 Good morning. I'm going to talk today about permafrost modeling, but before going to permafrost modeling I'd like to give a little background on what is permafrost and how we measure it. So permafrost is a frozen soil and the actual definition is it's the soil material which stays frozen for more than two consecutive years. Another definition which I'm going to use in this talk is the active layer sickness. So the active layer sickness is the depths of seasonal soil. So it occupies 24% of the land in the northern hemisphere and 80% of the land in Alaska. And when permafrost starts to sow, it influences on the infrastructure and you can see that from the pictures and it's also influenced ecosystem. And the example is if there's a catastrophic event such as a forest fires then the forest fire burned the trees but it's also burned the organic layer on top of the soil and if that burn is quite severe then that organic layer might be severely burned which means that and it turns out that that organic provides the necessary it protects permafrost from sowing because it has a very low conductivity during the summer and when it's gone then the permafrost is vulnerable to sow and if the ground has enough segregated ice then what might happen that small pound, small pound might cure and depending on the amount of ground ice that pound might be developed into a lake and that lake called the thermocars lake. Okay so if we're going to do any economic planning for the state of Alaska I think it would be wise to include some results from a permafrost modeling. So how do we measure permafrost and this is the basic setup we have permafrost this is how the permafrost observation station look like so it's the air temperature we measure air temperature we measure snow depths and we also measure the ground temperature and this is the example of shallow borehole measurement station so it's the sensors usually at every 10 centimeter depth and they measure temperature on an hourly basis but besides those temperature measurements we also have soil moisture sensors which measure soil moisture at the unsaturated zone and it's usually up to three soil moisture sensors and the depth usually up to 60 centimeters so and that soil moisture makes the modeling of permafrost a little bit complicated because depending on a soil structure it might hold different amount of water and you can see I showed on the table the different soil structures so and what happened when the temperature drops below zero degrees it steals some amount of that water remains unfrozen and of course we don't input that explicitly into our model I mean the information about soil texture what we do instead instead we try to fit our unfrozen we call it unfrozen water function or in this case it's a curve to the measured soil moisture data from the site and we do that by basically yeah fitting it and the A and B it's our parameters it's gonna be more clear on the next site here is the mathematical this model so it's it's one of the formulation of the heat flow equation and in this case we use the enthalpy formulation and the advantage of this method is that we don't have to explicitly treat the free south front three from a free cell moving boundary and that's in order to have a unique solution it's of course have to be equipped with boundary and initial condition and that's the general setup we have a air temperature at two meter depths and then the some ground flux at the bottom and it's usually and I was set up it's which I'm going to talk later it's 700 meters below the surface and initial temperature distribution and the unfrozen water function it has the following form so here is the volumetric soil moisture content and this is the coefficients which I talked previously this is the thermal capacity and the thermal conductivity latent heat okay so how we define active flare and there is a two major definition of the active flare one is based on the zero degrees divider so when whenever we hit that cap then we say that after that depth it's it's going to be active flare however in our model we define a little bit different way based on the unfrozen water content so we defined the so-called unfrozen water saturation coefficient and whenever that saturation coefficient intersects with unfrozen water function then at that whatever depths it happened we treat that depth as the active flare depths and this is the schematic representation of the model so it's GIPL2 MPI and GIPL stands for Geophysical Institute Permafrost Laboratory and we try to build our input in all in ArcGIS format format and the output is also in ArcGIS format but now it's in NETCDF as well and I'm going to talk a little bit about my project which I was involved so I did the modeling of in high special resolution for the state of Alaska of the permafrost dynamics for 120 years starting from 1980 till the end of this century and I used the GCM data which was downscaled by scenarios network planning for Alaska group to 2 by 2 kilometer resolution and this is monthly averaged air temperature monthly averaged precip input data also include soil organic matter initial temperature distribution and the most importantly the surfacial geology map for Alaska so each that class in the surfacial geology map has has a number of layer assigned to each class and each of that layer has its own thermal physical properties and has its own unfrozen water parameters we I equilibrate the model and after the equilibration I calibrate it for the several sites along the trans-Alaska and highway and this is the result from one of them and then after calibration we validate the model with the IPY dataset from 2007-2009 it's more than 60 stations it's a mean annual ground temperature mostly at 20 meter depths and here we have quite a good correlation and then I also validate our simulations with the mean annual ground temperature for US schools project for Alaska and those mostly shallow borehole stations from one to six meter depths and I went a little further and did the validation for the active layer thickness and here the measured data is from a circumpolar active layer monitoring station and my result look in a following way so I calculate the mean annual ground temperature which is above zero and which is below zero and for for every year and then I average that every 10 years so I have 12 decades and I mapped it for 2 meters 5 meters and 20 meters and on 2 and 5 meters I got the trend decadal trend which is 3.5 3.7 percent and for 20 meters I got it a little bit slower it's 2.4 percent so when I did all this modeling this is how the graphical results look like and now if you guys ask me the question so what's the very simple question what's the vulnerability and resilience of permafrost to sell if the climate will continue to warm I still won't be able to give a simple answer but if you're curious we can talk later about it I'll be in the neighborhood okay so I add a little intrigue I think I can move forward I was visiting CSDMS two weeks at the end of this summer and when I just got there we the idea was to incorporate GIPL model into CMT environment and we were able to do that in the very first day which was quite surprising but it's mostly because I did my homework before I rewrote the model so that it satisfied the IRF format and my feeling is that so far as your model satisfied that format it's quite easy to make it CMT okay and the ultimate goal of course to couple that was the ideological model and I hope that with a will funding and enough desire that can be done and at the end I'd like to acknowledge NSF and State of Alaska and express my gratitude to Arctic Research Supercomputer Center those people do a really good job over there they provide their really good services and I appreciate all this the consultations and of course CSDMS without them I want to be here thank you for those of you who have forgotten Elkin was a student award winning he did a student award winning job and his graduate thesis so questions for Elkin on permafrost modeling yes Michael this is very interesting so obviously knowing that the permafrost is is really important from a carbon cycling point of view I know that there are other efforts to try to do this because even your model requires if I understand this a lot of ground data so there are folks who are looking at things like and I don't know the details about this I'm hoping you call you will sort of say oh yes but there are folks looking at things like n-factor and degree throwing days to try to develop a proxy to then you know do these regional estimations have you compared your results to those type of proxies there is a yeah there is a thank you very much for the question it's very interesting but there is yeah many people and they are doing different they they basically work on different approaches and one of them it's doing the this freezing sound factor analysis and we have it in our lab we have different versions of the model and one of the version is actually which is exploring that approach yeah there is different approaches and there is different groups and the other group is I think CCSM they they doing that is a little bit different way but the idea is the same it depends on the on the model and depend on the dynamics and depend on the what the question you you employ and in this case it's a transit it's numerical based so it's it's more dynamic uh any other questions or Elkin okay thank you thank you