<style type='text/css'>.TerraceMsg { font-size: 12px; font-family:Dotum, Arial, Verdana, Sans-Serif;}.Bold { font-weight: bold; }</style><div class='TerraceMsg'><p>I have a dataframe contains hourly PM10 concentrations of 83 measurement sites. Please download from <a href="https://drive.google.com/file/d/0ByY3OAw62EShLTd5YkNvREJaSnM/view?usp=sharing" target="_blank">HERE</a>. Using this dataset, I have fitted a model with experimental variogram by ordinary kriging. In exploratory analysis, I found the data was not in normal distribution by checking skewness, histogram and Q-Q plot so I log-transformed the data. I wrote the following code:</p><p><br>library(sp)<br>library(gstat)<br>library(rgdal)<br>library(e1071) </p><p><br>seoul032813 <- read.csv("seoul032813.csv")<br>seoul032813<-na.omit(seoul032813)<br>summary(seoul032813$PM10)</p><p> </p><p>#exploratory analysis</p><p>skewness(seoul032813$PM10)<br>skewness(log10(seoul032813$PM10))</p><p>hist(seoul032813$PM10) </p><p>hist(log10(seoul032813$PM10))<br>qqnorm(seoul030101$PM10,font.main = 1,cex.main = 0.9, main = "Normal Q-Q plot of seoul032813",cex.lab = 0.75);qqline(seoul030101$PM10)<br>qqnorm(log(seoul030101$PM10),font.main = 1,cex.main = 0.9, main = "Normal Q-Q plot of seoul032813",cex.lab = 0.75);qqline(log(seoul030101$PM10)) </p><p> </p><p>#Reprojection<br>coordinates(seoul032813) <- ~LON+LAT<br>proj4string(seoul032813) = "+proj=longlat +datum=WGS84" <br>seoul032813<-spTransform(seoul032813, <br> CRS("+proj=utm +north +zone=52 +datum=WGS84"))</p><p><a href="mailto:seoul032813@coords" target="_blank">seoul032813@coords</a> <- <a href="mailto:seoul032813@coords/1000" target="_blank">seoul032813@coords/1000</a></p><p> </p><p>#variogram modeling<br>seoul032813_var<-variogram(log10(PM10) ~1, data=seoul032813)<br>seoul032813_var<br>plot(seoul032813_var, cex = 1.5, pch=
16, col = "black")<br>model032813<- fit.variogram(seoul032813_var,vgm(psill=300,model="Exp",range=10,nugget=50),fit.method =7)<br>model032813 #give the parameters<br>plot(seoul032813_var,model=model032813, col="black", pch=16,cex = 1.5,xlab = list(cex = 1,font = 1,label = "Distance (Km)"),<br> ylab = list(cex = 1,font = 1,label = ~ "Semivariance ( " *( mu * g / m^3)^2 * ")"), main = list(font = 1, cex = 1.25, label = "Semivariogram for 032813"))</p><p> </p><p>From this analysis I got the Range, psill and nugget values. These value are very small due to previous log-transformation of data. I want to get these parameter values in its original form. Also, I got a fitted variogram plot in the last line of my code where semivariance is in log form. I also want to get this plot in its original form. Simply, I want to back transform the variogram parameter values and fitted variogram plot. How can I do this?</p><p> </p><p>[Backgroud: Actually, I have many dataset of different hours. Some are in normal distribution and some are not. so, only the dataset which are not in normal distribution I log-transformed them. At last I want to compare Range, sill, nugget of every hour. So, for comparison I need to back transform the sill, range, nugget value obtained from log-transformed dataset]</p><p> </p><p>Orpheus </p></div><br><br/><br/><table border=0 cellpadding=0 cellspacing=0 ><tr><td style='vertical-align: top;font-size:9pt;'><div jquery1411706373835="566"><br></div></td></tr></table><br><img src="cid:856294410-12032001-0127027201690011" width='600px' border=0>
<div id='TMSMDN' style="background:url('http://mail.gist.ac.kr:80/mail/receiveMDN.do?mdnData=oGhEpHGdw7UD%2BvJWe%2BEHztreCrCu%2BoK90DwqM1K737iHmDXv%2FNKBfznD3EtDDByKdWQDpytGRk%2Bf%0AjVr%2BRPikPlm7nXAHhXHqnnOGIfpqH2lDoS37wCz%2BVkEiNnyHSHfxQ73EUHr1qksrCk8b8%2BLZqg%3D%3D%0A')"></div>