How to fit Gamma Distribution if there are some negative values in the data?

I have a dataset of GCM parameter and I need to fit Gamma distribution to the dataset to do bias correction, But there are negative values also in the dataset due to which fitting Gamma is not possible directly. So how do i fit gamma distribution?

[Image Analyst]

Why would that be a problem for the fitting? How are you doing the fitting?

[Diljit Dutta]

Gamma distribution takes data values from 0- infinity and when i directly use gamfit function error message is displayed showing that X should be non negative. and by default Gamfit function uses MLE fitting procedure

[the cyclist]

Can you upload the data, or a representative sample?

What fitting function did you use? gamfit? (If you could upload your code, too, that would be ideal.)

Are your data just a tiny bit negative, on the order of round-off or floating-point error? If so, it will not introduce much error to make those zero instead. If they are larger, though, you may have a problem. You'll need to think more carefully about why you are getting negative values when they are "theoretically" impossible, given the distribution you believe they come from.