the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
An Aerosol Climatology via Remote Sensing over Metro Manila, Philippines
Genevieve Rose Lorenzo
Avelino F. Arellano
Maria Obiminda Cambaliza
Christopher Castro
Melliza Templonuevo Cruz
Larry Di Girolamo
Glenn Franco Gacal
Miguel Ricardo A. Hilario
Nofel Lagrosas
Hans Jarett Ong
James Bernard Simpas
Sherdon Niño Uy
Armin Sorooshian
Abstract. Aerosol particles in Southeast Asia have a complex life cycle and consequently are challenging to characterize. The diverse topography and weather in the region complicate the situation. An aerosol climatology was established based on AERONET data (December 2009 to October 2018) for clear sky days in Metro Manila, Philippines. Aerosol optical depth (AOD) values were highest in August, coinciding with the summer southwest monsoon, due partly to fine particles from urban aerosol particles, including soot. Also, August corresponds to the burning season in insular Southeast Asia when smoke is often transported to Metro Manila. Clustering of AERONET volume size distributions (VSD) resulted in five aerosol particle sources based on the position and magnitude of their peaks in the VSD and the contributions of specific particle species to AOD per cluster based on MERRA-2. The clustering showed that the majority of aerosol particles above Metro Manila were from a clean marine source (58 %), which could be related to AOD values there being relatively smaller than in other cities in the region. The following are the other particle sources over Metro Manila: fine polluted (20 %), mixed polluted (12 %), urban/industrial (5 %), and cloud processing (5 %). Furthermore, MERRA-2 AOD data over Southeast Asia were analyzed using empirical orthogonal functions. Along with AOD fractional compositional contributions and wind regimes, four dominant aerosol particle air masses emerged: two sulfate air masses from East Asia, an organic carbon source from Indonesia, and a sulfate source from the Philippines. Knowing the local and regional aerosol particle air masses that impact Metro Manila is useful in identifying the sources while gaining insight on how aerosol particles are affected by long-range transport and their impact on regional weather.
- Preprint
(2649 KB) -
Supplement
(822 KB) - BibTeX
- EndNote
Genevieve Rose Lorenzo et al.
Status: open (until 31 May 2023)
-
RC1: 'Comment on egusphere-2023-197', Anonymous Referee #1, 25 Apr 2023
reply
General Comments:
“Climatology” is generally referred to data over at least 30 years. In this study, the Authors present far fewer years and these data are therefore considered long-term averages. The title needs to be changed. For example, it could be changed to “An Emerging Aerosol Climatology…” as inspired by Holben et al 2001. The Authors present a comprehensive analysis using several different aerosol and meteorological data sets to determine aerosol source partitioning over the Manila metropolitan region. The manuscript uses results from five clustered AERONET aerosol volume size distributions tied to MERRA-2 speciated AOD do determine the type of aerosol particles over the Manila region. At times, it is not clear how MERRA-2 data are being matched to AERONET data. As expected, Manila is dominated by coarse mode sea salt with fine mode pollution and sometimes cloud processed aerosols. Manila is impacted episodically by other aerosol sources such as smoke and pollution from other transboundary sources.
Overall, the manuscript is written and organized well; however, the data use and explanations resulted in several questions which are described below and may result in modifications to the data analysis. The questions below need to be clarified and changes supported by citations before publication.
Detailed Comments:
Lines 24 and 25: Please combine these two sentences for a more complete statement. For example, “Aerosol particles…are challenging to characterize” due to “the diverse…”
Line 26: Is it a climatology by definition? It seems more appropriate to name it a “long-term average” or perhaps “emerging climatology” as coined by Holben et al., 2001.
Lines 27 to 30: These sentences should be combined and revised to produce a more complete statement.
Lines 30-37: MERRA-2 aerosol particle composition data are on the monthly basis (as discussed in Section 2.1.2) rather than determined for the nearest time to the AERONET size distribution retrieval in which the size distributions could change significantly in aerosol urban and transported plumes. What is the difference between “fine polluted,” “mixed polluted,” “urban/industrial,” and “cloud processed” since they are all in the fine mode.
Lines 114-116: Schuster reference is not very appropriate here. More appropriate is Dubovik and King, 2000, Dubovik et al. 2000, Dubovik et al. 2002, and Dubovik et al., 2006.
Lines 132-151: Why not use the hourly MERRA-2 product (M2T1NXAER) to match up with AERONET?
Line 152-161: It is important to state here that MISR retrievals are much fewer than other LEO and GEO sensors. How well do the monthly AOD averages from MISR represent the conditions over the Philippines in such a meteorological diverse environment?
Lines 170-177: How were the NAAPS data products used? Where they used quantitatively or qualitatively?
Lines 178-181: Later discussion indicates use of other Worldview products. Please specify all products or images used.
Lines 184-185: The AERONET VSD is retrieved for discrete particle sizes and do not represent “bins” as mentioned here.
Lines 193-194: The dust category can include mixed aerosols as well as dust. Lower FMF (<0.4 or <0.3) is more appropriate. Also, the desert dust AOD may not retain high aerosol loading over 1000s of kms. What wavelength is used for AOD, AE, and FMF in the table? For example, the dust case identified over Manila for March 24-25, 2018, does not reach 0.3 at Level 2.0 AOD 500nm. Which FMF wavelength is used? Please state in the caption. Overall, these threshold values for AOD, AE, and FMF are only estimates as they are not rigorous cutoffs for these air mass types and this should be discussed in the text referencing Table 2.
Lines 214-219: When was the dust case identified? NAAPS itself is a model so how can it “confirm” the existence of dust over Manila? Also, HYSPLIT uses reanalysis data and it too depends on a model and a number of assumptions in which is provides a possible transport pathway for aerosols. Do you have surface based measurements confirming the dust reached Manila?
Lines 224-225: How were the NASA Worldview images used for verification? For example, NASA Worldview provides many products. Authors, need to use caution and understand the uncertainties related to MERRA-2 in regards to data assimilation, modeling, and determination of aerosol species. The MERRA-2 data set should not be treated as a measurement.
Lines 252-253: Where is the plot?
Line 278-279: AERONET data at Manila are considerably under-sampled during the months during the Summer Southwest Monsoon between May and October. For example, some years during this period, very few data were collected (e.g., 2013) due to the weather and the changing of the instrument. State in the text or in the figure the total number of observations used for each monthly average.
Line 284-288: Which level are temperature, relative humidity? It should be indicated that low-cloud fraction is from MERRA-2 with cloud top pressure > 680hPa. How is the distinction made between precipitating and non-precipitating clouds? It is very likely aerosols are washed out in precipitating clouds so partitioning by precipitating and non-precipitating clouds is important in the cloud processing assessment. Also, the cloud processing is difficult to determine monthly as these processes occur on the sub day temporal grid.
Line 308: What are the values in the legend and why is the scale so small (i.e., 10^-7)?
Line 334-342 (Figure 3): In Figure 3c, specify that these data are the SDA retrievals. What are the total number of observations and/or days for each monthly averages presented? The total number may explain some of the variations in the plots due to low sampling of AERONET data and quality controls. The Figures 3d and 3e are not clear on which data whisker plots refer to which wavelengths; please specify them on the plot. In Figure 3h, the blue line the ratio of RI between 440nm and 675nm but the red line is the ratio of RI between 440nm to the RI average of 670-1020nm. It seems either the wavelength should be 670nm or 675 consistently through the document.
Line 350-351- MISR monthly averages are based on more limited data per month due to its orbit and measurement technique. Also, MISR over pass is in the afternoon so these data are biased to the afternoon clouds. Does MERRA-2 have the same constraint? Therefore, is MISR the appropriate instrument to compare monthly AOD? Why not use MODIS, VIIRS, or possibly Himawari AOD?
Line 360-362: “0 for particles as large as cloud drops” – the desert dust aerosols can also generate EAE near and below zero. This statement “except for when the coarse mode has a large impact on the angstrom exponent” is not clear and needs revision.
Line 371: EAE values less than 0.8 EAE are less likely to be fine mode dominated and/or impacted by optically thin cirrus clouds (especially in the Southeast Asia/Philippines region).
Lines 458-461: However, Sinyuk et al. 2020 (https://amt.copernicus.org/articles/13/3375/2020/) showed that the AERONET real part of the refractive index is correlated to the fine mode size distribution contribution which makes is a less robust parameter.
Line 465-466: The Authors have presented SSA at four wavelengths from AERONET and not just one. AERONET SSA is computed using the four standard wavelengths (440, 670, 870, 1020nm) (e.g., Dubovik and King, 2000, Dubovik, et al. 2002).
Line 545 (Figure 5): Very interesting plot showing the variations of the size distributions from AERONET at Manila. Can you please provide the related VSD properties (effective radius, median radius, standard deviation, peak radius of volume concentration) for each mode (fine and coarse) in a table or plot in the manuscript or supplementary? Also, what is the average AOD and FMF related to each VSD?
Lines 551-570: The discussion of the VSD peaks needs further clarification. For example, instead of “The average coarse mode peak (0.04 um3/um2) was the highest…,” you could say, “The average coarse mode peak for the volume concentration (0.04 um3/um2) was the highest…” Several other sentences in this paragraph are similarly vague and should be revised.
Line 597-600: What are the total number of AERONET measurements in each cluster? Figure 6 indicates that a total of 1345 VSDs were used for the cluster averaging. How were the corresponding parameters correlated to each of these clusters? Previous plots show mainly monthly averages however individual VSD retrievals and even VSD clusters are not explicitly tied to a month. How are the corresponding parameters in Figure 7 grouped into the clusters? The SDA FMF is not within inversion product so what timing threshold is used to link to the inversion data? Also, the AOD 500nm is measured at a different time than the retrieval unless you are taking the AE and computing AOD 500nm between 440nm and 675nm.
Line 668: Figure 8b, how well does the total AOD from MERRA-2 compare to AERONET AOD at the same time? What wavelength is used for MERRA-2 data in the plot? Please indicate this information in Figure 8b. Also, if this is 500nm AOD then it appears underestimated at the specified times. Figure 8d back trajectory analysis only shows for three days so it is difficult to determine the source region. The Back trajectory analysis should be used between 5 and 7 days to better show the source region.
Line 697: Same as Figure 8 comments.
Line 716: Same as Figure 8 comments.
Line 741: Same as Figure 8 comments.
Line 748: Section 3.6 seems out of place and perhaps should be part of Section 2.4, 2.5, or new 2.6.
Citation: https://doi.org/10.5194/egusphere-2023-197-RC1 -
RC2: 'Comment on egusphere-2023-197', Anonymous Referee #2, 23 May 2023
reply
Lorenzo et al. have explored aerosol climatology over Manila using AERONET database. Besides, MISR AOD was considered with MERRA-2 data and other met. data to explain aerosol type and aerosol movement through wind. The manuscript in most cases lost direction and lack critical analysis, as too many datasets were used without much detail interpretation. Writing of the text also needs improvement.
General comments:
- The manuscript lack of novelty. Aerosol climatology has not been constituted with long term database. Authors have explored too many aerosols dataset without being conclusive on any of these. Determination of aerosol type lacks more analysis and improvement of hypothesis.
- Significant part of case studies is based on NAAPS model outcome which is only used for regulatory forecast purposes and has uncertainty in model forecast over the Philippines.
- Besides, there are some claims that need to be reverified by authors. For instance, availability of AOD dataset and related research over Southeast Asia (SEA) are plenty. SEA is one of the extensively explored regions of the world for aerosols because of its climate significance. However, in the Line 46, authors claim non availability of research on aerosols over Southeast Asia. This is certainly not true.
- Again, August is pointed to be the highest biomass burning month for Insular SEA. However, in fact, Sept.-Oct. is primarily reported to have widespread forest and peatland fires, burned over large parts of maritime southeast Asia, most notably in Indonesia, southern Sumatra and southern Kalimantan. In many reports, the increase in AOD in the Philippines during Oct. is a direct consequence of fire in neighboring Indonesia and corresponding wind movement in Sept. to Oct.
Specific comment:
- There are many contradictory sentences explaining the results. Like in L28, highest AOD in August is contradictory to L325 and L329 with AOD 0.23 in October.
- Introduction needs to be re-drafted. First paragraph does not conclude anything specific besides mentioning some previous research on the Philippines. Please include the findings on aerosols/aod/fire/trend from previous experiments. Content in the second paragraph is more relevant in the study area, section 2. Even the context of conducting this research is very briefly mentioned. Please emphasize on what new science questions can be answered by innovative analyses using combinations of data sets. Just because aerosol climatology has not been reported from a region using a new data product does not hold the novelty of the research.
- Drafting the manuscript in many cases is not appropriate. In L 88, authors emphasized that only AERONET data has been used to study whereas in table 1, it says for AERONET, MERRA-2 and PERSIANN. No discussion on MISR monthly AOD is included.
- 2.1.3 What was the purpose of comparing monthly MISR 0.5x0.5 data against AERONET and MERRA- AOD? This does not conclude anything scientific on aerosol climatology.
- The criteria for considering Table 2 has some ambiguity. Please explain what was the basis of considering FMF to sort fine and coarse aerosols? Again for a country like the Philippines with a very low annual AOD (~0.2), how accurate is it to separate marine and industrial aerosols based on FMF?
- Table 2: Its also strange that no biomass burning aerosols were considered as air mass type when biomass burning is an important contributor. Beside authors are working on smoke aerosol transport in section 2.3.1 while no such classification was made in Table 2. Please justify.
- The NAAPS model outputs are not always convincing enough to detect regional emission sources. NAAPS model aerosol forecasts are available on a 1°× 1° grid. However, use of NAAPS model forecast over Philippines is questionable as ‘number of AOT assimilations available in and around the Philippines is limited because of the pervasive cloud cover, making model outputs of AOT subject to uncertainty for this region (https://acp.copernicus.org/articles/22/12961/2022/)’.
- Section 2.4: Why not EOF was performed on monthly mean MISR AOD instead of a reanalysis product?
- Fig. 1 & 2 is not required in the main text, move it to the supplementary file. Reduce the related discussion on meteorology as this paper is not focused on aerosol -meteorology interaction but the aerosol climatology. Entire section 3.1 should be removed/deleted keeping Fig. 1 and Fig. 2 in supplementary.
- Fig. 3a: trend in AOD is not clear, make adjustments in y axis.
- L 330: The major contributor of biomass burning emission in peninsular Southeast Asia is emission from Indonesia with much higher fire spots in October compared to that of Philippines in March. It's highly unusual to have a greater AOD peak in March compared to October. Justify.
- 3.2.3: Author should add SSA to identify aerosol mass type as in Table 2. Besides avoid explaining individual aerosol optical properties in the result discussion part, instead focus on results. This is true for all sub sections in 3.2.
- Section 3.3.2: Include SSA as an additional parameter to characterize aerosol mass and re discuss the result.
Citation: https://doi.org/10.5194/egusphere-2023-197-RC2
Genevieve Rose Lorenzo et al.
Genevieve Rose Lorenzo et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 190 | 68 | 12 | 270 | 20 | 3 | 3 |
- HTML: 190
- PDF: 68
- XML: 12
- Total: 270
- Supplement: 20
- BibTeX: 3
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1