NCDC Snowfall Data - Interpretation?

Question

This is my first time using this dataset, so I have probably made an elementary mistake. I am having trouble understanding a file from the NCDC's climate normal dataset:

http://www1.ncdc.noaa.gov/pub/data/normals/1981-2010/products/precipitation/ann-snwd-avgnds-ge010wi.txt

Here is a sample of a few rows from the above file:

USC00064767 105P

USC00065077 31P

USC00065445 370S

USC00069388 66S

From the documentation:

ann-snwd-avgnds-ge010wi.txt Long-term averages of number of days during the year with snow depth >= 10 inches

From the readme and the above, I am assuming that the numerical portions of the entries in the second column in the above sample should all be numbers of days per year, so at most 367. However, some of the values listed above are greater than 367.

Any ideas for me, or other places to look? Did I miss any threads on this forum that discuss how idiots can understand NCDC data?

Answer 1

The readme file for all the datasets has a section on units. In this section, they describe how to interpret the numbers (which exclude decimal points in the data files).

For the "number of days" data it says:

[units are] "tenths of days" for the number of days per [time] above or below certain threshold, such as days above 90F. e.g., "256" is 25.6 days.

So, you need to divide all those numbers by 10 to get the actual values. Also, take note in the readme file of these pieces that correspond to the data file in question:

S = standard (no more than 5 years missing and no more than 3 consecutive 
    years missing among the sufficiently complete years)

P = provisional (at least 10 years used, but not sufficiently complete to be 
    labeled as standard or representative). Also used for parameter values on 
    February 29 as well as for interpolated daily precipitation, snowfall, and
    snow depth percentiles. 

-9999: missing or insufficient data; values cannot be computed
-7777: a non-zero value that would round to zero, for variables bound by zero.