Scale Effects and Regional Disparities of Land Use in Influencing PM2.5 Concentrations: A Case Study in the Zhengzhou Metropolitan Area, China
Round 1
Reviewer 1 Report
At least the statistical approach mimics general expectation.
1) How sensitive are the results to changes in state parameters, local farming practices?
2) How sensitive are the results to changes in Land cover and changes in PM2.5 (over space which was done and time)?
3) (i) Were the MODIS aerosol (PM2.5) dataset used or (ii) were they compared to the PM data or used to create the PM relevant data at least. If the former (i) then what was the atmospheric model used to retrieve the aerosol, where and when. Changes in the viewing geometry and atmospheric gases can contribute to some of the uncertainty in the results or compensate for one statistical approach short falls.
4) Did you compare the MODIS PM values with local PM data taken during the time of the satellite overpass?
5) The following statement is not completely correct if at all, “Because land use 331 showed spatially heterogeneous relationships with PM2.5 concentrations (Figure 4), the 332 land-bearing emissions and absorption (source-sink) capacity of pollutants differed across 333 Land 2022, 11, x FOR PEER REVIEW 10 of 12 regions. Thus, spatial optimization from the perspective of source-sink coupling may be 334 considered to achieve the target minimum PM2.5 concentrations”. First, the data sources being used have multiple spatial, temporal and spectral biases.
One can not just compare these values as provided. Sure resampling the 30 m data to the various scales helps the statistical aspects of the analysis, but not the physical-chemical contributions to the variation ‘water-down; by averaging the data. This is 1 place where the sensitivity analysis can help.
PM2.5 concentration (with concentration units) is a bulk value and its composition is a sum total of the entire sampling period plus any local chemical reactions that might occur due to environmental conditions, for example. The land cover changes might be more controlled and obtained at a finer resolution, but they could be influenced by landscape, environmental conditions and local farming practices. Further the land cover data is not representative of the same time scale nor spatial resolution of the PM data. Hence the statement can not be made.
We all need to build upon the global body of knowledge in this area and create a more robust scaling of the satellite data with aircraft and then especially to the highly resolved ground truth data for the parameters measured (PM2.5, and I think either surface spectral reflectance or derived NDVI or one of the other vegetative indices).
Once these points are addressed in the paper, the paper will be closer to being publishable
Author Response
Thank you. Please see the attached file
Author Response File: Author Response.pdf
Reviewer 2 Report
Article - Scale Effects and Regional Disparities of Land Use in Influencing PM2.5 Concentrations: A Case Study in the Zhengzhou 3 Metropolitan Area, China
The work present an actually approach and a statistic treatment propose. The considerations are minor, but the revisor considering relevant to understand the results better.
- The text introduction must present the PM2.5 definition in the first paragraph.
- Abstract present the work goal as "this study investigated the scale differences in spatial variations in PM2.5 concentrations, in spatial associations between PM2.5 concentrations and land use, and explored the effects 15 of the spatial heterogeneity and action scale of land use on PM2.5 concentrations". The same must be done in the introduction last paragraph.
- In the beggining of item 3.2 (line 197) the map of land use classification must be described to help understanding the results better.
- Figure 4 - the scale levels needs to be the same. The effects comparison stay more clear.
- Discussion and conclusions has differents objectives. The authors must separated them.
- In Discussion and conclusion the paragraphs are to long.
Author Response
Thank you. Please see the attached file
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Thanks for the improved manuscript. I recognized and accept some of the comments suggested were made.
Why not add the following to the discussion section?
This reviewer doesn't need the answers to the questions posed. The reviewer already knows those answers and of Hammer's work, for example. Your readers might not. There is much more literature that is relevant and not cited by the authors. The literature noted in the aforementioned if reviewed and included as appropriate would have enhanced the quality of this paper for hundreds of thousand more readers, and advanced the state of the understanding in this part of the field. The authors did address the questions and directed edits adequately. But reading through the responses to this reviewer and based on the author's admission they did not use most of the response, this reviewer believes those responses warrant inclusion in the paper as well. Some examples of what the reviewer is referring to follow along with this reviewers follow up comments/questions:
1.1 comment How sensitive are the results to changes in state parameters, local farming practices?
Response: We are grateful for the enlightening question. The result on bivariate Moran I of PM2.5 concentrations and the area of cultivated land indicated that PM2.5 concentrations have relatively stronger positive associations with adjacent cultivated land at larger scales. The GWR’S result showed that cultivated land has positive effects mainly around main urban areas of the five cities. Thus, scale effects are more sensitive to changes in local farming practices in surrounding areas of main urban. From the perspective of local farming practices, and integrating the results of this paper and related literature knowledge, we suggest reducing chemical fertilizer usage around main urban areas, especially where there are large areas of vegetable and flower cultivation, and we also suggest implementing the farmland shelterbelt project around the main urban areas, to block agricultural emission sources to transfer and concentrate into cities.
[Add a version of this to the section where this comment was placed]
1.2 comment How sensitive are the results to changes in Land cover and changes in PM2.5 (over space which was done and time)?
Response: We think the results are sensitive to changes in land cover over space (or the spatial differences of land cover), and land cover can also affect the long-term condition and trend of PM2.5 concentrations. For daily or short-term changes in PM2.5, it would be not sensitive. [Based on what is your thought based ?? Also if the anomalies in PM2.5 over a day or shorter are not related to LC/LU anomalies, then what might be and why is there a relationship in LC/LU and PM2.5-see response to comment below? If nothing else, let the readers know that this was considered. Add a version of this to the section where this comment was placed. ]
1.3 comment::(i) Were the MODIS aerosol (PM2.5) dataset used or (ii) were they compared to the PM data or used to create the PM relevant data at least. If the former (i) then what was the atmospheric model used to retrieve the aerosol, where and when. Changes in the viewing geometry and atmospheric gases can contribute to some of the uncertainty in the results or compensate for one statistical approach short falls.
Response: Yes, MODIS aerosol (PM2.5) dataset were used. This dataset combines AOD retrievals from multiple satellite algorithms including the NASA MODerate resolution Imaging Spectroradiometer Collection 6.1 (MODIS C6.1), Multi-angle Imaging SpectroRadiometer Version 23 (MISRv23), MODIS Multi-Angle Implementation of Atmospheric Correction Collection 6 (MAIAC C6), and the Sea-Viewing Wide Field-of-View Sensor (SeaWiFS) Deep Blue Version 4. The GEOS-Chem chemical transport model is used to relate this total column measure of aerosol to near-surface PM2.5 concentration.
Furthermore, the Geographically Weighted Regression (GWR) is used with global ground-based measurements from the World Health Organization (WHO) database to predict and adjust for the residual PM2.5 bias per grid cell in the initial satellite-derived values. The dataset is global annual PM2.5 concentrations product from the NASA Socioeconomic Data and Applications Center. The daily surface PM2.5 concentrations from each data source are obtained by applying the daily simulated AOD to PM2.5 ratios to the coincident daily calibrated AOD sources. Monthly means are calculated from the daily PM2.5 values. The monthly mean PM2.5 concentrations are then aggregated to annual mean values. A detailed description of the product, methodology, including satellite AOD sources, GEOS-Chem simulation and the algorithm is provided in Hammer et al. (2022) available at: https://sedac.ciesin.columbia.edu/downloads/docs/sdei/sdei-global-annual-gwr-pm2-5-modis-misr-seawifs-aod-v4-gl-03-documentation.pdf.
[Define what is meant by calibrated? Add a version of this response to the section where this comment was placed].
1.4 comment::Did you compare the MODIS PM values with local PM data taken during the time of the satellite overpass?
Response: Yes, we have verified this data. The dataset is annual PM2.5 concentrations product based on daily simulated AOD (as mentioned above), not just one time which the satellite overpass. We bought daily monitoring PM data in the study area. But there were too many missing values, especially there were no monitoring data at most stations in December, 2019. As we know, air pollution is serious and PM2.5 concentration is very high in winter in China, especially during the period from November to January of the following year. Missing values in December leaded to the statistical annual values were lower than actual PM2.5 concentration. The verified R2 between the statistical annual values and the global PM dataset is 0.7063, indicating the global PM dataset good accuracy in the study region. While, the statistical slope is 0.5044. It is caused by missing values heavy pollution period in December. Thus, we didn’t mention it in the manuscript. Besides, Hammer et al. (2020) verified the data and documented that the annual PM2.5 estimates were highly consistent with global ground monitors (R2 = 0.81; slope = 0.90).
[This is an important question. The question is not completely addressed. The comment/question is more about comparing 'apples' with 'apples'. What was done to ensure measurements used in the comparison were all made at the same time, same place with the same instruments. Naturally there are missing values and those contribute to uncertainties for example, or the low R2 you obtained.
... Add a version of your response plus the response to my new comment/request to the section where the original comment was placed]
1.5 comment::The following statement is not completely correct if at all, “Because land use showed spatially heterogeneous relationships with PM2.5 concentrations (Figure 4), the land-bearing emissions and absorption (source-sink) capacity of pollutants differed across regions. Thus, spatial optimization from the perspective of source-sink coupling may be considered to achieve the target minimum PM2.5 concentrations”. First, the data sources being used have multiple spatial, temporal and spectral biases. One can not just compare these values as provided. Sure resampling the 30 m data to the various scales helps the statistical aspects of the analysis, but not the physical-chemical contributions to the variation ‘water-down; by averaging the data. This is 1 place where the sensitivity
analysis can help.
PM2.5 concentration (with concentration units) is a bulk value and its composition is a sum total of the entire sampling period plus any local chemical reactions that might occur due to environmental conditions, for example. The land cover changes might be more controlled and obtained at a finer resolution, but they could be influenced by landscape, environmental conditions and local farming practices. Further the land cover data is not representative of the same time scale nor spatial resolution of the PM data. Hence the statement can not be made.
Response: As reviewer suggest that we removed the statement and provided new implications based on the sensitivity analysis. Please see lines 343-346 in the revised manuscript.
[OK]
1.6 comment::We all need to build upon the global body of knowledge in this area and create a more robust scaling of the satellite data with aircraft and then especially to the highly resolved ground truth data for the parameters measured (PM2.5, and I think either surface spectral reflectance or derived NDVI or one of the other vegetative indices).
Response: Yes, we think so! It is a significant and challenging work. Many researches on retrievals for PM2.5 data based on surface spectral reflectance have been conducted in different regions. And, advanced spectroscopic equipment for aerosol retrieval has been or is being developed, such as the POLDER-3 (Polarization and Directionality of the Earth’s Reflectances) instrument and the Gaofen-5 satellite DPC (Directional Polarimetric Camera) sensor.
[ Add the spatial resolution of the data from POLDER and the spectral resolution of the bands used to create the product. You would likely be comparing to MODIS. Are the product retrieval algorithms the same between MODIS and POLDER? How frequently are the POLDER data available? Is POLDER still operational. Add this response to text where the original comment was made].
Author Response
Please check the attached file
Author Response File: Author Response.pdf