I have a dataset contains hourly PM10 concentrations for 90 days of 83 sites. That means I have total (90*24) set hourly data each contains 83 observations. I want plot semi-variogram for every hour (total 90*24 sem- variogram) then I want to compare variogram parameters (Range, psill, nugget) among all hours.
So, suppose now I have total (90*24) different dataset. among them, some data set are already in normal distribution. The dataset which are not in normal distribution, I want make them normal by log-transformation. Because, as far as I know, normal distribution is important for kriging analysis.
Naturally, the form of range,sill, nugget value obtained from actual dataset and log-transformed dataset are different. The range, sill, nugget value obtained from variogram analysis of log-transformed dataset are comparatively small as data were transformed into log. So, I need to back transform these range ,sill, nugget obtained from log-transformed dataset as I want to compare the range,sill,nugget value among all hours (90*24).
I am using gstat package in R for variogram analysis. How can I back transform the range,sill and nugget value obtained from log-transformed dataset?
[If you need, I can upload a sample dataset and my R code. please let me know if you have any other query]
Dear Uzzal Dash,
You can use Matheron (normal distribution) or Cressie (no normal distribution) to create your semivariogram. In this case, do you not need transform data.
Thanks Cláudio L. Bazzi. Did you mean that I can use 'classical method of moments variogram estimate' for variogram analysis with the dataset which are in normal distribution and I can use ' Cressie”s robust variogram estimate' for variogram analysis with the dataset which are not in normal distribution?
Dear Cláudio L. Bazzi , thanks a lot. It would be easier for me if it works. It would be great help for me if you could give me some reference/paper related to it.
Attached two references. One in Portuguese language and other in english.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package