This manuscript presents a preliminary attempt to investigate whether soil N2O emissions can be inferred from NOx emission estimates determined from satellite data on NO2 columns and a box model. First, they examine aircraft measurements of N2O and NOx downwind of California croplands to obtain central estimates and distributions of N2O:NOx emission ratios. The then apply these emission ratios estimates of agricultural NO emissions derived from calculations of NOx emissions based on TROPOMI NO2 observations. They compare these N2O flux estimates to independent ground and airborne studies in the US Corn Belt and Mississippi River Valley, which are shown to be “broadly consistent.” The authors posit that the method is promising enough to warrant further studies.
I agree that the idea was worth pursuing and could be pursued further. I also feel that this the research was sound and the manuscript well-written, so that this effort can and should be documented in the peer reviewed literature. That said, significant challenges remain to prove that this approach will provide useful estimates of N2O emission, which I describe below. I also have minor suggestions for improving the clarity of the manuscript.
For this approach to be proven useful, it will need to demonstrate confidence in spatial and temporal variation in N2O emissions. For example, the flux estimates in Griffis et al. (2013), based on N2O concentrations measurements at 100m on a very tall tower in Minnesota and accompanying micrometeorological measurements and budgeting analyses, show temporal variation on the order of 0 to nearly 3 nmoles N2O/m2/s for the tower footprint area, with a distinct peak in June and much lower fluxes during the late summer and non-growing seasons. At the cornfield plot scale in Maryland, a micromet study using short towers showed fluxes ranging from near zero to a peak of about 8 nmoles N2O/m2/s following a fertilization and wetting events (Zhu et al. 2023). Wagner-Riddle et al. (2007) showed similar peak fluxes after fertilization and also smaller but important peaks of about 1-3 nmoles N2O/m2/s during spring snowmelt in Ontario. So, the question is whether the proposed method based on space-based estimates of soil NO flux, modified by N2O:NO ratios can reveal variation of similar magnitude. On the positive side, the fluxes shown in Fig. 3 are generally within a reasonable range of about 0-4 nmoles N2O/m2/s. A less positive result is that panels a and b show only means and ranges for a measurement period, and panel c has only a one-point-in-time estimate from this method, so we can’t tell if there was any temporal variation. Furthermore, panel b suggest that we might not be able to distinguish between a flux of 1 and 3 nmoles N2O/m2/s, which would be disappointing. The SI shows that the mismatch could be larger, depending upon the assumed NOx lifetime. Might this factor vary temporally and spatially, thus making the challenge of detecting spatial and temporal variation on N2O fluxes even larger? A necessary test would be to make estimates over seasons, including periods when fertilization is likely and when it is not and between regions that have a lot of fertilized croplands and those that don’t. This is a challenging task, and I understand why it hasn’t been done yet, but it should be identified as a key next step.
I give credit to the authors for revealing all of these uncertainties, and I reiterate that I think this was and still could be a worthwhile endeavor, especially given the lack of alternative approaches for space-based estimates of N2O fluxes. Despite my concerns about whether the method will really pan out, I think that this work should be presented in the literature.
Here are a few suggestions for minor improvements of the manuscript:
Fig. 1: The inset graphs seem to suggest that the ratios are all integers, but the text gives non-integer values. This should be clarified.
Lines 218-220: This is an understatement and it may not be understood by all readers. I had to return to this sentence after reading the results to figure out that the N2O:NO ratios used to estimate fluxes in the Midwestern sites were derived from the study in California. The Central Valley of California is generally drier than the rainfed agricultural areas of the Midwest, so I would expect lower N2O:NO ratios in California. This should be discussed and it should be made clearer later in the manuscript when discussing Fig. 3 results, that the N2O fluxes in the Midwest are based on ratios derived from a study in California. On this point, I disagree with the authors’ response to reviewer #1 in the previous round of reviews. Numerous studies in the literature demonstrate that variation in this ratio is large in both space and time.
In numerous places in the manuscript the ratio flips back and forth between N2O:NO or NO:N2O (I’m less worried about switching among NO, NOx, and NO2 when using this ratio, for the reasons given by the authors in the responses to the previous round of reviews). It would be less confusing to readers to pick whether N2O appears in the numerator or the denominator and be consistent throughout the manuscript.
In the section on future studies, another approach worth mentioning for estimating N2O:NO ratios would be output from a model like DayCent, which bases estimates of that ratio primarily on soil moisture (specifically on water-filled pore space of the topsoil the last time that I checked, which was a while ago). If soil moisture could be estimated simultaneously from space or from models using ground-based data on soil properties and weather, then a spatially and temporally variable ratio (and its uncertainty) could be estimated. |
In their paper, Adams, Plant and Kort propose to use NO2 space observations to estimate N2O emissions from farmlands. The paper is well written, has a good set of references and was a pleasure to read. The comparisons with three very different measurement approaches, as shown in Fig 3, is clearly illustrating the potential of the approach. I have a couple of points which I would like to see addressed before the paper is ready to be published.
The method relies on two key steps, first the determination of soil emissions based on satellite columns, and secondly the link between NOx and N2O emission fluxes. Several questions came up related to these two steps.
In the paper, NOx/N2O emission and NO2/N2O concentration ratios are not always clearly distinguished, but emissions are not the same as concentrations and lifetime and NO/NO2 chemical conversion plays a role. It would be good to be more precise at several locations, and describe in more detail how measured concentration ratios are computed back to emission ratios.
l 122: "we assume all the emitted soil NOx (primarily NO) has converted in the atmosphere to NO2, ..". "The inverse of our ratios is directly comparable to literature NO:N2O molecular emissions ratios."
Why is this assumption made? This is a potential source of systematic error. Normally the concentration of NO2 is larger than NO, but this depends on the chemical regime, distance from the source and availability of ozone. Also the soil NO/NO2 emission ratio may play a role. Using a chemistry-transport model could lead to more accurate results.
l 117: "To isolate cropland regions, analysis is restricted to locations >0.04° (~3.7 – 4.4 km) from regions with emissions in the top 1% of the National Emissions Inventory (NEI) (Strum et al., 2017), and to periods when the aircraft was below 500m elevation."
Pollution from isolated large sources can easily travel long distances (20-100 km). Is this assumption justified and effective in removing non-agricultural contributions? How well are agricultural emissions separated from the other emissions (industry, traffic etc)?
The explanation of the box model, section 4 equation 1, was confusing, and more discussion (maybe even a figure) could be helpful to increase confidence. The first two terms refer to advection. This would require computing gradients along the wind direction: when downwind concentrations are higher than upwind concentrations this indicates that emissions occur. The authors refer to a delta(NO2) as "the mean TROPOMI NO2 column enhancement (molecule/m2) above the background abundance, which we define as the 5th percentile of NO2 abundance in the domain of interest". But this is not the same as a gradient? Please explain more clearly how this is implemented.
The authors distinguish deposition and lifetime. How important is the direct deposition term? Normally I expect the reaction with OH to dominate. Please add some more detail on how the lifetime is approximated.
TROPOMI is analysed on a daily basis. However, box model emission results based on daily observations may be very noisy. Is noise a problem, especially when comparing with campaigns with just a few days of observations. Is this a problem?
The emission ratio results are shown in Fig.1. A very broad range of values is observed, from close to 0 to well above 1. This is an important results, and shows that the proposed methodology will not provide good results everywhere. Basically the authors suggest that these differences average out when looking at larger regions. But is this really the case?
Are N2O/NO2 ratios expected to be similar in other parts/regions of the world? Can the ratios determined for the San Joaquin valley be used in the other domains in central US (Iowa etc) discussed in Fig.3 ? As mentioned, ratios will depend on moisture, vegetation and soil type, vertiliser use which may vary from one region to another and is also time dependent. This may potentially cause significant biases in the results for a given target region.
In the conclusion: "As presented here, the largest source of uncertainty in the estimated N2O emissions derives from the large variability in the observed airborne N2O:NOx emissions ratio. Improved understanding and definition of this ratio, and what controls variation, could improve the fidelity of this proxy approach. "
This seems to point towards potential improvements in the methodology. If parameters like soil type, land, moisture and rainfall could be correlated with the emission ratios then this could improve the emission estimates and generalise the method to other regions. Please add some comments at the end of the conclusion how the method may be improved in the future.