The structure of crop hydroclimate vulnerability and its implications

I'm going to introduce Jennifer M. Bernie from Stanford, who's been doing or working in the intersection of food security, climate change, like hydrology, and kind of like connect all of that together to see the impact of these things on humans. Yeah, welcome, Jen.

Thank you so much. Um, and please, uh, just, um, it's really nice to be here, but please audibly interrupt 'cause I, I can't see you all. So, if there's anything you need clarity on, let me know. Um, yeah, it's a delight to be here today. Um, I'm going to talk about, um, what we're learning about crop hydroclimate vulnerability. Hopefully with some interesting, uh, data for this wonderful crowd, and, um, and some of its implications for how we think about sustainability.

I thought I'd step back and just say a second, uh, or two about, uh, what we do in my research group. Um, so, if that we can be of help to you guys in the future, you know where to find me. Um, you know, anyone in the sustainability and kind of climate world is somewhere in this diagram. Uh, my group is doing a lot of work thinking about, um, emissions, the physical impacts of those emissions, the impacts on human systems, and how we think about adaptation and policy. Um, and in particular, thinking about making sure we're not maladapting, creating, um, kind of negative, um, or negative impact, um, amplifying feedback loops in this coupled system.

And we do this in two areas, really. The first, which I'm not going to talk about today, but we do, um, work on short-lived pollutants, like how aerosols in particular behave in the system, thinking about where they come from, building, um, harmonized inventories, understanding their impacts. We've done a lot of work on pollution on, on crop productivity, impacts of pollution on crop productivity, and then how to think about consistent policy that understands the, um, climate-mediated impacts of these compounds as well.

Um, the bigger part of what we do, however, is actually thinking about what food and agriculture look like in the system. So, we have built emissions inventories for agriculture, food, and land use. Um, we do a lot of work trying to think about climate and air pollution impacts on ag, as I mentioned earlier, but, but really, and this will be where I dive into today, thinking more and more, what do we know better in the climate world? We're learning much more in recent years about seasonal and sub-seasonal, um, variability and how that is likely to change. Uh, we're learning with greater confidence thanks to sort of improved computational tools. Um, and so we've been thinking a lot about, uh, what we can learn about really detailed seasonal variability and, and how that's changing and why that might matter for food security. And then we've been thinking also about how do we now reconcile that improved knowledge with, um, with better adaptation and policy decisions. So, a whole range of field projects that I won't touch on today. We'll stick with the, the data, uh, the data side of it today. But, but that's what we do. So, if we can ever be of help, please, please reach out.

Um, today the outline is, you know, I'm going to tell you a specific piece of this ag climate story that we've been putting together. Uh, and I'll do this in kind of two and a half parts. Let's say three sections, but the last one will be last. The first is last one will be short, excuse me. Um, the first section is we're going to talk about where crop water comes from and, and whether or not it matters. Um, this is a sort of interesting story that started from a just a very simple question about how much water is coming from oceans versus land and, and does it mean anything or does it matter? Um, I'll show you that it is a very interesting story and it, it seems like it's a nice hook into the system to maybe, um, think about how to prioritize some investments.

The second piece I'll talk about is, um, is ongoing work looking for evidence of maybe maladaptation. This will be linked to, uh, the crop water story. Um, and, you know, if you have something like evidence of maladaptation and sort of an understanding of the physical system, then you can really think about how to, how do you do policy, uh, better to take that into account. And so, I'll end, um, with, uh, with some thoughts there.

Okay. So, the motivation for really all of this work in, in my group is, is a few fold. Um, I'll just say a few words, uh, about that to, to tee up the data story. The, the first motivation is really just the, the story about global food and nutrition security and that is, you know, that, um, in recent years we've seen decades of progress reversed. Uh, we were on target, you sort of in the mid-2000s to meet the World Food Summit goals of of having the number of hungry in the world, but we have seen stagnation and then, and then reversal, um, in recent years and some of that is, uh, conflict and violence. Some of it is, you know, bad policy. Some of it had to do with COVID, but there's a pretty broad consensus that climate is playing an increased role here and really understanding how to think about, um, adaptation to seasonal and sub-seasonal variability, uh, it is really an important piece of it.

The second motivation is actually the policy environment has changed quite a bit. Uh, we're sitting here in the United States, so we might not feel it, but globally, you know, we have actually made progress. A decade ago it looked like we were staring down the tunnel at four or five degrees of warming. Now it looks more like two and a half to three and a half. So, you know, that's progress. It's not enough, but it's, it's better. Um, and for a long time the policy world correctly focused on fossil fuels and, you know, industrial, industrial emissions, energy emissions, transportation emissions as the, the sort of correct area to, to work on mitigation, but since progress has been made, we now have to start think about thinking about other sectors and decarbonizing food is, is a really tricky one. Um, we just flat out don't know if we can do that. Um, and, uh, but, but we've started to see pretty strong, um, road maps laid out by all the big international organizations. You know, sort of zero gross deforestation targets, how do we think about decarbonizing food? And so, it becomes really tricky. How do you think about meeting tight targets but tight climate targets without sort of exacerbating this hunger problem that's already recently moved in the wrong direction.

The third motivation is again that that idea about wanting to avoid bad feedbacks and, and again related to what I just said, but if you look at the dynamics in these 1.5 or 2C compatible scenarios, I just want to show you a little bit about what these look like to, to help contextualize the, the problem framing what we're talking about today. If you look at, at, at just a few variables and the assumptions made in these, in these, uh, in these models, um, world population typically doesn't look as bad, so that's the upper left there. Um, the red dotted line is actually the UN estimate. So, most of these scenarios assume lower population. If you move to panel B, um, they assume that agriculture, forestry, and land use emissions, so that's that acronym AFOLU. They assume that we've kind of already reached net zero or going to go net negative very soon. Um, at the same time, they allow for like an increase in cropland. And all of that is possible in these scenarios because there's this massive increase in forested area. And there's really no evidence that any of this is happening. So, I'll come back to this, but it's, it's quite optimistic. Um, and, you know, it's to just frame it with some order of magnitude numbers, if you think about sort of a 20% increase in population, sort of mid to end of century, um, you know, if you, if we meet those needs with sort of expansion of cropland alone, that's, you know, way beyond the remaining carbon budget to stay below 1.5C and most of the remaining carbon budget if you wanted to stay under 2C. So, there's a real like issue here that's that's masked a little bit in these optimistic scenarios that suggest that we can do this. So, there's a lot of work to be done.

Um, and then the last motivation is thinking about like how we think about carbon policy in that environment. So, if you think about something like the social cost of carbon, um, we tally up all the, um, the harms that would likely accrue from the emission of the next, uh, you know, say ton of carbon, and we know that in a marginal cost and marginal, marginal benefits framework, we should be willing to pay enough to mitigate the damages of that next emission. Um, that's a little bit straightforward when it comes to, uh, there's, I mean, there's tremendous amazing science underneath it, but the concept is straightforward when it comes to, uh, fossil carbon. You're sort of moving carbon from the, you know, deep earth to the atmosphere. When we think about terrestrial carbon, it's a little trickier. There's, um, kind of known benefits. We get food, for example, if we can convert land to, to agriculture. Um, we also get emissions that we have to think about, and there are sort of a diverse array of emissions. We sort of know that we impact the carbon cycle. So, this concept is often called the carbon opportunity cost. So, if you change what you're using land for, if you deforest, for example, not only do you lose that carbon immediately, but you also alter future carbon uptake potential. So, you're changing the balance out into the future. And then, when we think about terrestrial carbon cycle, when you alter that, um, there's a bunch of impacts not just through the carbon that you put in the atmosphere. So, uh, in particular today, I'm going to pinpoint, um, this, this pathway through hydrological changes. So, when we think about land use change for food, what are we doing to downstream hydrology? Um, and we'll focus on the atmosphere here, but, uh, but the concept is there. So, you know, if we wanted to think about this, we, we can think about the social cost of carbon, we can think about this carbon opportunity cost, but then there's these other unknowns that we're trying to, to put into this picture.

All right, so with that, I'm going to start into this, this first part about where our crop water comes from. If you're interested in reading the paper, this is work, um, led by my fantastic postdoc, uh, Dr. Yen Jing, and, um, it was published, uh, a few months ago. So, I'll just show you this first part that's been published and I'll show you what we're working on, um, still in progress in the second half.

So, again, as I mentioned, this started from a very simple question, which is we know that most of the world's crops are rain-fed, and rainwater originates either from evaporation of oceanic water or evapotranspiration of water from land. Why does this matter? Like, on a basic plant physiology level, the water's the same, they don't care, the plant doesn't care. Uh, but it matters because we think that the climate futures of these two sources could diverge dramatically.

All right, from very simple physics, right? We know that, uh, evaporation from oceans is intensifying, and it will continue to intensify with a warmer atmosphere, um, turbocharging some of our regular circulation patterns, um, but evaporation from land really depends on how land use evolves. So, if there's less plant life transpiring water, there's going to be less, uh, land-originating moisture.

So, to do this, we used, um, observations of isotopes, uh, in the atmosphere, uh, from a few, uh, NASA instruments, the, the TES and AIRS instruments. And to do this work, we looked at observations of deuterium, or heavy water, in the atmosphere. What do these isotope observations tell us? Uh, water vapor isotopes, uh, naturally have different absorption spectra, so you can use observations to infer atmospheric abundance. Um, and one of the nice things is that standard ocean water contains naturally occurring deuterium with a very well-understood mixing ratio. But then because, um, you know, HDO compared to H2O is, is heavier, it's differentially evaporated, condensed, and precipitated. So, as you measure the abundance or the depletion of deuterium relative to ocean water, like along sort of hydrological pathways, you can infer something about how far that water has traveled. So, again, we sort of start from this standard concentration in, in ocean water. And then if you look jointly at sort of the, the relative depletion of deuterium and moisture content along with some basic theory, you can think about understanding in general where water is coming from. And the punchline here is that like if it's more depleted, like more deuterium has, has precipitated out, that water is likely from the ocean and further away.

So, I'll walk you through just a little bit of what that looks like. Um, again, we, we, uh, generate this number. This is the deuterium depletion, and it's this just sort of the change in the abundance compared to oceanic abundance. And we look at the atmosphere not right at the surface, um, you know, as anyone who lives kind of in the, I'm presuming a lot of you are in the Bay Area, but we get a lot of, you know, you can literally see the daily recycling at the surface level. So, we go a little bit up from that and, and sum the, the more weather-driven, um, water content higher up in the atmosphere. And this is what that looks like. Um, uh, there's some boxes around regions that I'll show you in a minute, but this is just a long run, kind of climatological average, um, over just under, you know, a decade and a, just over a decade and a half, um, and the darker colors indicate more, um, uh, depletion, and the lighter colors indicate, um, you, you're of a higher, uh, higher, uh, land dependence.

So, let me, let me show you what this looks like, um, in detail in some of these regions. So, I'm going to show you some paired diagrams of, um, that deuterium, uh, level, that deuterium depletion, and water vapor quantity. And a lot of panels on this figure, but, but you'll see really quickly that they're pretty simple. So, these are in pairs by regions. You got North America in the upper left, India in the upper right, um, South America chunk of it in the lower left, and, and the Sahel in the lower right. And these are now, uh, atmospheric pressure layers. So, you can think of the bottom of each plot as the surface level, like land surface level, and the upper troposphere at the top. And showing you just over the course of a year what the average, uh, deuterium depletion levels and water vapor content, uh, water vapor mixing, uh, values, uh, look like in these regions. And we can take these and jointly look at, um, the, the deuterium levels and, um, and moisture amounts in the atmosphere and compare them to compare them to some theoretical curves about what this would look like.

So, a few of these are, are shown here. Um, this purple curve, uh, we can think about something like a Rayleigh distillation model. You got perfect precipitation happening from oceanic water. You're going to see like this sort of steeper, deeper, uh, depletion levels, um, in that curve. If you look at this, uh, pink curve, that's more like a, a modified, uh, moist reversible adiabatic process model where this is where all the water is condensed and stays in the air, but you don't have any precipitation. And then if you look at something like an ideal mixing model, these are these are models where two types of, of, of air are mixing. Um, one where, uh, the, the origin is from ocean in blue and one where the origin is from land and green. Um, you can look think about sort of what theoretically these, these values would look like. And QF and DF here are are surface values. How much, uh, deuterium is in water evaporating either from the ocean surface or from the land surface. So, you get all these theoretical predictions of what curves could look like and then what we can do is take one region. So, I'll just show you a little snapshot from India here. And if we look at like India in April and integrate over these windows, we can look at what those values look like. We do this over say a few days to avoid all the super fast cycling, um, water and, and we end up with something like this. So, if we compare to what the land mixing model looks like, that's kind of an upper bound on water that that likely originated from land. So, we can calculate kind of the fraction of the water observations that lie above this curve and that's a pretty good estimate of the fraction that originated from land. So, that's a lot of, a lot of hydrological detail that that maybe just leads to this important point which is if you're above this theoretical land curve, we're going to assume that water came from land and this is going to give us this number F that I'm going to refer to the rest of this talk, which is the fraction of water that comes from land sources. So, in this case, in India in April, you have a very high fraction, like 77% of water on average coming from land.

Now, so, question, I showed you, Oh, yeah, please. Yeah, you, your models are pretty like location independent. So, I'm thinking of India, Himalayas. I should be thinking north of India. The Himalayas are a big barrier. Definitely would modulate this phenomenon, but then you're not accounting for that. So, to what extent is terrestrial features a thing here? Yeah, absolutely. What a fantastic question, 'cause it's definitely the next slide. Uh, so, if you look at these mixing, ideal mixing model curves, this is, this is sort of where we're going to anchor. You got to have terrestrial values to pin down the land numbers. And where do we get those? We get those from this network of surface isotope measurements. Sorry, it's turned out a little blurry in the screenshot. Um, but we, we get those delta D values from location-specific measurements that tell us a lot about, uh, you know, the, the locate, the sort of geographically modulated location-specific information. So, you can see for example, India and Europe and, uh, you know, South America look very different from each other. So, wait, so that's to get the dynamics? So, you're saying the, what's it called? The, the isotope ratios, is this where the isotope, like the sources of this particular mixture of isotopes, like as measured in those sites, or it's like transients as they see water passing by from the ground level, that's what they measure? Yeah, no, these are longer run averages. So, we're trying to get, um, really more just like the geographic spread of what we think the boundary condition for the, for a mixing model should look like. So, again, Thank you. Yeah, right now, everything I'm going to show you right now is really more of a long run F, like just what's the structure of the hydroclimate dependence look like. It is true that there is some, uh, some differences, of course, at faster scales, but this is going to be like a climatological F. And these, it sounds like we should think of it as air passing by these points and the measurements of deuterium ratios at those points in, in land, in this case. Yeah. Yeah, so these are like stream gauge measurements, land water source measurements that just tell you what the likely concentration is, uh, from water being evapotranspired from these areas, and then we put that into those ideal mixing models and think about what if that's mixing with oceanic air, what would we expect to see? Got it. Oceanic air has a known distribution. Yeah, exactly. Got it.

So, if we look at those calculations around the world, again, these are long-run, um, annual averages, um, but there's quite a bit of heterogeneity. Uh, and some places, you know, the upper Midwest in the United States, um, Central Asia have very, very high annual F values. Um, and, uh, and this is hiding a lot of seasonal, um, within-season heterogeneity that I'll show you, but, but there's just quite a bit of difference around the world. Um, but we all know, uh, annual values don't, you know, they matter, but they don't matter that much. We're really interested in thinking about how this intersects with when we want to grow food. And so, the thing that we, uh, then do is we use global observations of of vegetation indices. In this case, we're using, um, an ARV, and we use these to define the main growing seasons, kind of in each location coarsely. So, this is these are just MODIS data. Um, the green line here shows you, for example, what the seasonal curve of ARV might look like. Um, we can just coarsely fit that curve with, you know, something like a back model. Uh, and then when you just detect the rates of change in that curve, that turns out to be a pretty good proxy for, um, the growing season, um, portions that really matter. So, this first part in green is going to be, uh, sort of the vegetative period, the part in yellow, uh, is really the most sensitive period, the reproductive period, and then in brown, we have sort of, uh, senescence and harvest. So, we can define crop phenology this way and think about where, um, where we're more sensitive as well.

Um, so, here's what that looks like, uh, around the world. So, a few different isotope data sets here, the AIRS, um, and TS sets, and then, um, uh, the dotted orange line is the AIRS set truncated to the years that tests was making observations. So, to just compare across them, they're pretty consistent. Um, and you can just see over the course of the year, there's quite a bit of change in this, uh, this F number, the fraction of rainwater that's coming from land. And, um, and it coincides differently with the growing seasons, uh, the main growing seasons across these regions.

If you combine this information, so it's one thing to think about a percentage, but what we really want to know is, is kind of amounts of water also. If you combine this with, um, with total precipitation information, we can also think about, uh, when the peak of land originating moisture happens in a growing season. Um, and one thing that stands out, I'll just draw your eye to kind of India, third, uh, on the top, and, uh, maybe as the most important example there, but it's a strongly monsoonal system. Um, but we know, um, from all of our, you know, land hydrologist colleagues, that they do think that terrestrial moisture plays an important role in kind of kicking off that monsoon cycling, and you can kind of see that here, that the peak of the terrestrial water amount, like in actual millimeters now, um, it sort of hits at the beginning of the season and then the, sort of oceanic driven monsoon rainfall takes over.

So, I just show you this out of like pure delight at these data. They're kind of interesting. Um, but what really was amazing to us, uh, this is like the, the fun part of the story, is that there's this kind of natural division that just jumps out from this work. And so, um, there's a lot in here, but I'm showing you kind of coarsely gridded global data of the distribution of that F number over the main growing season in a region. And then, um, plotted in, in the color is going to be your seasonal water, um, supply value. If you're in blue, that means you have more precipitation than potential evapotranspiration, so you have like adequate rainwater supply. If you're red, that means that you've got a deficit and you're probably drawing down soil moisture. And there's this, this pretty nice, um, division that emerges around about 36%, where, um, just above that, if you're more than 36% dependent on, uh, water from land, you're just more likely to have water-stressed cropland. And again, it's not because there is, uh, anything different about water from the ocean or water from land that that matters for plants. It's really this interesting reality again, that's sort of just jumped out at us, that F, it turns out to be encoding a bunch of important subseasonal dynamics.

So, uh, if we look at these, um, uh, growing, this is now, uh, all the cropland that we looked at around the world, but with its growing seasons stacked up. So, not, uh, not by calendar year, but really by growing season, all aligned. And there is this flip right around, um, uh, whether you're looking at water supply, so P minus PET, or whether you're looking at root zone soil moisture. There is this flip right around, um, uh, 36. You know, if we bend these, it, it, it's, you know, there's a little bit of wiggle room in there, but, uh, right around this F value, where you start to see above it, you're getting, um, rainwater, uh, insufficiency and you're starting to draw down soil moisture over the growing season. Below it, you tend to have enough water and abundant, uh, root zone soil moisture. So, it's pretty interesting. You just have this one heuristic value about where water's coming from, but it really tends to line up very well with this whole rich set of subseasonal dynamics. Uh, and one thing that's interesting is that it does a much better job of this than something like say, the aridity index. So, this is doing the same thing for those croplands, but just grouped by aridity, and we didn't include the, the hyper arid and, and really arid areas, but just even moving from very humid to semi-arid, you don't see that same sort of distinction about soil moisture dependence during the growing season. So, this contrast is really what surprised us the most, that you have this very simple metric that that tends to tell you a lot about hydrology that you wouldn't maybe otherwise know to measure.

And so, just, can you give a little bit of a deep dive into the RGS and measure? Sorry, can you say that again? Can you go just deeper into what these two metrics measure again? >> Sure. The aridity index is just sort of a standard, you know, heat-based, heat and moisture-based index. And then the root zone soil moisture is is using satellite and, and reanalysis products to, to estimate that soil water balance. Cool. Thank you. Yeah, no problem. I appreciate the question.

The other thing is that there's, you know, there's a pretty strong relationship then between water availability and crop yields, of course. If we just use, um, greenness, you know, maximum NIRV as a, as a proxy for crop yields, it's a pretty good proxy. And then we just use a very simple model of like water supply during the growing season, during the vegetative and reproductive parts of the growing season, and we just look at water supply and root zone soil moisture, the sort of partial correlations, and actually the total R squared, which I'm not showing you here, but, but you can infer it from the partial correlations. You can explain, you know, 60, 50 to 60% of water variability, sorry, of yield variability with just these simple metrics at high F levels. So, you, we start to have some, some nice, just heuristics for crop yield and its dependence, you know, what's driving crop yields in different locations and honestly, some of what my group has thought about over the past a few years, it's just like when and where, like there's this kind of mystery, mysterious finding of like, it doesn't sometimes look like precipitation matters all that much for crop yields, but in some places it does and, and, and we think that this is like a big part of explaining why.

So, do you have any intuition for why, 'cause the source should be irrelevant, but clearly is, you know, correlates with something really critical. Yeah, I think it's just that, you know, you, it's something, it's fundamental circulation, kind of, it's the, the structure of hydroclimate on the planet. You get more inland in a lot of these places and you rained out your oceanic moisture, right? And I think it really is, is just basically like as we have sort of a characteristic length that oceanic water travels and when you're beyond that, you're more dependent on land and when you're more dependent on land, you have some really interesting things that that can happen. So, let me, let me just tell you about one of them. It's my, you teed up all my questions perfectly. Thank you. Yeah.

So, that, you know, the other thing we see, sorry, these are overlaid, but if we just look at a map of now, just when we have sort of agricultural droughts measured by just precipitation, sort of standard metrics, um, and we then look, sort of again, about the occurrence of these droughts and how they relate to to F. Um, I'm not showing you all the data here, but there's a pretty strong relationship between the length of a drought in a growing season and the F number. And that, I think, starts to make sense when you think about sort of soil moisture depletion and you get the sort of self, self-propagating, um, feedback. Um, but when we look down in this lower right, at, um, at, you know, sort of when you have evaporative stress indices, so, it's just an independent metric that that says you're in a drought, these deep red bars, and you look at cropland, um, you know, with F less than 36%, um, you still see negative impacts, right? Droughts are bad. Um, but when you look at that land-dependent cropland, when it has the same kind of drought, the impacts are much worse. And I think it's because of those subseasonal dynamics. You've already depleted soil moisture much more in those areas, um, over the, over the middle of the growing season.

So, uh, what we took from this was just, uh, you know, two things. A lot of global cropland has has high F. So, something like 40%, uh, of major crops, 50% for, for, uh, wheat. Um, so, there's, you know, there's a lot of our breadbaskets are in this kind of region. Um, and this F number is a really nice heuristic in that it's sort of a long-run climatological value, but it tells you a lot about subseasonal dynamics. And so, therefore, it provides some guidance on where you might think about, uh, prioritizing irrigation or water storage, um, because it would sort of pay the highest dividends. Um, we're, we're trying to work with now, um, getting some isotope-enabled Earth system modeling output saved so we can, uh, track and validate some of this, but, um, you know, we've, we've learned and, and had a great time learning about how these, you know, isotope observations, uh, tell us this really interesting agricultural water story. Um, you know, and that these high F areas are more likely to be water stressed, they're more sensitive in terms of their yield responses to water availability, uh, they're more likely to experience growing season droughts and longer droughts, and they're more sensitive when those droughts happen. So, you know, that's like a, like interesting in and of itself, but it's also really important when we zoom back out and think about the world food system and that this problem that we're trying to solve of, you know, sort of sustainable food in a warming climate. Um, upstream of evapotranspiration is really important for a lot of global cropland. Um, and the implication here, of course, is that reductions in any sort of reductions in evapotranspiration from land are going to matter for downstream cropland. And again, this comes back to this policy question, right, of like how do we think about carbon from terrestrial sources? Um, and, and we can think about land playing really three important roles of, uh, in, in the climate system in terms of its carbon, right? It's, it's a terrestrial carbon stock. The amount of carbon in the terrestrial system, like modulates the annual, interannual sequestration of carbon, and now, as we're sort of showing in more detail, it's a water source for downstream crops and ecosystems. And, um, and in particular, thinking about sort of South Asia and the Sahel as very sensitive areas that that have some of this very high early season dependence on, on land-originating moisture.

So, we took that and, and move into the second part here, but thinking about deforestation and land use change, um, you know, if that changes evapotranspiration, uh, and terrestrial recycling, it's going to affect downstream crop production, right? We know that these high F areas are sort of disproportionately sensitive to the amount of water, and so we can, we can start to look at, at whether that's happening.

All right, so I'll, I'll zip through this. Uh, this is ongoing work, but, uh, I'll focus in on, um, sub-Saharan Africa, again, thinking motivated by sort of the, the, the Sahel in particular. If we start again with that F number, um, you can see like the Sahel doesn't jump off the map, right, as, as sort of an annual, at the annual level, but it is very important seasonally. And it is, um, a pocket where some of the most precarious farmers in the world live. What we've done is now take like a source receptor model. This is a, a moisture tracking model and this is a, yeah, just a linking of sources and receptors that says, you know, water in grid cell A likely originated from all of upstream grid cells, you know, X, Y, and Z. Um, and so we can think about defining a precipitation shed. Um, and this has been, so here's, here's an example of, of what some of that looks like. You can see that, you know, you have to pick a threshold, but here we, you know, pick some small, you know, hundredth of a percent, um, two hundredths of a percent, and you can see what say the precipitation shed of that, that little yellow grid cell looks like. Um, and this has been used, these sort of source receptor models have been used, for example, in this recent report, like the Global Commission on the Economics of Water, to sort of talk about places that are linked in this way, but you really actually need something like the empirical observations that I showed you earlier to pin down those numbers.

So, if we think about, um, precipitation in a given location, it's going to come from a few different sources, right? You can think about like a very local amount, like that's sort of recycled, you know, within a grid cell. We can think about upwind, sort of precipitation shed over land and upwind over, over ocean. And so, when we think about, you know, a grid cell's precipitation shed, it's really useful to start with that F number that we derived because that tells us about the partitioning of land versus ocean, and so then we can map that with these, um, moisture tracking models to really put quantities of water that are coming from different places. So, here's just one example. In this particular example, we, we do this partitioning, the local, uh, you know, water is 1.4%, and then we can, uh, the land total land is 53% or something like that, ocean 45%. We can then also overlay with, you know, what type of land is it coming from? Is it coming from forest or mixed forest, or is it coming from, you know, other land cover, like, like cropland? So, again, can do this kind of partitioning, um, and we can look over the, the growing season as well as three different parts. And so, uh, again, for a little grid cell up in, uh, you know, the very, uh, northern, uh, plains here, you can see that these just, uh, the sources and likely distances from which water comes, uh, vary a lot, uh, over the course of the season. So, just an example to show you, kind of the, the, the work that we're able to do.

Um, so, question, uh, so, you focused on, uh, isotopes, or is it just for tracking, but it seems like that water is going to have minerals and impurities and everything and fertilizer residue coming across land. What does that do for, like, pollution? It's a great question. Uh, I don't, I, I don't have any answer for you. Um, there is a lot of interest in, um, you know, the sort of microplastics community is even sampling rainfall to, to get at this contamination question, but I don't think we have any sort of a comprehensive view of, of what's being transported. Well, and the other question is, uh, is climate change, well, changes. Uh, so, we're going to have, you know, the warmer oceans. It sounds like you have more water coming out. Like, are there other patterns that affect the dynamics of the system that climate change drives? Yeah, absolutely. So, we do expect obviously more, like, I think the naive assumption would be, uh, F goes down everywhere 'cause we have mo-, more oceanic water. Mhm. Right? So, that fraction should go down. Um, but, we do know that these, um, major modes of circulation link different places, right? Um, and so, these places that have a high F values already, maybe actually a little bit outside, say, a circulation pattern that's getting turbocharged with more evaporation from the ocean. I don't think we know. Um, we're starting to put together F over time to see if we can see any signal over the past 2 and a half decades. There's a little bit in some places. Yeah, that's a great question. Sorry, go ahead. No, no, thank you. That's great. Yeah.

So, if we look at, you know, these are just examples showing you, kind of the, you know, over, you know, major cropland areas, um, how much water's coming from different land types. Um, and I'll just draw your eye up to the upper left. When you look at the Sahel, right? High fractions of this, of of crop water are coming from forest. So, upwards of 30, 40, 50%. And really, it's coming from the Congo Basin. So, I'll show you that in a second. I'm just going to zip through for a sec 'cause I don't think, um, I want to make sure we have some time. Um, if we think about, uh, this is now zooming in on Africa and the sort of, if we think about the Sahel again in particular, the main growing season, um, it, you know, has has quite high, um, forest dependence for its water. Even though it's a monsoonal system. I think this sort of gets at your question. When you get up into the Sahel, like a lot of that oceanic water is, is coming out in the coastal Sudanese zone. And, uh, and when you get up to the north, um, you're, you're still really land dependent. And if we even just sort of flip that paradigm around and think about where, um, where is the Congo Basin the main source of water for, if you look, um, uh, in, um, in sort of Western Africa, this is where the main sources of, of water come from. You have a little bit of European forest, you know, there is ocean in there, but when you look at where land is, it's all from the Congo Basin. Um, same for East Africa, a little bit of transpired water from, from, um, the southern Amazon. So, just a very, uh, tight linkage to the Congo Basin.

Um, what we're working on right now is putting together these precipitation sheds and actually looking at, um, deforestation within a delineated precipitation shed. We don't quite have that done yet, but I'm going to show you the sort of first approximation of that. Um, and what we've done is take, again, outcomes like, uh, you know, NIRV or NDVI, greenness metrics, um, again, delineating that growing season, the crop phenology, and, and taking that maximum greenness, uh, as a, uh, a good proxy for yield. And we're going to look, um, at, you know, sort of a longer run change in average maximum yield. And this is to account for these other modes of climate variability. Like, if you have a strong El Niño year, right, that swings stuff around, um, everywhere. Um, both of these periods have have one, one big El Niño in them, so at least that's maybe a good, uh, a good metric. Um, and we've, we've done some work to, uh, leave out El Niños as well, as well, to sort of see how that changes, but I'll just show you the average here. Um, and we're going to control for, sort of how much crop area is in the cell, kind of in these later areas, and, and local land use change over that time period, and then we're going to ask whether greenness and the change in greenness, so, overall, kind of yield proxy and, and change in yields, vary based on whether there was deforestation in the, uh, surrounding cells, like the precipitation shed, in the interim period. And we'll do it, um, just with a pretty simple, like, statistical model, empirical statistical model, where we're going to look at the outcome, in greenness index, vegetation index, or a change in the vegetation index as a function of, uh, deforestation as you move out from that cell.

Um, I, uh, just to show you, there's a lot of heterogeneity across, uh, sub-Saharan Africa. So, average peak NDVI, right, varies a lot, and the change over those periods, you know, some places had increases, some places had decreases. So, there's, there's, um, there's signal to work with here. If we look at, um, these results, uh, now I'm just going to show you the, you know, the vegetation index. This is NDVI, but we've done it for a, for a host of them. Uh, and you think about deforestation locally and then as you move out, sort of in grid cells. Um, always controlling for what happens locally, you can see that if there's been any agricultural deforestation in those surrounding areas, you actually see a, a change in, in, uh, in greenness. And that's controlling for, sort of local changes that would actually change the color of the cell as well. Um, and if you see other deforestation, like from development, you know, you don't see the same thing. So, we actually do see some preliminary evidence that that maybe this water pathway is, is already showing up in, in crop yields. Something like, it's something like a 1% per square kilometer of agricultural deforestation in the surrounding area. Now, this is really looking just coarsely, not with the delineated precipitation sheds. Uh, we're trying, we're trying to do that, uh, now.

Uh, if you look at the NDVI difference, maybe you want to do a, sort of better local control for things that might, you know, might be going on that that change the greenness that that have nothing to do with this. This is now that, that differences formula. So, the change over time. And again, you see a, a bigger reduction in places that had agricultural deforestation, um, as well. And, and again, it, it declines as you move out, uh, which makes sense. And, um, we've also done some work looking at what's being grown. You see this in, in maize, so the top plot over here is, is a version of what I just showed you, but only for maize cropland. Again, with just greenness and then change in greenness. Um, you see it a little bit in cassava. Cassava's less sensitive. Um, cassava's also like the ultimate food security crop. You put it in the ground, you can leave it for six months, you can leave it for 18 months. You'd expect to see not as strong of a signal there. Um, and then when you look at something like, like oil palm, um, which is right, like, uh, a good kind of placebo test 'cause it's, it's already trees and that are going to, um, do something, uh, uh, similar. You're, you're not seeing that there. So, um, we think that this is some pretty good preliminary evidence that, um, that we're seeing this pathway show up in crop yields. Again, we're working on finishing, kind of more detailed empirical moisture tracing like with these actually, um, delineated precipitation sheds, and doing some more of that sensitive phenology work to make sure that we're, we're picking up water changes in the parts of the growing season that matter. If it's a, if it's a water change, you know, sort of the end of the harvest period, it doesn't matter. So, we're, we're making sure we're, we're, uh, getting everything, um, in the right order. But, I would say it's not, it's not so optimistic right now. It really does look like some examples, like an example of a bad, kind of reinforcing feedback from agricultural deforestation. So, you know, we know that, uh, crop water is coming from, in a lot of cases, coming from land-based sources, and then we know that there's a lot of cropland expansion, particularly in places like the Congo Basin, and that's potentially creating a, a bad feedback loop in this, uh, in this kind of climate and food security world. So, it could be, uh, you know, it, it could be unfortunately an example of maladaptation.

So, what do we, what do we take from that? Like, >> What if I, This is, we establish, you know, deforest, establish more agricultural land, it messes up our water, our yields are lower, we deforest more. Yeah. Okay. >> Yeah. I haven't shown you that third part, but there is pretty good evidence that more recently deforested land is not as productive as older established cropland. So, that would be the third piece, yeah. So, then the flip side, if we invested more in, you know, higher productivity in land areas, that that would suggest the, the reverse thing is that land would be extra, you know, the water supply would still be available for that land if you had the investments again. Yeah. And

One of the things we think about is also, you know, what drives smallholder agricultural deforestation? Like it's it's really poverty. Um, right? And there's some pretty great detailed survey work recently showing that like, you know, one of the last things smallholders will do, um, I think this is from Malawi, um, great survey that showed that one of the last actions they would take before skipping a meal, smallholder farmers would be to go cut down a tree and try to sell it for charcoal. And so, it's not it's not so obvious like, you know, big oil palm plantation or soybean farmers in Brazil or whatever. I think, uh, it's a much more smallholder poverty-driven stories.

You could also think about social safety net kind of interventions, right? If you can help, uh, farmers through a bad season who are at the forest margin, like with cash transfers or other types of interventions that might pay outsized dividends as well. Right. So, you can think about, yeah, taking us straight into into policy land, right?

I think, you know, I I'll just say one thing worth thinking about of course is like, not all carbon is equal and that there might be carbon that's more important to protect. Um, and this sometimes feels a little heretical, like I'm not letting fossil fuel companies off the hook, um, but, uh, we do know that some of this terrestrial carbon, right, is just so outsized in its importance. Um, we've been learning a ton, of course, about these cropland atmosphere connections. Well, my, you know, my group dorks out about this all day, every day. Um, but yeah, so we, you know, this land-originating moisture is just turning out to be very, very important for global crops. And and now we kind of know, right, where these areas are that are more likely to be stressed. And then we're sort of working through this that agricultural deforestation is having this knock-on effect in addition to releasing a lot of carbon into the atmosphere.

Um, if we think about how this fits, you know, into something like the Paris Agreement, right? Um, the idea was to limit global warming and reduce emissions, uh, and reach the peak as quickly as possible. The idea was that this was, you know, going to be determined, uh, going to be achieved with, you know, voluntary commitments. Uh, but also finance, capacity building, you know, policies, technology transfer. And I think some of this fits in very well there. Um, I mentioned earlier, you know, that there's no real evidence that this sort of rosy picture in the 1.5° scenarios is happening. Right, the number of hungry is rising. Uh, land use emissions are steady if not rising in some locations. I haven't shown you that here, but that's like a something my group works on. There's no evidence of massive deforestation. Um, and there's um, there is evidence of like cropland expansion in sub-Saharan Africa that's not particularly productive. So, that sort of flies in the face of this picture.

So, what could we think, however, you know, in a in a policy world about solving this? I think we could think about starting to treat carbon in biomass maybe a little bit differently than carbon from fossil reservoirs. Um, if you move carbon from fossil reservoirs to the atmosphere, you get all the warming effects, you get ocean acidification. We know that, but one of the nice things about like the social cost of carbon when thinking about fossil sources is that it doesn't matter where the emission is from, right? The emission is of carbon dioxide, it's a well-mixed, long-lived greenhouse gas, the ocean is big, and so we are getting pretty good at understanding what an emission does in terms of impacts everywhere, and depending on all the parameters that you want to build in more normatively about discounting and and whatnot, we have a methodology to to understand the social cost of those emissions and build policy around them.

If you think about moving carbon from the terrestrial realm to the atmosphere, you get all the warming effects and the ocean acidification, you get all the atmospheric effects, but you're also getting, uh, changes in reflectivity, you're getting a shift in like interannual sequestration potential and overall sequestration potential, and you're getting this vegetation hydrology set of changes that I've that I've outlined a little bit here. Uh, interestingly, of course, then the location really does matter. It doesn't matter for the atmospheric effect, but it matters for the other effects. And so, we started to think, you know, in my group we've been starting to frame out what it might look like to really link um this sort of source-receptor kind of I idea more formally into something like a social cost of of terrestrial carbon to kind of modify how we think about this, and that would really naturally lead to um to sort of thinking about where to put resources and could play in nicely to something like a loss and damage framework. So, that's that's where we're thinking on the policy end as we work through some of the details on the science here.

Um, so, thank you so much for you know, letting me talk to you about all the fun stuff we're doing, and please don't hesitate to reach out if if we can be of any help.

Well, fantastic. Thank you very much. Um, maybe I'll stop sharing so I can see folks. Yeah, yeah. So, I guess one question be Okay, so one area that you kind of went past but didn't dwell on its soils. So you have you know like ultimately goes from air through plants through soils and kind of back through ground water. So what's the underground portion of that cycle?

Yeah. Yeah, for sure. When I'm saying terrestrial, I'm thinking about vegetation and soils, like the whole sandwich of above ground and below ground carbon and and water. Yeah, but how do they modulate the process? Like what sense do they drink carbon nitrogen from fertilization or how do they mess up the story? Yeah, I don't think we fully know yet. You know, one of the things we know is when you know, forest deforestation happens, you lose all of the above ground, you know, biomass ends up in the atmosphere usually within a few decades and you lose about 25% of the soil carbon over short time scales. That's like a masking regional heterogeneity there, but but so we do know that there is also, you know, strong soil vegetation water linkages that are probably modulating this as well. Yeah.

Right and I guess to what extent another question about so what is surface temperature effect? I'm thinking this in addition to fertilization and you know putting in from deforestation to just cooling technologies, ways of cooling air in the region to improve farming and what not, but it's going to affect the water cycle as well, right? So what would those be effects be?

Yeah, so we've done some work using like satellite measures of surface temperature to try to understand as a case study in this sort of like what irrigation does, right? And that sounds like a very stupid question, but like it turns out at least in the regions that we've looked at, irrigation's benefit is about 2/3 just relieving soil moisture shortages, right? Providing water and about 1/3 heat stress relief because you're getting that evaporative effect and it matters to plants. So, that's just a very narrow answer to what I think is a broader story that must be playing out as well, right? We do know that like that energy balance getting that uh sort of changing the the sort of uh latent and sensible heat balance will matter. Uh, it matters for people, it matters for for vegetation. I don't think we have the full story yet, but but we do know that that that um the sort of heat relief is going to play in. Yeah.

Okay. So, now looking at the ocean, um I mean it it's a carbon sink. I mean let's do the dabble in it. I mean it is a carbon sink. To what extent can oceans be used, uh, to store the carbon that's emitted, uh, and do they re-emit it or is it like are we talking about per- permanent sinks?

Yeah, I think that's a trillion-dollar question right now. So, yeah, I mean ocean up takes a lot, right? And it's sort of like some of it moves down into the deep ocean, some of it burples back out, and sort of cycles around for a long time. Um, uh, and there's a lot of interest in, you know, like whether it's land-based, you know, sort of rock weathering or like liming the ocean, could you like enhance CO2 uptake in the ocean and sort of get it down into the deep ocean, um, more quickly? I'm not an expert in any of that, but but it is a really interesting question about whether you can like override or or turbocharge some of the the natural thermostats in the Earth system to kind of do that.

>> Right. And just thinking the other side of the of the ocean is desalination. Uh, you know, we're kind of uh cheating, I guess, by getting some of the ocean water via desal to land-based farms. To what extent is that feasible? And I mean is it just by the coast or rich countries or is this a safety valve?

Um, I I it's a little bit of a safety valve. Right now, it is coastal and it is more wealthy countries, but not exclusively. We've got some middle-income interest, right? Um, mostly makes sense probably for, uh, for drinking water purposes, right? Sort of for urban use. Um, but maybe for high-value agricultural use. I mean, you would probably never flood irrigate with something like that. Um, the, you know, the other thing that's been interesting is just the We're going a little far afield now, but like there's been a lot of work on sort of saline irrigation and what crops can do well in sort of, uh, not like right, yeah, fully desal, but like, you know, dates like salty water, some of it, some kinds of tomatoes. So, there's a pretty interesting, um, questions that along the line, uh, the lines of what you're asking. Yeah.

And I guess I mangrove forests, they specifically grow in salt water, but I don't think we eat those today. But maybe we can, uh, develop some sort of, you know, uh, something that grows on the mangroves.

Yeah, they they actually come with the laws of the earth, like the little sort of tidal thingies. Um, cool. Well, yeah, thank you very much. It was a fascinating.

Thanks so much for having me. I really appreciate it. And yeah, don't hesitate to reach out if if we can be of help with anything.