# How to do frequency analysis of an index when some years have zero value?

I want to do frequency analysis of drought severity, but unlike precipitation data here I have few zero values in some of the years. how to do this?

Here is the sample data:

1981    -1.34836065669339
1982    -1.35038460644290
1983    0
1984    -3.90332570519887
1985    -1.43196467810065
1986    0
1987    -7.61244713754898
1988    -1.02963066829013
1989    -2.87025610581835
1990    0
1991    -1.66903948845324
1992    -1.09433029589075
1993    -2.66739065155971
1994    -3.01088239849460
1995    0
1996    0
1997    -14.0056988537627
1998    0
1999    -9.85363948490749
2000    -1.36158063667645
2001    -5.34791858336523
2002    -3.11969637720832
2003    0
2004    -3.62559721791381
2005    -2.08214497384456
2006    -4.80944910848512
2007    -1.42804069956900
2008    -5.37009474560193
2009    -13.0916410702083
2010    -2.36624662191914

• Remove them and treat it as irregularly sampled data. There are ways to do this. – kwinkunks Nov 20 '15 at 13:10
• What do the zero values represent: no data or an actual zero value of drought severity? – Fred Nov 20 '15 at 13:26
• Could you assign them a near-zero value? – haresfur Nov 21 '15 at 2:28
• @haresfur No, I cant assign them near zero values, severity index starts from -0.99 onwards, what I can do is change the sign, that's it. Zero means normal condition, i.e., no drought. – DS Bisht Nov 21 '15 at 5:13
• Thinking about this some more, I'm not sure using this type of data is the right thing to do — I think it would be better to do your analysis on something more continuous. This dataset has essentially been thresholded on the drought condition. You might be better off using rainfall (say). What do others in your field do? – kwinkunks Nov 23 '15 at 2:42