Hi,
I have daily gridded temperature dataset for 30 years and doing quality control of the data, I wanted to know which is the most appropriate statistical method for estimating outliers for daily temperature in any computational environment?
Not knowing the parameters of the output, I would start with anything outside two standard deviations on either side of the mean and follow up by adding back in the variables eliminated by this simplistic outlier test one at a time to determine when the trend observed after the first batch of removed outliers is found to change.
Thank you Douglas, but I think by using 2 standard deviations there will be a probability of loosing some of the peculiar observations as outliers.
Thanks Marcus, I will try to explore 6 sigma, it will be great if you can you send me a link where it has been used in climate data quality control.
Dear Saumya, using 95th percentile as higher limit and 5 percentile as lower limit you can easily compute the outliers falling above 95th and below 5th percentile.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers