 Gordon is a senior scientist at NCAR and has served as the head of terrestrial systems and provides, is one of the scientists who provides the connection between ecological processes and the climate modeling, atmospheric modeling that goes on and that NCAR is known so well for. He has even written a textbook on this topic which would be highly recommended for everyone who wants to get an introduction of that like interface between ecology and the climate and please take it away Gordon, your the talk he's presenting today is re-inventing nature, environmental stewardship in the age of earth system models. Thank you, that was a great introduction and it's great to have this opportunity to really share my views on the importance of interdisciplinary research. As you mentioned, I'm an ecologist working at an atmospheric science research center and that really has shaped my views on what interdisciplinary research is and what it can contribute and that's what I want to talk about today. That's the theme for today and the title of this talk actually comes about from a book that was written in 2015 called the invention of nature. The premise was that Alexander von Humboldt really invented an interdisciplinary study of nature and earth as a system. The idea of the book was that this really presages the advent of earth system models that we use today and I think that's true but if so then we as scientists have sort of largely forgotten those ideas. Yes, we do have these earth system models that combine the atmosphere and the biosphere and the hydrosphere and geosphere but we too often talk across disciplines rather than among different disciplines. We frame our science and our models as providing boundary conditions to someone else's science or model rather than being truly integrated. So that's the premise I want to talk about today is that we need to reinvent this interdisciplinary study of nature and I can start this by looking at these two different views of climate. The one on the left is sort of what's called the classic sort of blue marble view of earth as if you look at earth from space far you see the blues of the oceans and the whites of the clouds. It's really this sort of geophysical view of the earth that emphasizes atmosphere physics and fluid dynamics but of course if you look more closely if you can peer underneath the clouds you're actually going to see the greens of the vegetation and so the emerald planet is more of a biogeoscience perspective of earth and it's really emphasizing the effects of climate or our ecosystems on climate and atmospheric composition. And we can see this sort of biogeoscience's perspective in the evolution of our climate models. You know in the 1990s we were just dealing with sort of the physical representation of climate. They considered the land surface but it was really in terms of providing energy, water, and momentum fluxes to the atmosphere and it used very simple representations of vegetation. On the right in the by the 2010s we'd actually gotten into the earth system perspective. It's really recognizing the biosphere, its ecology, and its biogeochemistry essential to understanding climate. And now in addition to dealing with energy and water fluxes the models are also considering the carbon cycle, the nitrogen cycle, chemistry climate interactions through for example BVOCs and ozone and methane and aerosol. We have biomass burning. We're definitely dealing a lot with land use and land cover change as an important aspect of these models in an important way in which we are changing climate. And so this evolution from physical climate models to earth system models really has occurred over the last 20 to 30 years. And so we put a lot of ecology in these models. We think the ecology is really important for getting the climate right but did we put the models solely, did we put this ecology into our models solely to improve our climate projections or is there value in the ecology itself? What I mean by that is can you use these models to actually study originally designed for climate but can you use them to study ecology? And the answer to that I think is really has been a resounding yes. You know the classic way that these models are used are for what we would call prediction primarily climate but now you're going to hear words about earth system prediction but it's one of these consequences of alternative social economic pathways. But then I think and that's how these models are used a lot primarily but I think really what's not appreciated by them is that they're really important tools for scientific discovery identifying ecological processes that determine climate for example stomata conductance and alternative sort of parameterizations of stomata conductance. And it's really these tools are really really or these models are really good at actually advancing our theory. It allows us to test the generality of ecological theories at the macro scale. So we use them a lot in terms of project climate projections but the ecological utility of these models I think is really strong for scientific discovery and for advancing theory. What I want to point out is you know the merging of atmospheric science and ecology is not really straightforward. You know they have they're different cultures in different ways of doing research and I think this picture of Paul Dirac really illustrates sort of the stereotypical view of a physicist which is a hour-looking man standing in front of a blackboard with equations. When we go to ecologists however we see something really different. It's usually somebody outside in the field collecting data smiling in a very inspirational setting. But what I want to get at in this talk is that if you actually bring these two different cultures and sciences together you can actually find a new way new science emerges from that and that is how the biosphere affects climate and so that and the idea that the biosphere is central to understanding climate change. And one of the classic sort of studies of this has been to look at sort of tropical deforestation. This was one of the first papers done on that in the 80s by Bob Dickinson and Anderson Sellars and what they were doing is they were looking at the effect of tropical deforestation on climate and showed very easily and the prevailing view now is that if you remove the tropical forest you're actually warming the surface temperatures you're decreasing evapotranspiration you're decreasing precipitation that's what they did in the study they ran the model they had a climate model they ran it once with an intact tropical forest they ran it again in which the tropical forest was removed and what you see is you see a warming of surface air temperature by several degrees the soil surface is also warming you're a decrease in precipitation over this region and a decrease in evapotranspiration and so this was to me one of the profound this is a profoundly ecological study because it was showing how the forests are affecting the climate of the region it was written in a meteorological journal but it's actually to me like an inherently ecological problem and I think we can actually see that in my own work sort of actually illustrates sort of this different ways of looking at vegetation or this the problem whether you're approaching it from an ecological viewpoint or a climate viewpoint in that you know I've studied the boreal forest a lot throughout my career as a phd student I was developing an ecological model of the boreal forest and how it would change with climate change and what we see on the left here is a review paper I written that talked about what are the important processes that if one is developing a global model of the boreal forest in its response to climate change what are the processes you have to consider but it was actually written in the viewpoint of we know what climate is and we want to see how the forest respond to that climate change on the right just three years after this paper I had moved to NCAR as a postdoc and actually at NCAR I was exposed to a completely different view which is achieve maybe the boreal forest is actually very important to getting climate right and you can see it in the figure here the snow has a very high albedo it reflects sunlight back to space it cools the surface the trees are darker they're absorbing it protrude above the snow they're absorbing more radiation therefore they're warming the surface and so this with paper was written with well the boreal forest actually has a huge impact on our climate both papers have been highly cited they're both but they're very different views of the boreal forest one is of the ecological view the other is that the boreal forest are a very important part of getting the climate problem right but what we've seen because of this research because of these climate models and air system models and because we've put ecological vegetation into them we're really changing our views of the importance of the biosphere and I think one of the enduring legacy of this work is that it's really changed how we view climate yeah it's really interesting in the 1800s there's a really large debate both in the U.S. and actually internationally about whether deforestation was altering climate there are really strong opinions and strong things said about it but meteorology was emerging as a science at that time and as meteorologists saw a physical understanding of climate they rejected the notion that vegetation affects climate and this is really shown in these quotes by Cleveland Abbey who was a actually really influential meteorologist and is recognized for services by the American Meteorological Society but he didn't believe that the vegetation mattered at all and what I would say now is we've come a long way since those statements that the biosphere doesn't matter we now know that that's not true and in fact what we talk about now is we talk about the climate services of forest whether it's tropical forest tempered arboreal forest we talk about carbon storage and evaporation and albedo influences so what are the types of things that are actually going on next with these models what opportunities are next no I just want to say that was a profound change in the way climate is viewed that the biosphere really does matter so what opportunities are next you know there is this idea of earth system prediction you're going to hear it a lot our models are now called earth system models so we now have to do earth system prediction but if you look at it from a atmospheric point of view or a climate point of view typically as the land is perceived as a source of predictability to the atmosphere if we initialize soil or snow or vegetation can we actually improve our predictions of weather and climate that's how it's narrowly perceived the land surface is never perceived from the atmospheric point of view but there's much more to that than just climate there's a lot of change going on in the land surface there's drought there's wildfires there's forest mortality there's insects insects and there's actually greening of the vegetation can we actually start using our models to predict what these changes are and how they're why they're occurring as that was the premise of this review article that had in science and I just want to give one example of what we mean by earth system prediction this is an example of using our earth system model to look at changes in net ecosystem production this is the net carbon uptake by by the ecosystems and we're looking at the predictability of it what we see on the left is a temporal trend our verification forecast in net ecosystem production and what we're seeing in the orange or the red is actually a forecast with one year lead time so with one year lead time can we actually reconstruct what this this actual verification forecast you can see it's actually doing a pretty good job and the correlation is fairly high it's around 0.7 or so at one year if we go to a two year lead time it drops off by three year lead time it's much smaller but the idea is that if you actually do these predictions if you can initialize a model correctly you can actually do a pretty good forecast of what carbon storage might be and that's what we mean I mean by earth system prediction let me try to quick I think a little slow on this let me actually say what are the challenges so there are a lot of opportunities I think with these earth system models the challenge is really sort of increasing model complexity this is looking at just the number of equations that go in the land component of our model in our technical nodes this is not the number of equations in the model itself this is just the number of equations that we report in our technical nodes and you can see over time we've gone from less than a hundred equations in our technical descriptions to well over a thousand equations we're adding more complexity and more processes in our model but that does that make it better and this I think is that a really big challenge to these models we've been very good at constructing the models and adding more processes now we what we need to do is we need to deconstruct these models into their fundamentals these sort of fundamental processes that all these models have in them but we differ in terms of how we implement the equations we make choices when we put these in our models but we're not very good at discussing these choices they're poorly documented and I think what we need to do is take the mystery out of our models to really understand why we get the answers we're getting and that was the premise of this other this modeling textbook I wrote which is what are the what are the choices we've made and all the assumptions that have gone into these models the last two things I want to last two slides I want to end with is just we going back to this idea of interdisciplinary science this painting by Peter Brogill the elder has really influenced my views on science on this and actually ended up sort of using his to motivate another textbook on this book is more like why the biosphere matters the other textbook is how to model the biosphere but if you look at this it's a it's an artistic painting it has merit for artistic value itself but it's also been used by ecologists and their textbooks to define the sort of the concept of a landscape and then it's been used in climate books to talk about climate change and in particular this painting was done in 1565 but if you look at it the mountains are in their mountains in the Dutch landscape and what this is is it records Brogill's impressions of the severe winters at that time and marked a period of extended colder than usual winters and so we actually are seeing a record of climate change in this so you can look at this from our art point of view you can look at it from a landscape ecology point of view or you can look at it from a climate change point of view all from the same information the last thing I'll just leave with I think as we open it up for questions it's just this quote you know from a Anton Kerner from 1888 the idea of interdisciplinary research is a really old problem it's not new he was a plant he was a botanist and a plant geographer and he pointed out that even at that time researchers becoming more disciplinary in their research and he has this nice great narrowness too often has huge risks as its consequence and I think that's very true for all of our research they've run a little over so let me just stop there and see if they're questions thank you for that Gordon we'll have a little bit of time for questions and so people can type them in if they're just coming in you can type your question into the chat or you've just been listening into Gordon it's like the fastest typers get the first question maybe while people are typing up I'll ask Gordon a question this is the advantage of having the unmute button and that is are the NCAR models thinking more about or the NCAR modelers thinking more about like sort of a dynamic surface and like so one of the things that the community service dynamics modeling system works on this is all these like changes in the surface itself so changes in the coastlines etc etc and is that something that is starting to become important in the earth system models on the global and predictive scale as well no we can we can deal with sort of changes in coastline in this in for a sort of long time scales of like paleo climate for example where the orography of the continents change or one of one of the sort of things you have to provide to them was where's the land what's growing what is the vegetation or properties of the land that so we can change that for sort of different ororographies depending upon long paleo climate climate time scales but there's nothing being done in terms of sort of sort of a dynamic sort of surface where you might sort of see erosion along the coastlines or even flooding we can't really move those as feet changes in in these models at this point the surface really the dynamics of the surface really is in terms of just the changing vegetation we don't even have to do we don't even deal with erosion of soil or we don't do changes in sort of soil texture and soil properties over time yeah I think Pam Sullivan's talk yesterday started like hitting upon that like root zones and soil processes being an important like part of like how do you calculate that balances to the atmosphere or the climate system again I'll I'll take a question I'll read out a question by Jeff he says it's a great talk you've been central to integrating ecological processes into models so what are your thoughts on the integration of social processes into the next generation of ESM model ensembles well yeah that's a great question and one of the things that you know these models have shown you're particularly with tropical deforestation with deforestation is that changing the vegetation land use and land cover change is really important feedback on climate so there's this real rich history now of work going on because of that to look at how forests can be managed to mitigate climate change you know can we plant can we do a deforestation can we be planting trees for sort of carbon storage and what might be the effects of that on albedo and evapotranspiration and other things um but um that you can approach that from sort of a scenario point of view which is if we planted trees this is what would happen but the alternative view a way of doing it is to actually sort of make that land use decision making part of your model itself so that as the climate evolves the land use evolves in itself that's a really hard thing to do uh there was a lot of interest in that I'd say beginning like five years ago that became a really you know hot topic I think people have sort of backed off from that now because it involves human decision making and we're not quite sure how to do that they are these integrative assessment models that do that they tend to work on very different time scales than the than what a climate model is capable of doing they have different ways of sort of modeling than what the these physically based or biologically based climate models do which are really much process oriented models so it's still being explored I don't think it's actually sort of uh I thought like five years ago it was actually going to become a really big area this sort of the new frontier I think it is being done but it's still it's not I think most of the modeling centers are sort of backing off of saying that that's going to be a major part of what they do department of energy U.S. department of energy does have sort of a coupled model that does these things we at NCAR don't have that type of a coupled model thank you another question is by Brad Murray and he says you raised the question of whether adding many more equations increase the accuracy of the modeling a lot do you have a guess about the answer yeah you know it's okay so yeah it's going to come down to what do we mean by accuracy because we can sort of say the viewpoint has always been if we put in more sort of process rich parameterizations we're going to we're going to be more we're increasing the authenticity of the models we're more faithfully representing processes realistically but that doesn't necessarily mean we're going to reduce uncertainty in the model or it doesn't necessarily mean we're actually going to get it even a better answer we're going to more faithfully reproduce the some say some temperature trend or some trend in carbon storage or something like that because as we add more processes it's harder and harder to get the model to work right and so typically actually I think what we're seeing now with these models is that as we add more complexity to them the models aren't necessarily getting better in terms of reproducing some sort of observational data set and the classic example I would have is like adding the nitrogen cycle into our carbon models the first carbon models had no nitrogen cycle at all and then people realized really quickly that well nitrogen provides nitrogen availability provides a really good limit a strong limitation to how much carbon the biosphere can absorb in a warmer world over higher co2 concentrations so then people started putting in the nitrogen cycle the nitrogen cycle is I'd say high particularly at a global scale is highly unknown uncertain we don't really quite know we know sort of the basic processes but we don't know how to mathematically represent them and now what we're finding is everybody has a nitrogen cycle in the model but we're getting widely different answers because we don't really quite know all the details of how to model the nitrogen cycle we know that if we don't have it in we're going to overestimate how much carbon can be stored by the biosphere but when we put it in we're now getting really widely divergent answers among our models so we think our models are getting better in terms of having a feedback that's important nitrogen but they're becoming more uncertain because we don't actually know the details and the specifics of how to do the nitrogen cycle across all the ecosystems of the world