The hard NOx life for crops

6 min readJun 2, 2022

The recent jump in grain prices is another reminder of the importance of having high cropland productivity. Although the Ukraine war has pushed markets over the edge, the underlying conditions were already strained, in part due to weather-driven shortfalls around the world. As scientists try to identify ways to boost productivity in the face of ongoing climate change, all options should be on the table, even those we haven’t historically paid much attention to.

One option that has gotten more attention from scientists in recent years is to improve air quality in major cropping regions, especially those in Asia that are typically exposed to high levels of common pollutants. The basic idea is that if pollution is hurting yields, then removing that pollution will lead to higher productivity. But there’s still a lot of uncertainty about how much we can really expect from improved air quality. It’s common to see estimates differ by a factor of two, and often by even more.

One reason for this uncertainty is that it’s traditionally quite hard to study how much specific pollutants affect crops. Ideally you’d prescribe different pollution mixes and levels, and varying weather conditions, over thousands of identical fields, and see what happens. That’s not possible, so what’s next-best? You can run an experiment in a greenhouse, but then extrapolating that to actual fields requires a lot of assumptions. Or you can try to relate real-world measurements of pollutants to observations of crop yields. This approach has two main challenges - one is that other things could vary with pollution that also affect yields, and another is that the places we measure pollution are often close to cities, far from most of the places we grow food. Matching up the two datasets then requires assumptions about how pollution in locations where it’s measured relates to locations where it’s not.

In a recent paper, we try a new approach - using satellite measures of both pollution and crop growth. This resolves the issue of having measurements at different locations. And because we can use satellite measures from around the world, it greatly improves our ability to minimize the chance that unmeasured factors are driving the yield response.

Why haven’t people done this before? The main reason is that reliable, high resolution satellite measures of pollution are a recent phenomenon. The TROPOMI sensor on board the Copernicus Sentinel-5 Precursor satellite is particularly impressive. It is measuring a suite of short-lived pollutants and longer-lived greenhouse gases, based on the principle that each gas has its own unique absorption spectrum.

Our study focused on nitrogen dioxide, or NO2, which is the main contributor to total nitrogen oxides (NOx). NO2 was attractive for two reasons. One is that it is likely one of the key pollutants relevant for crops, especially when you consider that NOx is a major determinant of ozone formation. Second is that NO2 is one of the best-measured pollutants from satellites, because it has such a unique interaction with light - as shown in the figure below taken from the TROPOMI algorithm explanation. Notice how the absorption profile of NO2 has lots of unique ups and downs compared to other gasses. The algorithm used by TROPOMI uses this signature to estimate NO2 concentrations from the measurement of light in many narrow bands along the ultraviolet spectrum (each band has ~0.5nm spectral resolution). So even though NO2 is completely invisible to the naked eye, TROPOMI can measure it very precisely.

TROPOMI measures NO2 on a daily basis, so our first job was to summarize how much a crop would be exposed based on the location and time of year the crop is grown. The figure below shows the average exposure for major wheat growing areas during the winter season. (Map shows average values for each pixel, and the right shows the histogram for each region and season). The highest exposures are unsurprisingly in China and northern India, but you can also see some nearly as high values throughout Europe. North and South America tend to have lower exposures, at least in the winter season.

Summer crop exposures tended to be slightly lower on average, but still fairly high, especially in China

The next step was to relate these exposures to crop outcomes. In this case we use a measure of crop greenness called NIRv that is well established as an indicator of crop growth (see paper for details). We then compare local variations in NO2 to local variations in NIRv. We define “local” here as a 0.5° latitude x 0.5° longitude area (roughly 50km x 50km). The figure below shows an example for China, with the grid lines on the leftmost maps indicating the grid cells.

The reason to use local variations rather than the raw data is that it reduces the chance that some other factors, such as weather or farm density, produce a spurious correlation between the two. Since this doesn’t completely eliminate the chance of a spurious effect, we also conduct a series of robustness checks which are explained in the paper and supplemental material. The bottom line is that the types of negative effects shown in the figure above were found consistently across different seasons and regions.

As a final step, we estimate the potential gains that would occur if NO2 could be reduced to background levels, which we defined as the 5th percentile of the observed values in each region. As shown below, the estimates are largest in China, but also close to 10% in Europe and some other regions and seasons.

Overall, the results support the idea that cleaning up air quality would be a substantial boost to productivity, and help to put more precise estimates on the potential gains. While it can be hard to imagine a rapid change in the economic forces that drive NO2 emissions, many regions (like the U.S. and China) have already shown progress. In fact, the decrease in NO2 could be one factor behind the observed greening of Asian croplands (we discuss this in the paper, but don’t directly evaluate it). And efforts to decarbonize energy and electrify transportation, among others, offer the promise of even bigger changes in the future. Given all the stresses on the food system, I view this as pretty good news. You might even say it’s a breath of fresh air.

About the Author: David Lobell, Gloria and Richard Kushel Director of the Center on Food Security and the Environment at Stanford University

David Lobell




Stanford's Center on Food Security and the Environment (FSE) leads cutting-edge research on global issues of food, hunger, poverty and the environment.