIS data. The evaluations also show that the proposed fusion framework can obtain stable results similar to MODIS data in different years and seasons, even when the temporal distance between the fusion data and the reference data is large. We believe that the STFLNDVI product will be of great significance to characterize the spatial patterns and long-term variations of global vegetation. The NDVI product is available at DOI: http://doi.org/10.5281/zenodo.4734593 (Guan et al., 2021).
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(2691 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on essd-2021-156', Anonymous Referee #1, 01 Jul 2021
This paper by Guan et al. aims to create a new NDVI dataset through fusion of AVHRR GIMMS 3g and MODIS NDVI dataset. The new dataset, namely, STFNDVI has a monthly temporal resolution and 1km spatial resolution, covering the period of 1981-2000. The authors applied several procedures during this process, including denoise, normalization, spatial-temporal fusion. The algorithm is evaluated during the overlapping period and resultant dataset is compared with LAC and HRPT AVHRR NDVI data which are at higher spatial resolution. Obtaining a global high resolution long-term NDVI dataset can be critical for global change studies. This study presented a first attempt to solve this issue, but from my point of view, it is still a rather premature dataset and have limited value.
I have this opinion based on the following points:
- When do we need a high-resolution dataset? The answer seems to be clear, when low resolution dataset cannot provide enough details. This include two major aspects, one is that there is enough spatial heterogeneity at finer resolution, the other is that there is additional information that can only be obtained at this finer resolution. One good example for this second point is the change of land cover, e.g., deforestation or reforestation. Using high resolution data can provide information on when these activities happen and how much do they contribute to the changes of vegetation in addition to the nature factors. Another example is tracking vegetation phenology when multiple biome types co-exist within a pixel, and they respond differently to climate change (see Zhang et al. 2017; Chen et al., 2018). Under these conditions, the sub-pixel spatial patterns within a coarse resolution pixel also changes, but the current algorithm cannot get this information. This is critical issue since getting changes of this sub-pixel patterns is often why people would use high resolution dataset. The current algorithm assumes there is no interannual variations in this sub-pixel variation since the reference sub-pixel spatial pattern is provided by one year of MODIS data. This greatly undermines the value of this high-resolution dataset and the author did not even discuss this aspect. Using long-term high-resolution observations such as Landsat may help solve this issue.
- Data quality control. One large difference between GIMMS and MODIS is the data quality control procedure. Since GIMMS does not provide effective quality flag for snow or cloud covered pixels, there can be large differences in early or late growing season in northern high latitudes, as well as the tropical ecosystem, where the authors found large discrepancy during the comparison (Figure 4 and 14). A good practice would be to remove these observations during the per-pixel normalization period based on the MODIS quality flag, and only use the good observations to build the MODIS AVHRR relationship. The author mentioned that they use Whittaker filtering method to reduce noise, however, due to the presence of cloud, snow and aerosols, the anomalies of NDVI are often negatively biased, which cannot be effectively handled by the Whittaker filtering method.
- Continuity of the dataset. The authors claim that they generated a high spatial resolution dataset spanning over four decades, I guess that they suggest this new dataset can be used in together with MODIS NDVI. However, using two datasets together may create additional problems. For example, the trend for the first period is provided by AVHRR while the second period is from MODIS. Previous studies have demonstrated that the trend from different sensors can be quite different (e.g., Jiang et al., 2017). There may be additional risks that due to the differences in sensor performance, the NDVI calculated from both sensors may have a non-linear relationship, i.e., the probability density functions (PDF) for each pixel may be different between sensors. This cannot be corrected using the linear regression method as proposed by the authors, but requires additional procedure, e.g., PDF matching. This issue can be easily tested using BFAST or other breakpoint detection algorithms.
In conclusion, this is a good attempt to generate a high-resolution dataset based on the fusion of MODIS and AVHRR, however, due to the above-mentioned issues, I don’t think this dataset meet the high standard of ESSD.
Detailed comments:
L35, visible->red
L59, decades->decadal?
L110, “limited attempts” means very few attempts, I guess the authors mean “a few attempts”
L130, MOD13A2 has a 16-day temporal resolution, I guess this should be MOD13A3?
L139, the GIMMS 3g v1 version extends to 2015 December. Did the authors use this newer version?
L175, why 1989-1993, why not longer?
L179, it is not common to use ecological communities, it usually refers to the group of people who study ecology. I suggest to use ecosystems or biome types.
L222, the authors use “prove” several times throughout the manuscript, it is a very strong word that requires rigorous test and derivation. I suggest to use “demonstrate” or “show”
L257-258, using one year of data as reference can be risking, for example if drought happens in a savanna ecosystem, the tree-grass difference is greater than normal years, which will affect the spatial patterns at sub-pixel scale.
L343: why do you need to mention “famous” here?
L370: I suggest the authors to make comparisons where land cover changes happen during the past decades, for example, “the arc of deforestation” in Amazon, Sahel region in Africa, Northern China, these are research hotspots where high resolution dataset is needed.
L385, to qualitatively analyze the difference, I suggest to add a fourth row showing the difference between the first and third.
L400, using this 3D plot does not quantitively provide the information of r since it is difficult to locate the absolute value and. The color scheme also changes for each subplot. You may consider just use 2D plot with year as x-axis, month as y-axis and use color to represent the value.
L401, Grammarly incorrect, please rewrite.
L490, is this r value calculated based on the average value of the 12 months? this should be very high since the spatial details are averaged.
Reference
Zhang, X., Wang, J., Gao, F., Liu, Y., Schaaf, C., Friedl, M., Yu, Y., Jayavelu, S., Gray, J., Liu, L., Yan, D., Henebry, G.M., 2017. Exploration of scaling effects on coarse resolution land surface phenology. Remote Sensing of Environment 190, 318–330. https://doi.org/10.1016/j.rse.2017.01.001
Chen, X., Wang, D., Chen, J., Wang, C., Shen, M., 2018. The mixed pixel effect in land surface phenology: A simulation study. Remote Sensing of Environment 211, 338–344. https://doi.org/10.1016/j.rse.2018.04.030
Jiang, C., Ryu, Y., Fang, H., Myneni, R., Claverie, M., Zhu, Z., 2017. Inconsistencies of interannual variability and trends in long-term satellite leaf area index products. Global Change Biology 23, 4133–4146. https://doi.org/10.1111/gcb.13787
Citation: https://doi.org/10.5194/essd-2021-156-RC1 -
AC2: 'Reply on RC1', Xiaobin Guan, 15 Sep 2021
Dear Referee #1,
We are particularly grateful for your deep thoughts and valuable comments. Although the NDVI product presented in our study still has many problems that need further investigation, it is the first attempt to produce the global 1-km long-term NDVI dataset that is helpful for global change studies. According to your comments, we have tried our best to revise the manuscript to make it better, and an item-by-item response can be found in the Supplement. In the revised manuscript, we added some experiments to further assess the spatial patterns and data continuity of STFLNDVI, and added more discussions on the uncertainties in the process and results. Furthermore, we are also reproducing the STFLNDVI product by referring to the mean value of MODIS and AVHRR over the overlapping period, in order to reduce the risk caused by the selected one-year reference data.
Once again, we are particularly grateful for your careful reading and constructive comments. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
-
RC2: 'Comment on essd-2021-156', Anonymous Referee #2, 21 Jul 2021
Guan et al. generated a new NDVI dataset, STFLNDVI, by merging the data of MODIS NDVI and AVHRR GIMMS 3g. MODIS NDVI product has good data quality and a high spatial resolution but it is available since the year 2000. AVHRR GIMMS 3g product has been provided since 1982 but it has a relatively coarse spatial resolution (1/12 degree) and relatively poor data quality. The authors then performed the temporal filtering, normalization, and spatial-temporal fusing, making a new NDVI dataset of STFLNDVI with 1-km spatial resolution, covering the period of 1982-2015. Furthermore, the authors checked the temporal consistency, spatial stability, and spatial consistency of the new product during the overlapping periods of MODIS, ANCUS NDVI data. This draft was well-written, but I still have some comments on the algorithms used in this analysis, and I think the novelty is insufficient for a paper in ESD.
Major comments:
- Doubt on the reliability of the spatial variations at fine resolution. The original AVHRR product at coarse resolution can not provide any spatial variations within 1/12 x 1/12 pixels. The authors claimed that this newly generated NDVI product at 1-km resolution has the information of spatial variations within 1/12 x 1/12 resolution. Such spatial variations for every year are derived from the reference year (2014) of MODIS data. This means that, for STFLNDVI, the spatial variations within 1/12 x 1/12 resolution have no temporal change. This is no realistic, and I think the “high resolution” of STFLNDVI seems like a “pseudo high resolution”.
- Doubt on the reliability of the short-term temporal variations. When doing the normalization (section 3.1.2), the authors just used a linear model to make the multi-year mean value and trend of AVHRR data as same as MODIS data (as shown in Fig 3). The interannual variability or temporal variations within the year of STFLNDVI are from the AVHRR data without any correction. The short-term temporal variations of AVHRR aren’t always consistent with those of MODIS data, for example in the regions around the equator. Merging two datasets may lead to some artificial variations.
Specific comments
Ln 36-37: The logic should be “NDVI can not only …vegetation coverage and growth status, which is associated with …”. And it would be better to use “associated with” rather than “correlated with”
Ln 295: freely downloaded => downloaded for free
Ln 300: I don’t understand why the authors show the mean NDVI of the year 1990 here? How about the other years?
Ln 336: Why r in Europe (forest) and South of China (forest) is relatively low (Fig 4a)?
Fig 5, Ln 357: It would be better to give the statistics of r and bias for each climatic biome in Fig 5. The patterns in Fig 4 show low consistency in some regions, but this information has been hidden in global statistics.
Fig 7, Ln 392: As shown in Fig 3, the mean difference between fusion results and MODIS is less than 0.1. In Fig 7, the bins of colorbar are 0.2 or 0.1. It is possible that patterns of fusion results and MODIS have some differences, but these differences can’t be shown because of colorbar setting. Could you please check this?
Fig 8 and 9: Colorbars are missing!
Citation: https://doi.org/10.5194/essd-2021-156-RC2 -
AC3: 'Reply on RC2', Xiaobin Guan, 15 Sep 2021
Dear Referee #2,
We are particularly grateful for your careful reading and constructive comments. Although there are still insufficiencies in the STFLNDVI product presented in our study, it is the first attempt to produce the global 1-km long-term NDVI dataset that may be helpful for global change studies. We have taken full consideration of your comments to improve the product and revise the manuscript to make it better. An item-by-item response can be found in the Supplement. More experiments are conducted to assess the availability of STFLNDVI on spatial patterns and temporal variations. We are also reproducing the dataset by referring to the mean value calculated from the overlapping MODIS and AVHRR time series, in order to reduce the uncertainties caused by the selected one-year reference data.
Once again, we are particularly grateful for your careful reading and constructive comments. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
-
EC1: 'Comment on essd-2021-156', Bo Zheng, 18 Aug 2021
Dear authors,
Thanks for submitting your manuscript. Both of the two referees have pointed out the issues with data quality control fusing MODIS and AVHRR products and the reliability of spatial-temporal variations in your new data product at fine spatial resolution. Please consider seriously the comments and suggestions from the reviewers.
Kind regards,
Bo
Citation: https://doi.org/10.5194/essd-2021-156-EC1 -
AC1: 'Reply on EC1', Xiaobin Guan, 15 Sep 2021
Dear editor,
The authors are particularly grateful to the editors and the anonymous referees for their careful reading and constructive comments for this manuscript (essd-2021-156).
We have seriously considered the comments and tried our best to revise the manuscript to make it better, especially the issues with data quality control and the reliability of spatial-temporal variations for our new data product. An item-by-item response can be found in the Supplements for the reply to the two referees. We are also reproducing the STFLNDVI product to improve its quality and accuracy, according to the comments suggested by the two referees.
We would like to ask whether we are encouraged to submit a revised manuscript or not. If yes, please tell us at any time you convenient. Once again, we are particularly grateful for your efforts in handling our manuscript. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
Citation: https://doi.org/10.5194/essd-2021-156-AC1 -
EC2: 'Reply on AC1', Bo Zheng, 27 Sep 2021
Dear authors,
Thanks for your careful replies to the reviewers’ comments. While the referees find your work of some interest, they in particular raise concerns about no temporal changes in sub-pixel variation and the reliability of interannual and seasonal variations. Your replies have not addressed their comments well at present. Overall, I feel that the concerns from reviewers are sufficiently important to generate a reliable and accurate dataset sufficient for a paper in ESSD.
I am sorry that I cannot be more positive on this occasion. I hope that you will find our referees' comments helpful when improving your dataset and revising your paper for submission elsewhere.
Yours sincerely
Bo Zheng
Citation: https://doi.org/10.5194/essd-2021-156-EC2
-
EC2: 'Reply on AC1', Bo Zheng, 27 Sep 2021
-
AC1: 'Reply on EC1', Xiaobin Guan, 15 Sep 2021
Interactive discussion
Status: closed
-
RC1: 'Comment on essd-2021-156', Anonymous Referee #1, 01 Jul 2021
This paper by Guan et al. aims to create a new NDVI dataset through fusion of AVHRR GIMMS 3g and MODIS NDVI dataset. The new dataset, namely, STFNDVI has a monthly temporal resolution and 1km spatial resolution, covering the period of 1981-2000. The authors applied several procedures during this process, including denoise, normalization, spatial-temporal fusion. The algorithm is evaluated during the overlapping period and resultant dataset is compared with LAC and HRPT AVHRR NDVI data which are at higher spatial resolution. Obtaining a global high resolution long-term NDVI dataset can be critical for global change studies. This study presented a first attempt to solve this issue, but from my point of view, it is still a rather premature dataset and have limited value.
I have this opinion based on the following points:
- When do we need a high-resolution dataset? The answer seems to be clear, when low resolution dataset cannot provide enough details. This include two major aspects, one is that there is enough spatial heterogeneity at finer resolution, the other is that there is additional information that can only be obtained at this finer resolution. One good example for this second point is the change of land cover, e.g., deforestation or reforestation. Using high resolution data can provide information on when these activities happen and how much do they contribute to the changes of vegetation in addition to the nature factors. Another example is tracking vegetation phenology when multiple biome types co-exist within a pixel, and they respond differently to climate change (see Zhang et al. 2017; Chen et al., 2018). Under these conditions, the sub-pixel spatial patterns within a coarse resolution pixel also changes, but the current algorithm cannot get this information. This is critical issue since getting changes of this sub-pixel patterns is often why people would use high resolution dataset. The current algorithm assumes there is no interannual variations in this sub-pixel variation since the reference sub-pixel spatial pattern is provided by one year of MODIS data. This greatly undermines the value of this high-resolution dataset and the author did not even discuss this aspect. Using long-term high-resolution observations such as Landsat may help solve this issue.
- Data quality control. One large difference between GIMMS and MODIS is the data quality control procedure. Since GIMMS does not provide effective quality flag for snow or cloud covered pixels, there can be large differences in early or late growing season in northern high latitudes, as well as the tropical ecosystem, where the authors found large discrepancy during the comparison (Figure 4 and 14). A good practice would be to remove these observations during the per-pixel normalization period based on the MODIS quality flag, and only use the good observations to build the MODIS AVHRR relationship. The author mentioned that they use Whittaker filtering method to reduce noise, however, due to the presence of cloud, snow and aerosols, the anomalies of NDVI are often negatively biased, which cannot be effectively handled by the Whittaker filtering method.
- Continuity of the dataset. The authors claim that they generated a high spatial resolution dataset spanning over four decades, I guess that they suggest this new dataset can be used in together with MODIS NDVI. However, using two datasets together may create additional problems. For example, the trend for the first period is provided by AVHRR while the second period is from MODIS. Previous studies have demonstrated that the trend from different sensors can be quite different (e.g., Jiang et al., 2017). There may be additional risks that due to the differences in sensor performance, the NDVI calculated from both sensors may have a non-linear relationship, i.e., the probability density functions (PDF) for each pixel may be different between sensors. This cannot be corrected using the linear regression method as proposed by the authors, but requires additional procedure, e.g., PDF matching. This issue can be easily tested using BFAST or other breakpoint detection algorithms.
In conclusion, this is a good attempt to generate a high-resolution dataset based on the fusion of MODIS and AVHRR, however, due to the above-mentioned issues, I don’t think this dataset meet the high standard of ESSD.
Detailed comments:
L35, visible->red
L59, decades->decadal?
L110, “limited attempts” means very few attempts, I guess the authors mean “a few attempts”
L130, MOD13A2 has a 16-day temporal resolution, I guess this should be MOD13A3?
L139, the GIMMS 3g v1 version extends to 2015 December. Did the authors use this newer version?
L175, why 1989-1993, why not longer?
L179, it is not common to use ecological communities, it usually refers to the group of people who study ecology. I suggest to use ecosystems or biome types.
L222, the authors use “prove” several times throughout the manuscript, it is a very strong word that requires rigorous test and derivation. I suggest to use “demonstrate” or “show”
L257-258, using one year of data as reference can be risking, for example if drought happens in a savanna ecosystem, the tree-grass difference is greater than normal years, which will affect the spatial patterns at sub-pixel scale.
L343: why do you need to mention “famous” here?
L370: I suggest the authors to make comparisons where land cover changes happen during the past decades, for example, “the arc of deforestation” in Amazon, Sahel region in Africa, Northern China, these are research hotspots where high resolution dataset is needed.
L385, to qualitatively analyze the difference, I suggest to add a fourth row showing the difference between the first and third.
L400, using this 3D plot does not quantitively provide the information of r since it is difficult to locate the absolute value and. The color scheme also changes for each subplot. You may consider just use 2D plot with year as x-axis, month as y-axis and use color to represent the value.
L401, Grammarly incorrect, please rewrite.
L490, is this r value calculated based on the average value of the 12 months? this should be very high since the spatial details are averaged.
Reference
Zhang, X., Wang, J., Gao, F., Liu, Y., Schaaf, C., Friedl, M., Yu, Y., Jayavelu, S., Gray, J., Liu, L., Yan, D., Henebry, G.M., 2017. Exploration of scaling effects on coarse resolution land surface phenology. Remote Sensing of Environment 190, 318–330. https://doi.org/10.1016/j.rse.2017.01.001
Chen, X., Wang, D., Chen, J., Wang, C., Shen, M., 2018. The mixed pixel effect in land surface phenology: A simulation study. Remote Sensing of Environment 211, 338–344. https://doi.org/10.1016/j.rse.2018.04.030
Jiang, C., Ryu, Y., Fang, H., Myneni, R., Claverie, M., Zhu, Z., 2017. Inconsistencies of interannual variability and trends in long-term satellite leaf area index products. Global Change Biology 23, 4133–4146. https://doi.org/10.1111/gcb.13787
Citation: https://doi.org/10.5194/essd-2021-156-RC1 -
AC2: 'Reply on RC1', Xiaobin Guan, 15 Sep 2021
Dear Referee #1,
We are particularly grateful for your deep thoughts and valuable comments. Although the NDVI product presented in our study still has many problems that need further investigation, it is the first attempt to produce the global 1-km long-term NDVI dataset that is helpful for global change studies. According to your comments, we have tried our best to revise the manuscript to make it better, and an item-by-item response can be found in the Supplement. In the revised manuscript, we added some experiments to further assess the spatial patterns and data continuity of STFLNDVI, and added more discussions on the uncertainties in the process and results. Furthermore, we are also reproducing the STFLNDVI product by referring to the mean value of MODIS and AVHRR over the overlapping period, in order to reduce the risk caused by the selected one-year reference data.
Once again, we are particularly grateful for your careful reading and constructive comments. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
-
RC2: 'Comment on essd-2021-156', Anonymous Referee #2, 21 Jul 2021
Guan et al. generated a new NDVI dataset, STFLNDVI, by merging the data of MODIS NDVI and AVHRR GIMMS 3g. MODIS NDVI product has good data quality and a high spatial resolution but it is available since the year 2000. AVHRR GIMMS 3g product has been provided since 1982 but it has a relatively coarse spatial resolution (1/12 degree) and relatively poor data quality. The authors then performed the temporal filtering, normalization, and spatial-temporal fusing, making a new NDVI dataset of STFLNDVI with 1-km spatial resolution, covering the period of 1982-2015. Furthermore, the authors checked the temporal consistency, spatial stability, and spatial consistency of the new product during the overlapping periods of MODIS, ANCUS NDVI data. This draft was well-written, but I still have some comments on the algorithms used in this analysis, and I think the novelty is insufficient for a paper in ESD.
Major comments:
- Doubt on the reliability of the spatial variations at fine resolution. The original AVHRR product at coarse resolution can not provide any spatial variations within 1/12 x 1/12 pixels. The authors claimed that this newly generated NDVI product at 1-km resolution has the information of spatial variations within 1/12 x 1/12 resolution. Such spatial variations for every year are derived from the reference year (2014) of MODIS data. This means that, for STFLNDVI, the spatial variations within 1/12 x 1/12 resolution have no temporal change. This is no realistic, and I think the “high resolution” of STFLNDVI seems like a “pseudo high resolution”.
- Doubt on the reliability of the short-term temporal variations. When doing the normalization (section 3.1.2), the authors just used a linear model to make the multi-year mean value and trend of AVHRR data as same as MODIS data (as shown in Fig 3). The interannual variability or temporal variations within the year of STFLNDVI are from the AVHRR data without any correction. The short-term temporal variations of AVHRR aren’t always consistent with those of MODIS data, for example in the regions around the equator. Merging two datasets may lead to some artificial variations.
Specific comments
Ln 36-37: The logic should be “NDVI can not only …vegetation coverage and growth status, which is associated with …”. And it would be better to use “associated with” rather than “correlated with”
Ln 295: freely downloaded => downloaded for free
Ln 300: I don’t understand why the authors show the mean NDVI of the year 1990 here? How about the other years?
Ln 336: Why r in Europe (forest) and South of China (forest) is relatively low (Fig 4a)?
Fig 5, Ln 357: It would be better to give the statistics of r and bias for each climatic biome in Fig 5. The patterns in Fig 4 show low consistency in some regions, but this information has been hidden in global statistics.
Fig 7, Ln 392: As shown in Fig 3, the mean difference between fusion results and MODIS is less than 0.1. In Fig 7, the bins of colorbar are 0.2 or 0.1. It is possible that patterns of fusion results and MODIS have some differences, but these differences can’t be shown because of colorbar setting. Could you please check this?
Fig 8 and 9: Colorbars are missing!
Citation: https://doi.org/10.5194/essd-2021-156-RC2 -
AC3: 'Reply on RC2', Xiaobin Guan, 15 Sep 2021
Dear Referee #2,
We are particularly grateful for your careful reading and constructive comments. Although there are still insufficiencies in the STFLNDVI product presented in our study, it is the first attempt to produce the global 1-km long-term NDVI dataset that may be helpful for global change studies. We have taken full consideration of your comments to improve the product and revise the manuscript to make it better. An item-by-item response can be found in the Supplement. More experiments are conducted to assess the availability of STFLNDVI on spatial patterns and temporal variations. We are also reproducing the dataset by referring to the mean value calculated from the overlapping MODIS and AVHRR time series, in order to reduce the uncertainties caused by the selected one-year reference data.
Once again, we are particularly grateful for your careful reading and constructive comments. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
-
EC1: 'Comment on essd-2021-156', Bo Zheng, 18 Aug 2021
Dear authors,
Thanks for submitting your manuscript. Both of the two referees have pointed out the issues with data quality control fusing MODIS and AVHRR products and the reliability of spatial-temporal variations in your new data product at fine spatial resolution. Please consider seriously the comments and suggestions from the reviewers.
Kind regards,
Bo
Citation: https://doi.org/10.5194/essd-2021-156-EC1 -
AC1: 'Reply on EC1', Xiaobin Guan, 15 Sep 2021
Dear editor,
The authors are particularly grateful to the editors and the anonymous referees for their careful reading and constructive comments for this manuscript (essd-2021-156).
We have seriously considered the comments and tried our best to revise the manuscript to make it better, especially the issues with data quality control and the reliability of spatial-temporal variations for our new data product. An item-by-item response can be found in the Supplements for the reply to the two referees. We are also reproducing the STFLNDVI product to improve its quality and accuracy, according to the comments suggested by the two referees.
We would like to ask whether we are encouraged to submit a revised manuscript or not. If yes, please tell us at any time you convenient. Once again, we are particularly grateful for your efforts in handling our manuscript. Thanks very much for your time.
Best regards,
Ph.D. Xiaobin Guan
Citation: https://doi.org/10.5194/essd-2021-156-AC1 -
EC2: 'Reply on AC1', Bo Zheng, 27 Sep 2021
Dear authors,
Thanks for your careful replies to the reviewers’ comments. While the referees find your work of some interest, they in particular raise concerns about no temporal changes in sub-pixel variation and the reliability of interannual and seasonal variations. Your replies have not addressed their comments well at present. Overall, I feel that the concerns from reviewers are sufficiently important to generate a reliable and accurate dataset sufficient for a paper in ESSD.
I am sorry that I cannot be more positive on this occasion. I hope that you will find our referees' comments helpful when improving your dataset and revising your paper for submission elsewhere.
Yours sincerely
Bo Zheng
Citation: https://doi.org/10.5194/essd-2021-156-EC2
-
EC2: 'Reply on AC1', Bo Zheng, 27 Sep 2021
-
AC1: 'Reply on EC1', Xiaobin Guan, 15 Sep 2021
Data sets
STFLNDVI: A long-term 1km NDVI time series since 1982 by fusing MODIS and AVHRR products Xiaobin Guan, Huanfeng Shen, Yuchen Wang, Dong Chu, Xinghua Li, Linwei Yue, Xinxin Liu, Liangpei Zhang http://doi.org/10.5281/zenodo.4734593
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 2,088 | 890 | 94 | 3,072 | 65 | 95 |
- HTML: 2,088
- PDF: 890
- XML: 94
- Total: 3,072
- BibTeX: 65
- EndNote: 95
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
