Discover the Ideal Water Bill Format for Research and Development

Water bill format for Research and Development

thank you to greg and louis and dan bucchoff we give a nice introduction to dtc and first i want to acknowledge my my co-authors uh this is alisha bernadette xiaowu pao and nero min sergey these are the people and the dtc who are working on the hurricane problems and i also want to take the opportunity to acknowledge the tremendous cooperation we have at emc under the leadership of biola panda vijay yang kwang and many others who make this collaboration possible we heard a lot of discussion about uh history of hurricanes and also modern days and i would say over the past several years past three years the one of the most important development in the hurricane problem is noaa's hurricane focus improvement project i don't know how many people knew about this but this is a very ambitious goal the goal here is my understanding is in 10 years we're going to reduce the track arrow by 50 for day one to day five and intensity error again a 50 reduction we want to increase the probability of detection for rapid change up to 90 percent and reduce the false alarm rate 10 percent and the last quarter is very important we want to be able to make hurricane forecasts out to seven days and this is really in my opinion very ambitious goal in particular when you talk about the last bullet you want to extend the forecast out to base seven that means that you really got to number one be able to do a good job on hurricane genesis number two almost the regional model we're not going to be very useful you got to be able to be doing this from global models so how are we going to do that how are we going to accomplish such a ambitious goal in your opinion of course that if you want to do that you almost got to do significant research and development on all fronts that including model development model improvements that for example the physical parameterizations from surface hungry layer to competitive parallelization and you have to have to worry about the ocean and waves how they couple to the atmospheric model you want to run high resolution so that you can capture the intensity you want to do a better job on vortech initialization and also another important point is that you got to have good observation both near the core and also over the environment you also got to have a way to properly use those information to improve your model initial conditions and as uh rui and and many others already coming that ensemble prediction is becoming a very important focus tools and that's necessary for probability prediction data simulation and also the multiple model ensemble and uh when i look at this goal i mean a lot of people have found that many groups are funded to do research for hurricane improvement but in the end who is going to deliver the promise i would say come back onto the operational center that is going to be emc and nsc these are the people who are under 9 have to deliver those forecast improvements and so how are we going to channel all these forecasts all this research on hurricanes and so that we can have transition to operation and deliver those promises and this is where the dtc comes in and ddc as uh as louis introduced that it was formed around the late 2000s and the primary function really is to have a functional equivalent operational environment to test and evaluate new nrp systems over extended periods so that the research advances can be made into operation and louis also pointed out before that in order to accelerate those kind of exchange it also it would be extremely valuable to have a research community to be using the code that has been used for operation so in terms of dtc uh our current focus on hurricane is focused on the the edgework modeling system and we like to facilitate the transfer of that technology between research and operation by creating a framework for the nsap and the research community so that they have a common framework to collaborate and our initial focus here is to support the community to use hwarf which is the current operational system for hurricane at nsap then we would like to develop and maintain a test and evaluation infrastructure at dtc so that it can help us for pre operational implementation testing and to make sure that the code is robust and the nsa can use it with confidence and uh i'm going to please to tell you that we had a good agreement between emc and dtc in terms of code management for each wolf and our goal here is that the nsap and the research community can use the same code at least for the atmospheric component and we have actually done quite a bit of make quite a bit of progress uh porting the edge the h2o operation hwarf back into the world repository and in the coming year we're going to be implementing additional features for example and a plan for the 2010 upgrades uh into the work repository and later on also we'll be looking at features that could be implemented for 2011. so you may think that this is an easy job but actually it's not it's actually quite a bit of challenging task you know what uh the wolf system repository has been maintained by the mm division of ncaa and it's been evolving for example uh in 2004 we have the wolf version 2 and ever since every year there are new releases and currently we are scheduled to release the world version 3.2 and at the up at the nsap the h-wolf model under the leadership of the neurome who developed the h-word model was originally based on the version 2 and has not been communicated back to the world repository uh until until recently so you can see that with almost four or five years of separation there is a quite a bit of differences in the code even though the opposite and even the enemy core of the code so this year our major effort here is to put the ncep's h wolf operational code back into the world repository so that it can be uh the the component that are important for hurricane will be residing in the warps 3.2 that are going to be released and then later on we are going to evaluate those components that are intended for the 2010 operation implementation again based on the wolf version 3.2 will be imported into this 3.2 plus then later on we also be looking at the possible feature that we want to use nz1 to use for 2011 operation then into the same code then through significant testing evaluation we would have to be able to have a configuration a work h1 configuration for operation basically nsa will be able to configure this the operational code from the general world repository i want to tell you this is actually big very significant there's several reasons for that number one basically any university folks aoml or foreign research lab who want to do a hurricane research problem they can use this same code and whatever they do with them with feedback uh have a pass to feedback to operation and number two i know there's a lot of people are concerned about my code is arw your code is nmm i want to tell you that by putting the the hwarf code back into the web repository the arw gain can also use all the other features like the princeton ocean models the web coupling all these things that have been used for the operational edge work can also be used for arw so this actually allows the entire community to work on common code and there's a path to transfer from uh research to operation and i want to again give tremendous amount of uh kokuto to um our team here including xiaowu regia and and they they did three minutes of my work in doing this powering porting potting exercise and uh for example this is they look at 177 cases for the storm that occurred in 2008 and we have a virgin 2 which is the blue 9 and the version 3 which is a gridline and this is crypto forecast just to show you the uh as a reference and you can see that in terms of track arrows these two versions are almost right on top of each other and in terms of intensity forecast there are some still some small differences in early 30 days early early phase but they are not very large but you could also see interestingly that this is version 2 and version 3 between day day 4 and day 5 we actually are doing better than the version two so i want to tell you that we actually have met tremendous amount of progress and this um the version three will be available for the research community so that you can work on this and there will be a path for you to transfer to uh to operation and i want to emphasize that dtc uh the primary responsibility of dtc is not development our focus actually is to facilitate the research to operation and operation to research a transition and this is done through cooperation with many of our partners i guess this is over here and i made me visit a few and then uh we also be supporting the h5 project for the research to operation transition and for our testing we our goal here is to make sure that the code is robust and and that ncf can use it with confidence and confidence and in order for ends for dtc to be really be useful we also want to make sure that uh answer can rely on us for some pre uh to do the pre implementation testing and in order to for this to go well we need to have some protocol and our protocol is that we like to uh the the test the code in the in the community repository because if we otherwise we're gonna have too many versions and so we prefer to test those officially with these codes and uh in terms of testing here i want to we want to emphasize a few points one is that this is a functionality equivalent at the incept test environment and we also want to emphasize its end-to-end coupling rounds with the full operational cycling that like the nsf is doing and we also be using the same verification standard that the operational uh folks are doing at the at the national hurricane center at uh and at the amc and uh and so at this point i want to say that if you want to be participating in this testing we welcome very much on that and uh we'll be happy to work with you to incorporate your development into the world which repository so that your improvement can be then be evaluated for operation and also i want to you know that after this workshop we are going to have a three-day tutorial and with this we're going to release the h wolf code to the community and also and acknowledge that we're going to have many partners will be working with us in providing this community support that include mq and the version 3.2 maintenance and the release and the principle the princeton ocean model with the university of rhode island and volatile initializations with emc and copper with emc and the of rhode island and volt extractor from gfdl and so the the dtc will be working with mq so that the documentation website and this help desk and code management all available to support the broad community and now i'd like to take a uh just a few minutes to reflect on what we see at the future i want to say that the two point is that one is we need to make sure that dtc has a functionally equivalent environment with the emc so that we can do uh pre-implementation testing that emc can use it with confidence and we also want to work with h5 so that the problem promising new development can be incorporated and also another point i want to make is that there are several research projects been funded by h5 that are using arw and co-m's and we also have to have a way of making sure these developments can be also been available for operation and looking to the future i also want to say that even though our initial focus is on the each world we should not be limited by that i think we need to be more forward-looking for example have your approaches for for hurricane prediction by using the multiple model ensemble there are scenes that we can test we'll be very happy to do that we also repeatedly been finding that for example in some of common fields the data simulation seem to be always outperforming the 3d valves and these also something that we could also work with the emc to help make that available and another point is that as i mentioned if you want to make a seven-day forecast you got to be dealing with the global models and in thunder road the dcc should also be involved with the global model predictions and uh i remember somebody told me that every time you hear rick anthony's giving a talk he always uh showed two slides at least two slides one of the one is there's a little girl at the at the beach at the beach with a red dress and the other one is cosmic and he didn't have time to show any uh cosmic slides so i take the uh i'm obligated to show a few i want to say that in terms of that seven-day forecast and particularly for hurricane genesis uh you know one of the big problem we have is there isn't not very much observation over the ocean and with a radio quotation now we actually could have a lot of observation over the ocean and i'm not sure everybody know about this radio quotation but the whole idea here is if you put a gpu receiver on a small satellite then you can track the signals from the from the gps then you can actually have a radio quotation sounding at the point of rate we call the rate radio uh the parity point and you can actually get vertical profile electron density temperature and water vapor information from those measurements and the question of course are they any good are they how could they be helpful a good thing about this kind of observation is it's not limited by by land so you could actually have a globally homogeneous measurement and it's not affected by crowds and precipitations and shortly after the launch of cosmetic in 2006 we happened to run into a very interesting storm called hurricane and nastol and it shows that it's a category one storm and actually uh formed over the uh western atlantic and this shows the satellite images and if we make a forecast using the nsap operational analysis that time without any data simulation for cosmetic this is what we got there isn't that much of a development however if you assimilate the cosmic data into the n7 analysis this is what we got and clearly show that the cosmic does have improvement for the follicle prediction and further analysis we actually find the main reason is because the occultation can penetrate crowd and precipitation and provide tremendous amount useful information on water vapor and this is work by done by yongsan chen huaydio and jeff anderson and we also have a similar kind of walk on the western atlantic this is a xiaomi case for 2008 and we also have the same improvement that we find that if you incorporate the cosmic data you can do a genesis much better and if you do data simulation without cosmetic we find actually you got over development and i just want to finalize by showing you a few slides that this is the cosmic sounding we have in a day this is the cosmic sounding we will have in a week and no one now is working hard to implement a cosmetic two program and this is what we expect to see in a day so we should we have tremendous amount of opportunity to improve uh hurricane genesis and seven day forecast thank you very much