Modeling (Moderator: Colleen Petrik)
Regional/global modeling (Gennadi Lessin)
Shelf-coastline interaction, extreme events (Stanley Nmor)
Coupling benthic ecology and diagenesis (Subhadeep Rakshit)
Benthic ecosystems and fisheries (Colleen Petrik)
Breakouts report back
Learn more https://www.us-ocb.org/becs/
welcome back to our afternoon session where we’re going to be addressing modeling I’m going to try to keep us to time now so we’ll get started right at two with our first talk about urom and benik modeling using this model thank you so uh during the discussion session I thought that what we would like to do as modelers is to make everyone a little bit modeler so to make you understand modeling better so we can communicate and work together so I did not see much enthusiasm so um we usually uh at laboratory in the modeling group around three hour sessions uh handson workshops or introduction to B modeling 20 minutes B is a good start and we can continue afterwards so uh modeling talks can include lots of pictures of lots of equations so I decided to go for pictures uh and so XM byal model uh that is a mechanistic model we will try to incorporate up to date understanding of the processes but that means that we within this framework we also incorporate up toate lot of understanding so in order to advance the model we need to improve understanding but let’s start from uh the beginning where this model originate from uh it was developed in 1990s as an international effort of different institutes around the North Sea uh and initially it was meant to uh model uh ecosystem of the North Sea but since it evolved into a generic modeling uh tool for simulations at different scales in terms of time and space so we can apply it at very localized uh scales so here is a domain where Earth is around coup to e it’s optional model of l sound surrounding area Regional scale this is Atlantic meridional margin domain seven uh kilometer resolution is our Workforce where we usually apply it and recently it’s been applied also at the global scale so we can also apply it in 1D where it can serve as research tool to understand process Dynamics and in in Zero D or within Box model to reproduce mesop conditions so I W back and first to switch because I have lots of animations transitions so uh the model has uh around 100 State variables and uh 400 parameters that we deal with and it represents variable stomry of py ofon we have four pyop types in the standard configuration we have detail microbial loop with explicit representation of bacteria uh both inic and inv ventic uh we have cycles of all major nutrients including iron we have classification in the sediments we have three uh layer structure so these are Dynamic layers they change with they Horizon changes with time and we have oxic oxidized and an oxic layer I will tell about them in more detail in a second and we have explicit representation of biota and again this is a standard configuration uh and we have two bacteria uh types aerobic and aerobic we have two microne suspension and deposit heers we have different uh diet compositions and different habitats from where they take their food and where they uh pair pieces and we have one group of myone so since recently about five six years the model has been re implemented within framework for aquatic bical models or P I don’t know people are famar with that maybe modelers would be familiar so uh p is a couler it’s more than just coupler uh what it allows is uh to code your model as separate instances that communicate with each other and it allows to uh couple your bu chemical model to a big wide range of um fomic models because a model is agnostic of all the environment around it it’s just receives all the necessary input and uh provides you with uh sys and sources so for example if we have um model like npzd model then uh NP Z andd are separate instances nutrient phlan and we can only Fanon requires light other components don’t require light so we only provide what is required for each of these components and we then combine them together into a Boke ecosystem so what it allows opportunities that F structure allows us is that we couple Same by geochemical model to wide range of uh physical models so this is not the exhaustive Leist there are more that are U available we can customize model structure for example we can take elements of the model and combine them in the way that we want we need uh for example we can replace complex pic structure of LM with simple model like for variables but use the whole btic model of LM or vice versa we can use simplified Bic model but we don’t want to do that and we can also couple as to other models because s only represents lower tropic level but we can couple to uh models like Miser fish and fish we can model aquaculture by coupling with aquaculture models models of casses trolling and so on so model uh code is available online anyone can download it uh documentation is available as well and you have also a big team of modelers who will be happy to help and answer questions you have wide user base and lots of applications that involve Le full two-way couple are they online couplings uh this framework it can be one way it can be two way so for uh fish Miser it can be coupled one way so we can only provide bu chemical input but it can be also tool so yes so fish will provide organic matter pieces into budet chemical model so what this structure means for the Bic model is that also uh we have this uh modular approach where uh there is a separation between model code or model development and uh between configurations or between users where model code can be represented by building blocks of similar uh functionality so for this is blocks of disol Organics partic organic matter dis organic matter zos and bacteria and so depending on system we want to represent we combine this loocks into this scope structure so so here is a representation of this classical LM configuration the way it was developed originally and we did use uh Lego bricks to represent that because it’s very nice to show how things are connected between themselves but then we got email from Legos so they said that we can do that so so this is class we have dissol inorganics and these are produced or consume this depending on what species we’re dealing with and exchange with theic we have um microna bacteria two types of particulate organic matter one is labal another one is refactory so la is available to for ref is available only to bacteria and there are other things more details about that as well as dis organic matter so if we want to we can go into one way and simplify the model and uh create something we call bentic returns where this is something that is used in global models for example where everything all the organic matter settles to the bottom and mineralized and nothing else happens or we we can go the other way and create something more complex introduce more functional TP of microphone and try to model diversity of microphone and what it means for example for carbon Dynamics in the sediment canic the logic exchange so this is another representation of the same model with can do boxes and arrows we can do Lego not Lego bricks and this is I I did that based on output of the model where it was run in 1D in Gotham the water column and just trying to represent dynamics of the model if we look at the sediment uh being within the sediment like like what the poy books see sees in the sediment so these are the layers so oxygenated layer with free oxygen this is oxidized where we don’t have pre oxygen we have nitrite and uh this is an oxy layer this is a representation of habitats of different types of biota so the darker the color gets the more the higher is biomass and this red curve is uh conceptual representation of organic matter so this model assumes that organic matter has this exponential representation and what you will see is that as time goes we have deposition so this Curve will move to the right because we get more carbon at the surface then we will get increase in bacterial biomass as they consume uh this carbon and then increasing deposit leers which slower then bacteria start to consume the organic matter and then these thickness of these layers will become smaller because oxygen will be consumed and nitrate will be consumed and be denitrification But first you also will see adjustment of initial condition so don’t worry if suddenly all of these layers go down and there will be lots of oxygen in the sediment and I cannot show you this illumination because it doesn’t work it works so there is the position so more organic matter at the surface anerobic bacteria first and then aerobic and anerobic consumption of oxygen nitrate uh this is the uh average penetration dep of carbon so as deposit feeder speed on it carbon moves a little bit down due to boration that’s it that was one year of simulation so uh then I go into examples of what we’re doing with this model and uh all uh here is uh Southwest UK and pl mode and just some 10 kilomet south of pl we have Western Channel Observatory and station L4 where uh orographic and bical data have been collected for decades so opic more than 100 years with increased frequency about 25 years and we have also the regular btic survey collecting bentic fora several times a year so we have an idea of a composition of pona uh and the seasonal changes so what we wanted to do is to look at uh this time series of B I try to reproduce them with the model and from this states of Bon see what are the fluxes how this V form contributes to carbon cycling in the sediment so uh those were our questions and so what how different types of microphone respond to food availability so deposition from the pic how this food availability shapes community structure and how this structure is reflected in the consistent functioning so we have two functional types of P so we have to uh divide them into two groups deposit tent Fe based on their traits feing U references and these are outputs of the model so we don’t run full model Bic andic we use phop planton time series as a proxy for food availability and we are running this model within optimization framework to get the best Feit between model output and the data so what you see here these uh black lines are result of almost 1 million simulations it’s just a nice number and what we see is suspension feers are responding very quickly to changes phop so they are kind of mirroring what is going on in the pic when food is available they start to grow and when food become scarce they are decreasing Bas decreases very quickly and deposit feeders have more dump uh response so here to suspension feeders it’s about one week and for deposit feeders months and more than one months so from that we can reconstruct the fluxes and see that as soon as organic matter goes into sediment this tangent e quickly grab it eat it and there are considerable losses which we think it’s not just respiration it’s also losses that are not represented in the model explicitly such as uh reproduction and loss into Pic muron and the pration by is for example and the Posite feers are receiving carbon which was already processed by so uh another example is where we kind of inspire by this work we uh Diversified types of btic PHA in the model so instead of two we have six and we run it at this same station4 did the validation and then uh tried to look at the uh Northwest European shelf scale at how uh the position Dynamics shapes the diversity of Bic form on the shell so uh what we see on that figure is that high deposition leads to high biomasses but not necessarily highest biodiversity so there is this red area which is kind of a sweet spot where the position is just right enough to support diverse community of paa where there is enough organic in the sediment and just above the sediment so we can have both deposit feeders and suspension feeders at the same time uh this is another example of importance of organic carbon supply to the sediments and that btic system is responding to that variability of inputs so what’s going on here this are output of climate simulations on the northwest European shell it’s 100 years and we are using three versions of SSM in pilic so default classical XM one with fixed toet F and another one is a simplified pic structure all three models run with same physics and same bentic model so this is response of the same bentic model to different dynamics of deposition and carbon supply to the sediment so what we can see is variability uh both at the same location and different response to different uh supplies but also difference between different regions so when we look in the future then different regions will have different response and sometimes like in this case in the northern nor sea for oxygen and oxy l have uh different pic model but same physics can produce results that go in different directions so here we are talking about long time scales and here about large spatial scales so uh this is diagram of midusa one of the models uh used within uh Earth system modeling so it’s a component of UK a system model and as most of these models uh it includes very simplified representation of ventos but what uh XM can offer due to its modular structure is to take the XM btic model and plug it into Medusa so it has same pic structure as we had before but more evolved Bic system and uh we did that and on the right you can see distributions these are not scientific results this is just demonstation of concept where you run same her system model and uh but the results that you see are from bentic model of so of course one uh issue is that Medusa has also to be CED in this problem framework uh this is area of Interest currently and that we are working on because one one area is to improve the model itself another uh is what we apply the model for and in this case white topical area is impact of trolling on carbon budget so as you know trolling impacts sediment in different ways it kills organisms for it mixes sediments and it resuspend sediments so these are Direct effects are stroll passes but there are also indirect effects due to these uh things changes in not cycling and uh fluxes of nutrients and the big question which is controversial and everybody is interested in is how much trolling contributes to release of uh CO2 so that’s what we’re working on andm is quite suitable tool for this because it has all of the necessary components it has B phone that we can uh heal in the model yeah in the model and we can Mi sediment and it also includes the suspension so we can include all the direct effect and model will the evolution of the model will provide secondary effects on the nutrients so um and because it’s coupled to pic model then we can also look at what how it affects the pic environment and CO2 uh these These are preliminary results that we before so not looking at CO2 looking only at the effect on the sediments within Shel C by chemistry program uh so the message here is just that although we impact on V full removing it and uh looking at the distribution of sediments there are areas where uh this heing and removal of for will uh also decrease the amount of sediment but there are also other areas where the uh content of carbon will increase because this is a btic system is quite complex and model is detailed enough to give these complexities so if we remove it all depends on dynamics of deposition and dynamics of siment working and vertical distribution and so on so it’s not all straightforward overall impact is less organic carbon but local impact is variable uh uh so this is something we are working on and we have now two uh projects starting looking at uh impacts of trolling one of them will be looking at the whole shell and another one will run high resolution model in Irish c um so I will skin through quickly that we have also uh we’re working in Project necton which aims at Improvement of uh products delivered by cernus Marine system uh Marine Service services and although aen provides products for these uh Capal services and bentos is integral part of that model but B products are not delivered by cicus Marine Service so what we are trying to do here is to improve P component to validate it and to make it available as product to everyone so we are looking at I’m sorry I’m gonna have to stop you there I will go I will just conclude okay I’ll conclude with questions so basically working on the model the main questions here and the first one to be addressed that the priority of all of this work with ptic model is uh quality quantity and time and the programing Supply from the pilic so main issues of btic modeling are coming from the pilic environment if that is information is available and we get it right then we can work on all the bentic processes and look at this bentic form what it does what happens to it and how it affects uh carbon fat in the sediment so all of these questions are not questions to the modelers they are questions to Observation experimental community and these questions from experimental Community will help to develop models and advanced understanding of the system so it’s not about model or observations is about working together and using models as tools to advance [Music] understanding up next we have Stanley who will be discussing extreme events okay uh good afternoon everyone um I guess you probably my name now Stan um so I want to talk about event and I think M of the chemistry um so this is work that I did along LA and also some contribution [Music] fromr U so um for this presentation I will show some of the work we been doing trying to understand how this industry so I tell my presentation [Music] um I just you know start from high level and then hopefully down to something more specific so we’ve been discussing a lot over the past couple of days about govern projects and all the different processes system and there’s been quite a number of efforts within the community to try Andy uh these U the fluxy of carbon across you know this land continum and there’s been this kind of Rec of this kind of B Transport of carbon and also process I don’t want to go through all of the numbers here too complicated but I think really nice Pap um if you really want to look at what is of that come to understanding in trying to quantify Caron Del but I just want to highlight the role of poal dels um because um despite fact that they have a very small surface area they do play very key when it comes to the retention of carbon or t within Coors so at the moment I think this is the value being reported in literature about 44% of particular car dep uh so they really play key Ro when it comes to these kind of that get delivered VI the river but also they really important when it comes to acting at the biochemical reactor quite a lot of intense recycling of carbon this particular area to so it’s really important for us to really understand how these F function but because there are strong connection to the river so because they have very strong connection to the river they us rep this high input of um fent delivery especially during period of SL and that is exactly what got me interested Pro about the role of the EP will call it event so us of event OCC when there a very large or maybe High pration event that result to this very large delivery of large quantity of sent material over very short amount of time in a very local set and this event usually they don’t occur often you know something that you you get to see once in a while but um um this uh because of the unpredictability because of the way the Varity of current I really really hard to study them um but something that been getting a lot of trash from within the community um there’s been this increasing intensity and the urrent of this sort of event so here I’m showing on the right hand side with my flight you see kind of like the number of times when of heavy rainfor been reported in theeran see this of like an increasing tendency for to OCC so these are something that you know people are really looking into lately but then the question then Whata dep of and how does AFF at the moment we do not have a very complete picture how work so that is where um my research has been focused on so I kind of think of it like when um what happen when about the chemistry so the question then that are been of of interest to me and my group has been trying to really understand what is the techical response of Po siment when these of FL events OCC and what are kind of primary drivs of siment thre point this and hopefully the goal for us to try to quantify the size and the changes with the so um so I will take my presentation and just kind of give you guys a little bit of a feel of what I mean by this extreme of FL event so I’m taking you guys to France for those of you who have one way or the other find talk vacation close to the south of France so um so there this river called The Ron River um quite well I would say one of the most predominant Riv system thean SE um um if um about 5 to 10 met of T been delivered by this River up to the um area and um 80% of that you know Ur dur the flow um so the the flood when they do come they come with a band really really important for this transport I just want to give a little bit of feel of this area so close to the river M what we delate the proximal Zone there been very very high rate of experimentation 30 could could be as high as 30 cm per year and just a few radi be from that have Del area also really high in terms ofation also very intense byal activity going move offsh of like this region that ising less input so um so this is where my study CED or at least the place I’ve been looking at up researching this work and uh so in this particular just to give a little bit of Fe so right now I’m showing you like an area photograph captured satellite with to one of events and it shows the exent [Music] the when they do luckily enough when my team out there in the field one of this um period they retrieve about 30 cm of sediment within a few days given the fact that this particular region the aage sedimentation rate is about 10 cm per year so you have a TR of more sediment being delivered within just a few days and you have noral period so these are quite um really important event but not only just in terms of the quantity but also the quality of Mater been delivered yeah I’m showing you on the right side um a figure of a figure from a work done by a colleag in the lab where they try to quantify the T contents temporal variability this FL bring in material large quantity but also ter of the quality they do matter and uh so these these are things that I’ve been thinking with the project so cryto ear on this morning he showed some work on trying to observe of event and find way to really them doing observations but I will take a modeling perspective this afternoon show you way thinking about model systems so classically um for those of you who familiar with sment models we have these El genetic models are onedimensional sometimes or most times one dimensional models and I like to think of it like there are two end members of spectrum when when it comes to mod of system so you have the steady state systems are us time independent um way of dynamic and this is a more common approach usually done in the context of system but there also you know there also framework where it’s like time dependent Dynamic models that usually have it sort of like SE often very smooth very time in of temp process but then because of the sort of event that we try to understand are quite really discontinuous really very rare So this sort of Two end members doesn’t really work out for us that is exactly what um I’ve been doing the last C of years trying to develop framework and tools to really capture this non3 processes with that led to the development of model that we worked on so the goal of the project was to try to explicitly stimulate not bu that look like this FL event I’m not going to go into all the structures here because it will take me aot of time to think about it but I just want to just highlight some of the way some of the key thing that we need to get right when we want to so we need to really you know find a way to in or horning on the quantity of carbon that found with with this kind of but also we need to account for that quality we’ve been discussing a lot about quality quantity of car um lately this workshop and you know with the quality also it tied back to kind of the the input and theen of different contributing to the eventual at a point of sample so we need to find a way to unfold all this information within the model and that is what we try to capture with INF there so the idea is so you have like a backround FL which is like normal signal variation but also because this FL event introdu sort of continuity the system so we need to find a way to capture that that is what we try to here in this what I think the Bic premises of model is that prop model follow time period of time um time of the event you have a day of and then that’s within that there there are also some properties that we need to includ in terms of like the reactivity of theal that is what this are represent with all of that we could me put them all together within the time point and then we propagate those information time and we have to do that in a way that it’s all continuous in time having to stop the that is where um I I say the unique part of come in here so I will just want to just show some result of how sort of SL event chemistry I’m sure some way we thinking about that so here I’m showing you time series of um um one campaign that we conducted in the winter of 201 21 to 22 I guess um so this is a very unique campaign because we were able to kind of really capture you know or at Le in the few get some measurements long period of time at least of three months or so um so right now you seeing kind of the water discharge the opp side of the slide below that you see [Music] the water and quite clear you could be like a really um really pck event indicating that this flow really with the call so we were able to kind of constrain what we thought out in the field doing this [Music] time and we able to kind that they were kind of like deposition of lay of new petent there so we could we could take in all this information and could try to you know see how model represent that that is what I show there so right now you see some prop the left hand side of the slide point from extraction for dip for s for anded in different colors and then on the right side is where the model people so if we put them together we’ll get something that looks like this um giving a feel of how the model before event interest between model and data I think what is clear from here you see a system that is more less like in the steady stuation reduced and then produced with Serv and part lay produ and then obiously so this is more like more the stady situation we find but then with 25 cm of sediment that we got during this particular period of time you see a different different feature So within 25 layer of the siment you see s penetrating and deer down in siment little very little was form in particular layer to lay and then obviously main also kind of show very little faity there but also what I wanted to point out here also a slow reorganization or siment this sort of compare all the different profiles to see how profil trying to you know reorganize sort of EV but also what is also interesting from this very slide um is the fact that at least you know qualitatively you could argue Tom mod was able to kind of capture some of the Dynamics that we see here but also interestingly you see that ship in the s m transition zone from our 30 cm just the of FL and to somewhere around CM after so they kind of event to really res to this of sh maintain transition so this a really important um thing that I want to highlight here but also we could even go further to really kind of use this model to kind of really query what exactly going on deeper down in we’ve been discussing yesterday about this radi ladder and all the different Pro so there Pro canol anything taking those process and they really yeah I’m highlighting some of the interesting process that involve s me gener more deer than siment you have S reduction and just between that you have aob cation of processes that involve M it and also that so what we could do with you know with tools or at Le data set that we have is to try to quantify um how this FL event really affect D we try to do here so what we did we try to quantify this tetri ratio between the being produc and the S being interaction and that is what we showed there with r c to S ratio um so this theoretical umry that you normally attain if those Pro the one operating and then what we try to do here we try to quantify what is the change in this CTS ratio in respect to thec relationship and that is what we show here so just before the FL event we see that things overing between these two mod soft consumption the siment but then after the FL event see like split between the two model surface reduction so and part of the layer of theim become moreing another part of the become more and in so of like disconnection between two mod of ruction that is being by particular and with models we could also kind of look at it more and that what the model was trying to show you showing this Shi in the surface and this a maximum deeper kind of and consequence for main main shiftx to and it might have consequence and eventually to the water to the atmosphere so these are some of the way that we think flood event really might about chemistry Al I just want to touch on on just what it means in terms of the DI FL we’ll be mentioning a lot about FL FL so when this kind of flood event happens this this is entirely different time of the year this in8 we have two just two data points so with model to look what exactly happen when come to the that is what we showed here so during this particular flood event um in the spring you have about 30 cm of C deposited that 30 cm of was very poor in organic carb and that resulted to a different sort of profile from that of the Fall where we have a small sediment 10 cm of sediment but very rich in organic carbon and this resulted to two different mode of dip cying within the segment so that what we see here in this particular slide I just want to just put a slide here to really capture what I’m saying so if integrate all of this stuff together to get a small kind of siment budget and that what you see here so th siment material that fall in organic carbon compared to that of sent layer reach in organic you could see a differential um output when it come to VIP a FL out of so in one mode you have kind of internal cycling of di storage of di during because of this very large but then in the other layer you have something that is more kind of interface cying of a small layer of siment deposited then relase more so this kind of FL event depending on the quality and also the quantity of organic carbon that comes with it it can result to different mod of are something I just want to highlight and then lastly then the question is okay what then happen when we have multiple flood event occurring so we have one in the fall but I’m one in the spring and also probably one in the fall what happen What what will the consequence of a flood event in the path affecting a flood event in the future and that is what we try to quantify with once again mods and you know that brought into the idea of memory effect so so what is the effect of FL FL that’s what I want to show here so what we see I from this little experiment is that at you within the S you basically see very little effect when it come to memory but yeah I’m showing you a plot of the different pathway for [Music] carbonization again on the on the right hand side you can see the relative difference between the FL the past and FL FL future as well as the relaxation time scale timea back to own State what you see that in the Sur espe the prot in the very little marginal effect when to but we go much deeper down the siment doesn’t see the the effect of the path FL in more more more present in the future and that is prominent foreally like s reduction and meten that has a much higher longer relation time scale but also much longer much sler processes if you will so these are some of the ways we trying to see how FL event really affect chemistry so last I just want to just point out like some of the work I’ve been doing on the side and with I’m thinking about where we go from here so basically what a work have done is looking at things from a local perspective hopefully the goal is to try to upscale what I’ve done not only just on a larger facial scale but also perhaps some different region that might have similar characteristics like mine and currently there ongoing project working with car set at LS in France we trying to develop a 3D couple models and uh yeah hopefully investigate what is the last and then we to think about what other kind of extreme event um I’m only speaking here of FL but might also beeme event that I have and one of that is the to ding so when you have a strong connection between the deep sea and the the on the river Jon canons so they bring in very large input of sent material with the so what would be the impact of this ofing delivery are something tried doing my PhD but for the LA of time I couldn’t and then the last thing I just want talk about is you know some of the way that the community will try improve some of the mod that we have you know earlier this morning I prop the IDE of something like like to models so with that I just want to say thank you and if there are any questions just let our next speaker in the modeling session is hello everybody um seems like stany has shown a lot of things that I’m going to show so any um that was a great talk standing so today I’m going to talk about uh B diagosis and and how we can build a model from the observation okay so we all know the importance of the sediment processess so I’m not going to go deep inside this you know importance but just to you know make an introduction these are important sites for nutrient Recycling and has been overlooking the global models as you know you know and that has been one of the you know aim of this workshop and the benic processes in the share settings are becoming more and more important as the different governments are pushing towards CVR approaches which is kind of using the um sediment as a storing you know storage for carbon dioxide so our overall aim is to construct uh you know to to fill data Gap and to construct like a regional to you know Regional to Global Bic model so when we talk about the model the models can be you know categorized into two broad um you know types one one is empirical and other is mechanistic model so empirical means you know when you have a lot of data and you can use this this data to construct uh like a parameterized model which really works uh well in predicting the data uh and and on the other hand the mechanistic model does what it does is you know models the underlying principle of different processes this this type of models may not be as good as the empirical model inting the data but can give us insight into the processes that are you know happening in the in the system so I’m going to you know show you example of this mechanistic kind of model which is basically work from my last chapter of my PhD phis which I defended um around 6 months back so the study that we have done is based on bod Basin so this is the Basin and the B Miss is a 70 M deep seasonally hypoxic F which is connected to this shelf in the Atlantic Ocean um through a 20 M deep seal and and the V for B is located in in h as Canada noia Canada just you know in the coast of Atlantic so this places also a very interesting place because you know you see this you know red triangle this place is a t City station which is collecting water column data for since last 30 years in a weekly resolution so uh one very surprising thing is that you know even though you had a lot of water colum measurements there was no dedicated study on the benic processes before my thesis so just to give you a climatology of the botm water oxygen concentration you can see this in this region um you know the oxygen concentration in the bottom water increases every year during the winter winter months and slowly decreases consistently uh throughout the summer and now you can see that there is a really nice seasonal cycle of this oxygen so we can kind of call it as a natural laboratory so you can you know study different kind of uh gentic processes at different oxic levels in the the natural which occurs naturally um so what we basically did we went out in the field and collected sediment core through a multicore and in different seasons for around 3 years uh what we did and on The Collector S course we um measured the oxygen profile through oxygen micro sensors we collected the four water nutrients through Ryon sometimes also by cation and then we performed whole cor incubation in xt2 to measure the vtic fluxes and also we did some anop incubation experiments which by which I mean we basically section the sediment four at different um depth intervals and then incubated them in anoxic medium and then by the you know change in concentration of DIC and amonia we can kind of calculate the reation rate at different deps so now that we have a bunch of data oh no before that going into this so I’d like to uh you know uh I would i’ would like to you know show some of the observation interesting that we have done you know what we saw from our P Water profile so you can see um so these are the three for water profiles of uh these are the C water profiles of ammonia DIC and nitrate and these are not only from a single set of you know field but like for the seasonal average of different years so you can see that um the the ammonia and dr2 concentration kind of increases for around 5 cm and then consistently decreases towards the bottom remember that this place is very uh you know high in organic matter the organic matter in the siment is around 5 to 7% so you know it’s very I mean it’s surprising to see why this kind of you know I mean ideally you would expect your Ammon and D conation will increase with dep but it doesn’t happen and in case of nitrate you can see there is a kind of a very high concentrations of nitrate you sometimes see in your in our P Water profiles whereas the oxygen penetration depth in this segment is only one mm um so that is another interesting fact that we found also and some of the things that we have found during the during the um you know you know field sampling we saw that these sediments were you know inhabited by long tube WS which kind of can do deep bio so can those be responsible for this kind of profiles we’ll see that in our model and another interesting fact that we found some circumstantial evidence that these sediments may be inhabited by foraminifera which are cap of doing the nitrification so can this nitrate be leaking from their um cell so that is another questions also we have so now that we have a bunch of these data now what we can do so we can use this data data to model and look inside this data and often we have seen you know observation and modeling together can give us insights into the natural processes and help Discovery new things so just to give a general overview of the type of model which is basically you know siment diis model which are reactive transform model and Stanley has shown a lot of those equations already but I’m just going to give as uh like a general overview when you do when we do the siment uh modeling we have two phases solids and four water and um when and when we have a reactive transport model uh if we consider the transport those are basically the adicction diffusion biot teration and bio irrigation and in terms of the reactions we can you know do a lot of reactions according to the complexity that we want in our model but they generally are aerobic respiration and arobic respiration ification etc etc accordingly I mean the complexity you know as part our complexity that we need and so these are the basically the model equations which I’m not going to go too much details because Stanley has already shown uh but the thing that I would like to show that you know so these two um equations are basically for po water and for solid phases but interestingly one more equation that we would like to you know use in this kind of settings because we want to consider the intercellular storage of nitrate in foraminifera which can be you know you know expressed through this kind of formulation where our infection and the diffusive terms are zero but we have a nonlocal exchange of you know of a particular solute in nitrate um so and this is the reaction uh schematic reaction background that I have um considered um so in this in this in this diagram you can see that I have considered all the major chemical you know major components major you know um elements that are important for vtic diagosis um in coastal regions like carbon oxygen nitrogen um Iron manganese sufur cycles and then I have used this model to represent a yearly average by solving it in a stady state the transport processes that I have considered is the fusion biod derivation and bio igation and I used all the measurements that I have done for example this organic material imation rate for water profiles and B fluxes to constrain my model so this is the model results um so when we run the model you can see that you know the model result which are the solid line represents very well the seasonal average of the poor water and you know depth profile for example the model really nicely presents or simulates the uh solids the TOC CBN and the iron the reactive Adon and pyite and also the you know different solutes and and if you see this Di and ammonia you’ll see that by uh considering uh like a deep bio irrigation this nicely represent you know nicely you know predicted the you know this kind of you know the typical shape that we have seen in our um field data so and when we look at look into these fluxes we see uh that uh so first of all like let me explain this figure a bit so the the white bar that you see is basically the average uh four water flux sorry average btic flux that we have measured and the red one is the total model flux which is further divided into diffusive flux and irrigative flux U so you see that oxygen flux was dominated by um uh diffusion however when we saw the um DIC and ammonia flux the bio Iration flux was almost 50% of the um total flux that we have measured and in case of nitrate bio you know diffusive or irrigative flux was alone could not you know explain whatever we whatever the influx that we have measured however when we considered the foral uptake our model was able to replicate what our you know data was suggesting and indeed when we then looked into the for mineral abundance from our study side we found that the forer abundance was dominated by St for for which is a dentify forum and when we calculated the abundance and their in cellular storage and the the the rate that we obtained was very much you know in the same order of magnitude whatever our model also suggested so now that we know we see that you know we have a working model we can ask a lot of questions to this model for example what are the pro what is the main um electron acceptor which is you know um you know what you say you know responsible organic matter oxidation so you know as expected in coastal coastal regions sufate reduction is the major um you know organic matter degradation pathway and other question that we can ask is what is the rate of dentification and who dri the dentification uh is is it is it the coupled coupled you know nitification dentification direct dentification or pular dentification and in this case the answer is uh for denitrification was driving the majority of the denitrification that we saw in the um you know in our field so that was just an example of how the field observation can shape a dtic modeling so now when we you know you know start going towards a global Bing model uh what we need now we have to consider like what are the things that we need what are the assumptions we have to make or simplification that we have to make which you know can be useful in you know which can be you know helpful in you know modeling the world or the region so first of all is when we are doing something in a local scale it is very easy to sample and we have a temporal resolution so we have a lot of data which can be you know used to shape our modeling but in case of a global model that is not the case so you have a lot of you know data uh you know data availability is one of the constraint uh so so for for that we need to have you know um as much data we have is good and for example if we can have like a b flux compilation data that would be a great place to you know you know verify whatever a model is doing is it a real one or you know it’s not it’s not a real representation so that’s why you know getting a benefic uh CL comp compilation data is a good one and then we have to do one more thing is by is finding Universal parameters so you can see you know when we have a lot of reactions we have reaction rates in reaction rate is one of the parameter so uh if we are doing something for for for region a or region B and we are doing it independently then we can tune the parameters to match the U uh observ that we have seen in say region a or region B but when we are doing a global region or like a global model then we have to like use a parameter which kind of works for every region and kind of gives us a not too wrong value right so for that we have to find the parameter and we have also have to like find uh natural controlling factor for example temperature uh let’s say you know when you have a high temperature your rate kind of goes up you have low temperature your rate goes down according to the Q10 uh kind of formulation so you know we have to you know kind of find those kind of controlling natural controlling factor which will scale you know which will be which we can vary from one region to another region naturally and that will you know translate into different rates which can you know give us like some you know nice representation another thing could be like the de and the bottom water oxygen uh could be some of those natural controlling factors for example when we have low bottom water oxygen that time a lot of this mro can die and biotar or bio can go down so that is another kind of factor that we can have like a natural controlling Factor now when we I mean when we let’s say we design a like a btic model then we have to implement it some so so we have a lot of of art system models for example ROM University of Victoria esm and know gfdl esm so we have to use any of these Global models to uh use you know we have to use our venting model in combination with like a more bigger General uh art System model for implementation uh but the challenges remains is the scarcity of data and the difficulty in ground uh ground TR whatever we are our model is generating is it is it a real observ you know real representation of the field condition or not um so and and some of the some of the challenges includes that we you know in in a in a um open ocean or in a you know um where we have like a like a l low data density that that way we have to like you know kind of rely on the snapshot of the obser ations rather than having um like a seasonal observations that we can have in the postal region but not in open mocean um yeah and then yeah so now when we start building a uh Global model some of the uh model constraint that we need to have is like for example porosity it is a permeable sediment or impermeable sediment you know your transport very different in you know either of these two regions so I recently discover I mean found that there is a paper uh you know Martin 2015 uh um which kind of has used observations and some machine learning algorithm to kind of generate a uh like a Geor referenced map of global porosity which kind of maybe of use and similarly you know the sediment density also so has been similarly modeled with a mchine learning algorithm which can be used as a like a forcing forcing you know you know forcing you know parameter or in our model and then um certain things we can get from the earth system models for example the bottom water conditions oxygen concentration nutrient concentration which are important boundary conditions uh and depending on the um uh you know depending on the System Mod we can also have uh the particular particular organic flux to the bottom and sedimentation rate um so these are the things that we can obtain from our system model and I would like to um you know emphasize that correct representation of a p plus to the C floor is very important for Ving module because um the availability of the organic matter is what drives all the next processes and last but not the least biology can influence a lot you know the Bic processes by biotargeting the sent or bi you know by biting the sment so can we use the you know vtic bio data abundance biomass as a scaling factor for bioturbation or bio irrigation along with oxygen concentration so these are some of the other factors uh to consider as a you know consideration of yeah so as a starting point we can you know think about think of as a like a simple um simple representation of the complex model that we have saw that we saw with we we which just can have arobic respiration dentification and an arobic respiration and and for First Step what we are do I’m doing basically um you know taking a like a map of the word and kind of parameterizing the parameter in the simple model from by the with the observations at different regions of the world and trying to find the simple you know like the common parameters or like a common parameter set which kind of represents um the most of the observations you know different from different stud sites um yeah and by doing so we can you know so now the jordanon paper can come very handy you know when you know this kind of paper which are basically a nice you know um nice data set you know like a extrapolation of a like observation data set can come very handy to evaluate the uh output of a benting model because you know this we know is made from observations and how close our you know model output go to you know to this observation kind of can give us nice confidence or you or help you know correct the parameterizations in our model and so I mean so this is for the oxygen but if we can have something for diic or you know ammonia or nitrate can also will will will come very handy um for model development so in conclusion we have seen uh the development of a local sediment diting model from which can be shaped by observations and have gone through some theoretical basis of dtic model and gone through the early consideration before we start developing the mod for uh for a global system and any suggestions people have uh which can help uh you know some of the ear considerations are uh I am your last speaker for this session and I’ll be talking about benic needs for couple Global models of fisheries and marine ecosystems so I often think about Marine ecosystem structure in this idealized simplistic kind of two food chain way um where we’ve got a pagico represent how um they you know interact in Shallow shallow areas but in the deep sea they’re disperate so we’ve got kind of our important players here for someone like me who tends to study fish and zo Plankton we’ve got our phytoplankton that are primary producers that are producing oxygen they are fixing organic matter forming the base of the food chain and they are an important uh player in the biological comp we then have our different consumers our Z Plankton fishes and benic invertebrates uh that also play roles in carbon cycling so our Z planted and Fishes um have roles in the biological pump both with passive and active U movement of carbon and then our benic and vertebrates are more are you know at the other end of that pump they’re at the the bottom where they can potentially sequester carbon or prevent its sequestration uh and then our fish and shellfish also have another important role in that they provide food and jobs for people so uh I have been using models to study the drivers of fish population abundance from a bottom up perspective uh there have been number of studies on this already for in this study by fre lad all 2012 show that net primary production is not a good predictor of Fisheries catches across the global ocean so here we’ve got large marine ecosystems plotted as the dots and you can see a pretty insignificant scattershot of a relationship and you know NPP doesn’t really fully capture the amount of energy at the base of the food chain or these two Foods chains you know because it doesn’t take into account things like the microbial Loop so let’s go a step up and instead look at the amount of productivity going to Z Plankton and going to the C4 here we could quantify it with the Z Plankton fraction or the Z ratio and indeed there is a strong positive relationship between the soul plank in production and fish catches and then also a strong strong positive relationship of the PE Ratio indicative of the export fls and fish catches not only do these metrics of secondary production and the location of that secondary production tell you about total Fisheries catches but they also tell you about the type of fish you’re going to get in these catches so here’s a study by van dender and it all that look at the ratio of NPP that stays in the plaget versus NPP that goes to the Sea Flor and this ratio dictates what type of fish are found in observed catches so when you have a greater fraction staying in the blagic get more large blagic fishes like Tunas when a greater fraction goes to the Sea Flor you have more large Predators like demeral fish CS flers Etc so we’ve got this importance of secondary production and particularly the amount of production going to the Sea Flor being important for total fish catch and for demersal fish uh well how important are demeral and benic Fisheries here I have some statistics from the latest um fish agriculture uh organizations report on the state of world fisheries and aquaculture so here you can see the time series of wild caught catches and just the numbers of those catches in weight over time I’ve highlighted some invertebrates here in Orange uh and then you know box some of the numbers that are important so for example demersal fishes uh two specific species make up they’re in the top 10 species for finfish total so Alaska poic is 5% of all finfish catches and Atlantic Cod is 2% of all fish catches um when we look at crustations and mollusks they are less than 10% of that of fishes but still large amounts and our benic invertebrates are also contributing to aquaculture particularly again these crustation and Moss categories that have just continued to increase over time where Crustaceans are making up 10% of total aquaculture and mollus 26% and must note that the mollusk does include things like squids um but again we’ve got a not insignificant contribution it’s particularly easy to see the contribution in a regional context or if we instead change from looking at catches and weight but in terms of economic value so there are some regions that are highly dominated by demeral fish and meic vertebrates like crabs so you can see this in the bu of main e are very sea area so um even if those uh catches were not a huge proportion of the global total very important for the people in that area and then uh here we’ve got Seafood export value our demersal fishes are 10% of that and now you can see how much greater those benan vertebrate catches become when we see the value of things like crabs and lobsters which can be quite expensive so they have an oversized contribution to economics so given the importance of Fisheries to people and economies um and then you know they also their function in the ecosystem I am part of this Global Group called fisheries and Marine ecosystem model intercomparison project there are over 80 modelers across the world that have either Global or Regional Marine ecosystem or fishery model and we are trying to answer questions about fish and Fisheries seafood supply Marine biodiversity and Marine ecosystem function and our goal is to bring together these disperate models to better understand and project the long-term impacts of climate change and use these findings to inform policy there are about 10 Global models right now and then four Regional models but these Regional models have instances or implementations in multiple places so I’m going to focus on kind of the global models here uh these ecosystem components can be represented with different size classes for instance uh the boats model has has you know maybe three fish types that are defined by their maximum size and then within that fish type there are multiple size classes or instead we could Define fish by their functional groups not necessarily fish but higher tropic levels so the feisty model has small plag fish large pic fish demersal fish and beic invertebrates other models in fish represent tropic levels only some represent all species um and then others have a more like history straight stage framework these are all mechanistic models and they are representing the links of both mass and energy so they could do this with P2 networks um they might use diet composition data from observations um or they might actually just transfer uh keep track of the energy transfer and we simulate the fisheries and marine ecosystems by coupling in a one-way framework to Earth system models and so we get different outputs from the earth system models depending on the specific Marine ecosystem model so for instance boats uses NPP and C surface temperature as it’s forcing The Feisty model uses fled temperature bottom temperature meso plankt in biomass and mortality and bottom detritus BLS uh I think we’re pretty familiar at this point with Earth system models but just uh as a reminder these are Global and they are trying to capture all the interactions between the atmosphere ocean land and ice and they specifically close the carbon cycle by simulating biogeochemical Dynamics most of these models and most of the biogeochemical models within them were really developed to understand the carbon cycle and not developed with Fisheries in mind but let’s first think about the fish models themselves our um our studies like the freed Lin at all study have shown the importance of export production how many of these Global models are using it or something related to that and then also given the importance of demersal and Bic Fisheries how many are representing them out of our 10 Global Marine ecosystem models only two are actually using export carbon flux uh the aosm looks at this export in the iic and mesopic whereas The Feisty model actually looks at it going to the C4 and then only three of these 10 are simulating demeral fish or benic Inver so four have you know one or both of these components not great out of 10 and then let’s backtrack to what we are forcing these fish models with and look at our Earth system models given the importance of export production or just kind of secondary production in general how well do they simulate export detrius and sediment Dynamics so I’m just going to show a little example from Kelly Carney uh and her colleagues at the Pacific Marine environmental lab and the Las fishery Science Center so they’re studying the Eastern B sea uh and this is a biogeochemical model inter comparison they did for their region using two Regional models the Banis model and the best npz model and then one Global model Cobalt and you can notice these have varying degrees of complexity if we just look at the number of State variables being tracked from very very simple with the B at all model U to very complex with the best best NPC model that is uh not only keeping track ofic processes but Bic processes and Bas processes we can zoom in on the water sediment Dynamics again similar levels of complexity as the number of State variables a very simple model is just do a sinking varial the Cobalt intermediate complexity model is doing sinking with burial remineralization and denitrification and then we’ve got a little bit more complexity with the best MPC that is also refer presenting uh some of this material becoming Bic trus that is the food source for Ina that then um also contribute back to uh nutrients with excretion and respiration so this is just an example of two Regional and one Global model we can look at how this is simulated in our models that are used in the climate model intercomparison project this table is from safarian at all that was trying to track improvement from CIP 5 to CIP six um and so here we’ve got this really detailed model of or detailed table of the different models from SE 5 to cement 6 so that’s what each of these alternating rows is how they are representing nutrients and elements sediments organic matter um tropic levels actual name of the biogeochemical model and the ocean model that they’re using um as well as a few details about vertical and horizontal resolution so if we focus on this column about sediments you’ll notice we’ve got a lot of nodes soem models just aren’t even representing it at all and then we’ve got some really simple representations of a balance model or a box model but then it looks like some places are getting a bit more complex especially moving from SE five to se six of using a meta model or using an actual layered sediment model how well do these models perform when we compare them to observations uh so one set of observations for comparing against is export production out of the eodc Zone this is our fifth seith five round or seit round uh export estimates uh these were compiled by L clutter adult compared against two different observational products for export production and you see they just kind of have moderate comparison with the observations the correlations max out at six their standard deviations are all greater than the reference um and then the root mean Square errors are kind of all over the place what about improvements from C 5 to C 6 uh this was shown in the sarian at all paper by looking again at Carbon export at 100 meters uh compared against observations and what you’ll see in these little box triangle plots is the bias of export production in five and the bu so uh you’ll notice a lot don’t change very much whereas others actually get a bit worse in C not encouraging so um I’m just going to go back to this in comparison study by FY Carney at all who actually pulled text from the stock at all 2014 description of that Cobalt Global model that is talking about these caveats that you know a number of simple ifications are made in how they treat organic matter reaching the sediment one of these is that burial and remineralization happens instantaneous uh they found that these Solutions are not sensitive in deep ocean waters where the resident time of nutrients is measured in centuries but it may influence results in shallow shell environments these are poorly resolved by present Global models but may be more meaningfully captured in future high resolution simulations so the physics is already pretty bad in these coastal areas because of the low resolution but there have been so many developments into high resolution Ocean Models both globally and regionally so once we’re getting the physics rate we really need to start thinking about how to better represent the sediment Dynamics as well so uh as we are developing these sediment models or Ben thic models I’d like to propose that we meet these two modeling needs at the same time instead of just developing our sediment models for carbon cycling can we also develop a model that could be useful for Fisheries studies as well so we want to improve export DET trus and sediment Dynamics for the carbon cycle but also think about the importance of these things to our consumers and some of those consumers are commercially harvested While others are prayed for commercially harvested Fisheries uh and so we’ve got you know a number of ways to go about this I think we should just kind of stick with the framework we use for the project um where we are using functional types to represent organisms it is one way to distill things into their roles in the ecosystem so for instance our phytoplankton um often get uh defined by their size which impacts the nutrient kinetics um and light Dynamics but then also you know what nutrients do they need or not need oril Plankton are probably the most least well represented in these models only thinking about size um Fish can be defined by size they preferences their fishery status habitat for is another thing there or kind of reproductive Dynamics and life cycle Dynamics um and I think those would be great places to start for benthic organisms as well their size is important because physiological rates scale with that um feeding mode seems to be incredibly important in this benic environment but also thinking about habitat are they on the sediment in the sediment um and what is their reproductive cycle like as well and then also thinking about the specific processes we need to represent in theic for instance just focusing on carbon we’ve got fixation respiration passive sinking active migrations most of these Global models do not have active migrations or they only run them in like very very small um specific studies and then and benic what are the important processes we’ve been talking about during this meeting deposition resuspension mineralization bioturbation bio irrigation and burial uh and happy to discuss more of this uh but I will just acknowledge my coordinators in fiship and if anyone does want to create a model intercomparison program um I’ve got lots of insight to that one of the co-coordinators uh and then I’m also just going to acknowledge my spy model developers we are four minutes early so I will take questions and you can also ask any of the other speakers questions that you may have yes J so in relatively shallow shelf ERS we know there’s a lot of Bic micro alal biomass in the settl we also know that there’s a huge difference in Z plank in abundance in daytime and night and nighttime and a lot of people a lot of people but there have been some U studies that have suggested that demersal zanin live in the sediments during the daytime and feed on those microalgae consuming that carbon that’s potentially in the sediments and then go up into the water column at night where they’re eaten by other organisms and so it’s a car trans a potential significant carbon transfer mechanism that would throw your balances off because you have carbon deposition and an assumed carbon abundance in the sediments yet it doesn’t balance what’s what which should be available in the water comp so I just wonder if anybody if that’s been any discussion that’s happened with with these Fisheries modeling things oh no not at all because or at least not on the global scale um because no one’s resolving benic primary production everybody assumes it happens completely PL and it’s not um benic algae whether they’re micro or macro yeah so yeah another another real important thing to consider in coastal areas two two questions comments so you’re export production right the the I don’t model it but biochemistry models to but a lot of the models we see to consider silic cycling and in the crystal Zone with anic inputs with fertiliz but nitrogen phosphorus what’s happening is that silica is actually limiting nutrient so what would be like the base of a healthy move so I wanted to bring that point up um I think in some of these models it’s just like a remineralization recycling efficiency at some water depth and that maybe I think that’s incorrect conceptually um yeah some explicitly like consider Sila ballast um and I think it has a different remineralization rate than carbon would um but I think maybe probably the bigger uncertainty would be those Rivery inputs yeah into these models yeah groundw inputs too I think um and the second question that I have is are you or is there a way that you can consider viral not you can consider viralis because viral Lis can account for almost % of vital plank in di how for example so that’s not a grazing pressure that’s not like a like something these have different loss terms based on which biogeochemical model there are some have like just a linear loss term that you could think of as metabolism and a quadratic loss term that you could think of as Predators but viruses would probably also function more on that density dependent scale uh as a quadratic term um others actually add it as a separate term or maybe just one just the pisky models but people are I know they do consider viruses as a loss term for phol yeah I was going to ask you about the for vtic unification right and I thought that was really interesting is putting the the bio and Biogen chemistry so can you just constrain that by the nitrate for water because I was wondering how you kind of mechanistically implement that or you just kind of yeah remove the nitrate and the PO Waters to get the the profile correctly is that how you can con strain that the model yeah so in the model um what I did is basically like con intercellular storage of nitrate and then kind of use that nitrate as only available for the nitrification not to other but then the PE that you get like know the yellow color line in the that you get is basically uh probably like the the one that you measure in the cold water that is probably because of some lices or like leak from the but ideally they should not there it will be inside there not but whatever we measure is kind of could be indication that got someh something we kind of got a very high nitri and the way I like you know constrain the rate is by you know like I tuned my transfer function according to the like the whatever the that I’m you know my liation CL and cannot match that I see when I my transfer function to the intracellular and by balancing that to and kind of getting that going to that that’s how