GFESuite Training - Smart Initialization Exercises </head> <body> <h1> Smart Initialization Exercises</h1> February 15, 2012<br> <hr style="width: 100%; height: 2px;"><br> <br> Performing these steps in an operational environment will mess up your guidance. You should only perform these on an non-operational setup. <p>If you run out of time, please be sure to <a href="#cleanup">clean up your system</a> for the next participant. </p> <p><a href="#Exercise1">Exercise 1. Override the NAM algorithm for Temperature</a> <br> <a href="#Exercise2">Exercise 2. Adding a new weather element and algorithm</a> <br> <a href="#cleanup">Clean up the system</a> <br> <a href="#Answers">Answers</a> <br> <br> </p> <hr width="100%"> <h2><a name="Exercise1"></a>Exercise 1. Override the NAM algorithm for Temperature</h2> The purpose of this exercise is to learn how to override existing algorithms. We will try three different examples. The first example will simply set the T to 32 degrees everywhere as a verification step to ensure that you have changed the algorithm correctly. <br> <h3><a name="1a"></a>1a) Set the NAM algorithm for Temperature to T=32</h3> Create a new file in /release/edex/data/utility/edex_static/site/SITE/smartinit called MyNAM12.py. This will be your override file. Perhaps the best way to create this file is to copy the original /release/edex/data/utility/edex_static/base/smartinit/NAM12.py into /release/edex/data/utility/edex_static/site/SITE/MyNAM12.py and then modify it. Strip out all of the functions in the class except the calcT(), since that is the function we will be overriding. Refer to the documentation on overriding a class -- you will need to make additional changes to the header of the file, the constructor, and the main declaration at the bottom of the class. <p>Change the calcT() function to simply set an entire grid to 32. Note you cannot simply set T=32, you need to create a numerical grid first of zeros, and then add 32 to it. </p> <p>Save your file. Run smart init by hand using the appropriate command, similar to this, from the bin directory: <br> <tt>ifpInit -t 20011117_1200 -a MyNAM</tt> </p> <p>Look at the NAM T grids on the GFE, did they get changed to the constant 32 value? </p> <p><a href="#answer1a">Answers for exercise 1a</a> <br> </p> <h3><a name="1b"></a>1b) Set the NAM12 algorithm to use the 2m Temperatures directly</h3> Modify your previously created MyNAM12.py. Change the algorithm to use the 2m temperatures (t_FHAG2). Don't forget that you want degrees Fahrenheit and the model has temperatures in Kelvin. <p>Save your file. Run smart init by hand. Verify that the grids did get changed. </p> <p><a href="#answer1b">Answers for exercise 1b</a> <br> </p> <h3><a name="1c"></a>1c) Correct your algorithm for model and real topographical differences</h3> Modify your previously created MyNAM12.py. Change the algorithm to use a fixed lapse rate based on the standard atmosphere. We will assume 6.0 degrees Celsius per kilometer. Your topo and stopo fields are in meters, the t_FHAG2 is in degrees Kelvin. Don't forget to convert the answer to Fahrenheit. <p>Save your file. Run smart init by hand. Verify that the grids did get changed. </p> <p><a href="#answer1c">Answers for exercise 1c</a> <br> </p> <h3><a name="1d"></a>1d)Putting your new algorithm into operation</h3> In this exercise, we will be putting your new algorithm into operation and verifying the correct steps have been made. Normally you will need to stop EDEX, edit the localConfig.py file, change the set of INITMODULES to use your new MyNAM12, then restart the server. Since you really wouldn't see any changes until the next model run, before you restart your server, you should manually delete all NAM12 databases under release/edex/data/hdf5/gfe/NAM12; then connect to the metadata database and issue SQL "delete from gfe where dbid like '%NAM12%';". <p>Start EDEX, bring up the GFE, call up the NAM12 T grids. When smart init runs your module, the T grids should appear. </p> <p>To clean up the system, stop EDEX, remove the localConfig.py INITMODULES entry, and remove your MyNAM12.py* files from the /release/edex/data/utility/edex_static/site/SITE/smartinit directory, purge the grids again from the release/edex/data/hdf5/gfe/NAM12 directory; then connect to the metadata database and issue SQL "delete from gfe where dbid like '%NAM12%';". Finally, restart EDEX. </p> <p><a href="#answer1d">Answers for exercise 1d</a> <br> <br> </p> <p></p> <hr width="100%"> <h2><a name="Exercise2"></a>Exercise 2) Adding a new weather element and algorithm</h2> In this exercise, we will be adding a new weather element to the NAM12 database, called DewPointDepression, and then defining a new algorithm to automatically create it. Since this is a new element, you first must create the localConfig.py entry to add the weather element. <br> <h3><a name="2a"></a>2a) Define your new weather element.</h3> <p><br> Add the DewPointDepression with the following characteristics to your localConfig.py: <br> DPD = ("TdD", SCALAR, "F", "Dew Point Depression", 50, 0, 0) </p> <p>Be sure to identify this as a new parm for the NAM12 model. Hint: parmsNAM12 is the variable name you need for your parm grouping. Set the time constraint to TC6, since the NAM12 comes in every 6 hours. </p> <p>Stop EDEX, then restart it. Bring up the GFE to verify that there is now a Dew Point Depression weather element in the NAM database.<a href="#answer2a"></a> </p> <p><a href="#answer2a">Answers for exercise 2a</a> <br> </p> <h3><a name="2b"></a>2b) Create your algorithm and test it.</h3> Create or modify the MyNAM12.py file in /release/edex/data/utility/edex_static/site/SITE/smartinit. You may leave in your calcT() from exercise #1 if desired. Add a new calc function for Dew Point Depression. The equation is basically T - Td. You can choose to calculate T-Td in one of two ways: <ul> <li>use the T and Td from the output IFP NAM12 database</li> <li>use the t_FHAG2 from the input D2D NAM12 database, and then calculate dew point from the t_FHAG2 and rh_FHAG2.</li> </ul> Save your file. Run smart init by hand using the appropriate command, similar to this, from the bin directory: <br> <tt>ifpInit -s xxx -t 20011117_1200 -a MyNAM12</tt> <p>Look at the NAM dew point depression grids, compare with the T and Td grids with the sample tool to make sure you have done the equation correctly. </p> <p><a href="#answer2b">Answers for exercise 2b</a> </p> <hr width="100%"> <h2><a name="cleanup"></a>Clean up your system</h2> To clean up the system, stop EDEX, remove the localConfig.py INITMODULES and dew point depression entries, and remove your MyNAM12.py* files from the /release/edex/data/utility/edex_static/site/SITE/smartinit directory, purge the grids again from the release/edex/data/hdf5/gfe/NAM12 directory; then connect to the metadata database and issue SQL "delete from gfe where dbid like '%NAM12%';". Restart EDEX. <br> <hr width="100%"> <h2><a name="Answers"></a>Answers</h2> <h3> <a name="answer1a"></a>Answer for Exercise 1a</h3> <tt>from NAM12 import *</tt> <p><tt>class MyNAM12Forecaster(NAM12Forecaster):</tt> <br> <tt> def __init__(self):</tt> <br> <tt> NAM12Forecaster.__init__(self)</tt> </p> <p><tt> def calcT(self, t_FHAG2):</tt> <br> <tt> t = self._empty + 32</tt> <br> <tt> return t</tt> <br> <tt></tt> <tt></tt> </p> <p><tt>def main():</tt> <br> <tt> MyNAM12Forecaster().run()</tt> </p> <h3><a name="answer1b"></a>Answer for Exercise 1b</h3> <p><br> <tt>from NAM12 import *</tt> </p> <p><tt>class MyNAM12Forecaster(NAM12Forecaster):</tt> <br> <tt> def __init__(self):</tt> <br> <tt> NAM12Forecaster.__init__(self)</tt> </p> <p><tt> def calcT(self, t_FHAG2):</tt> <br> <tt> return self.KtoF(t_FHAG2)</tt><tt></tt> </p> <p><tt>def main():</tt> <br> <tt> MyNAM12Forecaster().run()</tt> </p> <h3><a name="answer1c"></a>Answer for Exercise 1c</h3> <tt>from NAM12 import *</tt> <p><tt>class MyNAM12Forecaster(NAM12Forecaster):</tt> <br> <tt> def __init__(self):</tt> <br> <tt> NAM12Forecaster.__init__(self)</tt> </p> <p><tt> def calcT(self, t_FHAG2, topo, stopo):</tt> <br> <tt> elevDiff = topo - stopo</tt> <br> <tt> t = t_FHAG2 - (elevDiff * 6.0 / 1000)</tt> <br> <tt> return self.KtoF(t)</tt><tt></tt> </p> <p><tt>def main():</tt> <br> <tt> MyNAM12Forecaster().run()</tt> </p> <h3><a name="answer1d"></a>Answer for Exercise 1d</h3> Entry in localConfig.py: <p><tt>from serverConfig import *</tt> <br> <tt>import serverConfig</tt><tt></tt> </p> <p><tt>serverConfig.INITMODULES["MyNAM12"] = ["NAM12"]</tt> <br> <tt>del serverConfig.INITMODULES["NAM12"]</tt> <br> <br> </p> <h3><a name="answer2a"></a>Answer for Exercise 2a</h3> Entry in localConfig.py: <p><tt>from serverConfig import *</tt> <br> <tt>import serverConfig</tt><tt></tt> </p> <p><tt>DPD = ("TdDepression", SCALAR, "F", "Dew Point Depression", 50, 0, 0)</tt> <br> <tt>parmsNAM12 = [([DPD], TC6)]</tt> <br> <br> <br> </p> <h3><a name="answer2b"></a>Answer for Exercise 2b</h3> <tt>from NAM12 import *</tt> <p><tt>class MyNAM12Forecaster(NAM12Forecaster):</tt> <br> <tt> def __init__(self):</tt> <br> <tt> NAM12Forecaster.__init__(self)</tt> </p> <p><tt> def calcTdD(self, T, Td):</tt> <br> <tt> return T-Td</tt><tt></tt> </p> <p><tt>def main():</tt> <br> <tt> MyNAM12Forecaster().run()</tt> </p> <p>The alternative form, where you use the t_FHAG2 and rh_FHAG2 would be the following. Note that this method will not take advantage of any topographical corrections you have made in the temperatures:<tt></tt> </p> <p><tt>from NAM12 import *</tt> </p> <p><tt>class MyNAM12Forecaster(NAM12Forecaster):</tt> <br> <tt> def __init__(self):</tt> <br> <tt> NAM12Forecaster.__init__(self)</tt> </p> <p><tt> def calcTdD(self, t_FHAG2, rh_FHAG2):</tt> <br> <tt> td = getTd(t_FHAG2, rh_FHAG2)</tt> <br> <tt> depression = t_FHAG2 - td</tt> <br> <tt> return depression * 9.0 / 5.0</tt><tt></tt> </p> <p><tt> def getTd(self, t, rh):</tt> <br> <tt> # input/output in degrees K.</tt> <br> <tt> desat = clip(t, 0, 373.15)</tt> <br> <tt> desat = where(less(t, 173.15), 3.777647e-05, t)</tt> <br> <tt> desat = exp(26.660820 - desat * 0.0091379024 - 6106.3960 / desat)</tt> <br> <tt> desat = desat * rh / 100.0</tt> <br> <tt> desat = 26.66082 - log(desat)</tt> <br> <tt> td = (desat - sqrt(desat*desat-223.1986))/0.0182758048</tt> <br> <tt> td = where(greater(td, t), t, td)</tt> <br> <tt> return td</tt> <br> <br> </p> <p><tt>def main():</tt> <br> <tt> MyNAM12Forecaster().run()</tt> </p> <br> <br> </body> </html>