Start by thinking of this as a search for errors in the data. Such errors may be in recording, or in the equipment. Start by looking for stupid things. Negative rainfall values, possibly negative temperatures (depending on where the data come from), and look for values that are too large or too small. Rainfall should have been 2.57 cm, but the decimal point is missing and 257 cm fell. If the data are all rainfall to the nearest hundredth and you find a value to the nearest thousandth, that might be a problem. If available, I would compare weather stations. If one station records 5 cm rain, and the station 1 km away reports cloudless skies then I might suspect that something is wrong.

Of course, most of the errors will be invisible if you only get yearly totals. It is much better if you can get the raw data as it was recorded, and make all the calculations yourself.

Now you have to decide what constitutes an outlier. Will you use raw values? Is this part of a regression, and you are looking at outliers in the residuals? Why did you choose 3 standard deviations from the mean as an outlier and not 2.8 or 4? Will you delete outliers? How do you justify this action? If I delete enough data, I can make the remaining values say whatever I want them to say.

There are several methods of detecting outliers. For example Dixon Method. You can use Xlstat  software for this purpose. It has few methods built in.

Thank you sir. Is 'Generalized Pareto Distribution' suitable for the same?

Dear Suman Kumar Kundu,

The following links will be very helpful to you.

http://people.eng.unimelb.edu.au/liuw/Webpage/wuliuchawla.pdf

http://download.springer.com/static/pdf/121/chp%253A10.1007%252F978-3-642-12519-5_7.pdf?originUrl=http%3A%2F%2Flink.springer.com%2Fchapter%2F10.1007%2F978-3-642-12519-5_7&token2=exp=1468910139~acl=%2Fstatic%2Fpdf%2F121%2Fchp%25253A10.1007%25252F978-3-642-12519-5_7.pdf%3ForiginUrl%3Dhttp%253A%252F%252Flink.springer.com%252Fchapter%252F10.1007%252F978-3-642-12519-5_7*~hmac=73ee76c3acbbe27342627516af81fab3a225f5748be43c1d9d0f3b52f44b734b

http://www.sciencedirect.com/science/article/pii/S0022169496032842

http://ac.els-cdn.com/S0020025506002891/1-s2.0-S0020025506002891-main.pdf?_tid=70021192-4d78-11e6-a8b2-00000aab0f6b&acdnat=1468909215_d4a82d61ce9b24058160cc92095269c7

http://users.auth.gr/~prinosp/downloads/Galiatsatou_PAPER_416.pdf

http://www.bepls.com/feb_2014/17.pdf

I hope this helps. Good Luck !