No Waiting, No Angry

export C_INCLUDE_PATH=/usr/local/Cellar/lzo/2.10/include/lzo:/usr/local/Cellar/lzo/2.10/include export LIBRARY_PATH=/usr/local/Cellar/lzo/2.10/lib pip install python-lzo

make install

make install

Now you can use git command which will ask you username and password only once.

This error may be related to some InfiniBand shortcomings. Set the following environmental variables in your profile and use intelmpi/2017.1.132, then the problem will never appear again.

gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -sOutputFile=output.pdf 1.pdf 2.pdf ... n.pdf

GSI is data assimilation system for NWP. Many use it as a black box, but some tips you need to know: 1. set the number of vertical levels in anavinfo to those in wrfinput. 2. set iuse column in convinfo an satinfo. 3. for cycle, GSI can ingest wrfout or the file generated by write_input=.true., but caution with hydrometers and soil state.

For performance, I usually call Fortran in python. And f2py is an excellent tool to convert Fortran subroutine to a shared library which can be import into python. Sometimes Fortran subroutine call external libraray, the external library can be linked by directly putting it in python command line or -Lpath -llibrary.

In meteorology, positive u component points to east and positive v component points to north. The wind directon is defined as 0=north, 90=east, 180=south, and 270=west. dir=270-atan2(v,u)*180/pi spd=sqrt(u*u+v*v) u=-spd*sin(dir*pi/180) v=-spd*cos(dir*pi/180)

In one of my program, i use netCDF4 library to create netcdf file with large dimension variables, the program become slow after wrote out a few variables. I can't figure out where is the problem. Finally, I used pupynere the problem was gone.

show global variables like 'max_allowed_packet'; set global max_allowed_packet=1024*1024*64;

db = MySQLdb.connect(‘localhost.localdomain’,'root’,“,'test’)   cursor = db.cursor()   cursor.execute(“SELECT VERSION()”)   data = cursor.fetchone()   print 'Database version : %s' % data   db.close()  

then you can connect to mariadb by “mysql -uroot”

when using kriging, please make sure that the scattered observations are unique, that is, their locations cannot be duplicated. Another thing is to make sure the observed and predicted coordinates are the same order, that is, don't make such stupid mistake, for example, use longitude/latitude for observations, but use latitude/longitude for predictions. I have spent a couple of day to debug these carelessness.

Adding the third part toolbox to matlab is simple, first extract the toolbox into $MATLAB/toolbox directory, then add that path in $MATLAB/toolbox/local/pathdef .m. After that start matlab, enter rehash toolbox cache, then the toolbok can be used as the built-in toolbox.

In matlab, rows vary first then columns for a matrix. So the function image or imagesc, plot the matrix one column after another column, that is, its pixels start from upper left conner, that is (1,1) is corresponding to upper left point, go down from top to bottom, then go right. But the function contour or contourf starts from lower left point, that is (1,1) is corresponding to lower left point, go up to top, then go right. It is important to understand this fact so that we can master matlab plotting.

providing radar and stationary satellite give high frequent data, ensemble assimilation and cloud model with topography may be a good way for nowcasting. I will try it.