the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Hybrid Instruments Network Optimization for Air Quality Monitoring
Nishant Ajnoti
Hemant Gehlot
Sachchida Nand Tripathi
Abstract. The significance of air quality monitoring for analyzing the impact on public health is growing worldwide. A crucial part of smart city development includes deployment of suitable air pollution sensors at critical locations. Note that there are various air quality measurement instruments ranging from expensive reference stations that provide accurate data to low-cost sensors that provide reasonable air quality measurements. In this research, we use a combination of sensors and monitors, which we call as hybrid instruments and focus on optimal placement of such instruments across a region. The objective of the problem is to maximize a satisfaction function that quantifies the weighted closeness of different regions to the places where such hybrid instruments are placed (here weights for different regions are quantified in terms of the relative population density and relative PM2.5 concentration). Note that there can be several constraints such as those on budget, minimum number of reference stations to be placed, set of important regions where at least one sensor should be placed and so on. We develop two algorithms to solve this problem. The first one is a genetic algorithm that is a metaheuristic and works on the principles of evolution. The second one is a greedy algorithm that selects locally best choice in each iteration. We test these algorithms on different regions from India with varying sizes and other characteristics such as population distribution, PM2.5 concentration, budget available, etc. The insights obtained from this paper can be used to quantitatively place reference stations and sensors in large cities rather than using ad hoc procedures or rules of thumb.
- Preprint
(719 KB) - Metadata XML
- BibTeX
- EndNote
Nishant Ajnoti et al.
Status: open (until 21 Oct 2023)
-
RC1: 'Comment on amt-2023-173', Anonymous Referee #1, 20 Sep 2023
reply
Overall the paper is well written and clearly presents its methodology and results.
I find that there is a conceptual problem with requiring emissions information for PM2.5 as an input to your optimization framework, when, in the absence of a monitoring network, this may not be well quantified (as is the case for the Mumbai example you present). You may want to discuss this as a potential limitation of your work and present some possible solutions, e.g., using existing model or satellite-derived information as a proxy for local concentrations during the network design phase, and/or iteratively chancing the network as newly collected data update the prior estimates of concentrations in the different grid cells. Furthermore, there is a potential disconnect between PM2.5 emissions and PM2.5 concentrations (which are to be measured by the network), with the possible impacts of secondary aerosol formation and pollution transport not being accounted for by using emissions information alone; maybe emissions are being used as a proxy for concentrations, but that was not clarified in the text.
Similarly, while you clearly state that you are aiming to optimize public satisfaction through your sensing network design, there are many other potential objectives which might be the goal of a monitoring network, e.g., improving estimates of population exposure, monitoring the largest known sources, etc. I would suggest adding some commentary to your conclusions discussing how your approach might be modified to achieve these other objectives.
While you note that sensors and monitors have different capabilities, in your formulation they are treated equally in terms of your utility function (i.e., people will be equally satisfied to be located near a monitor or near a sensor). Could you justify this further, or discuss how your results might differ if monitors were given a higher weight?
Specific Comments
Line 10: Use of “reasonable” here is a bit unspecific; I suggest “less accurate” or a similar description instead, to contrast them with the reference stations
Line 11: Remove “as”
Line 17: “selects locally best choice” should be “selects the locally best choice”
Line 27: Remove “the”
Line 54: “limitations that” should be “limitations in that”
Line 56: Suggest replacing “in the previous-to-previous paragraph” with “previously”
Lines 60-61: The distinction between sensors and monitors has already been defined earlier in the paper
Line 64: This definition for hybrid instruments has already been stated
Line 67: “noble” should be “notable”
Line 70: “Therefore, following” should be “Therefore, the following”
Line 77: “Next section” should be “The next section”
Line 80: “of hybrid” should be “of a hybrid”
Line 84: Consider restating the objective to better explain “people satisfaction”, e.g., “Our approach focuses on placing sensors in order to maximize a utility function quantifying popular satisfaction with the sensor placements”.
Line 87: “g(d) be” should be “g(d) must be”
Equation 1: describe how the parameter theta is set
Line 101: While emissions have an influence on local PM2.5 concentrations, secondary aerosol production and pollution transport also play a role.
Lines 106-107: Move this sentence right after the first one in this paragraph.
Line 110: “where monitor” should be “where a monitor”
Equation 4: It is not clear why sensors deployments should be required, but monitor deployments should not be.
Equation 5: Similarly, it is not. clear why monitor deployments are restricted, but sensor deployments are not.
Line 115: Please define d(a,b).
Line 129: “or” should be “of”
Line 156: ”carried” should be “carried out”
Line 164: describe how the parameter alpha was chosen
Line 171: I believe that a maximization expression is missing in the equation here.
Line 175: “reduce” should be “subtract”
Line 179: Stopping criteria are not described for the greedy algorithm.
Line 190: Please provide citations or links to the World Bank and TERI datasets used here.
Line 220: “of the Mumbai” should be “of Mumbai”
Line 224: “maintained consistently as above in” should be “the same as in the example for”
Line 225: ”we have larger number” should be “we have a larger number”
Line 244: “solution is” should be “solution that is”
Figure 5: It is unclear how the size of the grids is being varied; Is this the same example for Mumbai? Are the sizes of grids being reduced, or is the area of coverage being increased?
Table 1: Describing g(d) as a function of d is not very informative; consider expanding the description and referring back to Equation 1 for the definition.
References: It appears that a citation is missing for Lerner et al. 2019
Citation: https://doi.org/10.5194/amt-2023-173-RC1
Nishant Ajnoti et al.
Nishant Ajnoti et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 225 | 51 | 8 | 284 | 6 | 7 |
- HTML: 225
- PDF: 51
- XML: 8
- Total: 284
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1