USHCN Code Released

My USHCN code is available here as a zip file.

Once you unzip it you will see these files.

Main.cpp  Makefile  USHCN.cpp  USHCN.exe  USHCN.h

USHCN.exe is a Windows executable. Makefile is a Gnu makefile. If you aren’t a C++/STL programmer, you will probably have a difficult time understanding how it works.

In order to run it, you will need two files in the same directory as the executable.

http://cdiac.ornl.gov/ftp/ushcn_daily/ushcn-stations.txt

and any one of the the data files below. You will have to unzip the data file after you download it using gunzip.

Usage is : ushcn.exe USHCN_DATA_FILE_NAME [month=0-12] [max] [min]

Examples. Open up a command prompt window. Navigate to the directory with USHCN.exe, ushcn-stations.txt, and state41_TX.txt

ushcn.exe state41_TX.txt  > Texas.csv   – dumps out csv information for Texas to file Texas.csv

ushcn.exe state08_FL.txt  month=5 max > Florida_May_Max.csv   – dumps out csv information for Florida May maximum temperatures

ushcn.exe us.txt > US.csv – dumps out all US information

The code makes no attempt to do any adjustments or corrections. All it does is find all valid temperature recordings for all stations on all days, calculates averages – and finds min/max records per day of year for each station. There is no time bias. Earlier years are neither favored nor disadvantaged.

The reason I don’t do any corrections is because I am assuming a Monte Carlo distribution of error, which is the normal way that real scientists handle large data sets. There is no valid reason to believe that a huge data set created by thousands of observers over 130 years would have any systematic time-dependent bias to it – other than UHI. People read thermometers the same way in 1980 as they did in 1934.

If you mess the code up, take responsibility for it and don’t try to blame your mistakes on me.