Significance value of climate data trends using cdo

Question

I'm performing trend analysis on certain parameters like LAI,NDVI etc. I use CDO-trend operator for the same. But CDO doesn't give the significance value. Is there any other operator in CDO or way to get trend of any variable along with the significance value?

score 2 · Answer 1 · edited May 09 '18 at 11:10

2

You can use NCO to access GSL's gsl_fit_linear() function, then follow these steps to compute the p value.

edited May 09 '18 at 11:10

Bart

9,825
5
47
73

answered May 09 '18 at 10:20

Charlie Zender

5,929
14
19

score 0 · Answer 2 · answered May 09 '18 at 01:56

cdo appears to perform a simple linear regression against time. Statistical significance of the trend slope would be highly affected by seasonality and any other serial correlation in the data. The seasonal Mann-Kendall test is designed to cope with this. (See the Kendall R package). Another approach might be to model the time series using ARIMA coefficients along with trend coefficient.

Significance value of climate data trends using cdo

2 Answers2