Upload
erick-mccoy
View
220
Download
3
Embed Size (px)
Citation preview
Spatial Interpolation (Kriging)
Objectives
• Understand the general procedures for spatial interpolation
• Explore the use of Kriging for spatial interpolation
General Procedures
• Examine the spatial continuity in sample data used to construct a semivariogram
• Choose a proper model to describe the observed spatial continuity
• Perform interpolation of values at un-visited locations
• Examine the quality of the interpolation
Review of Kriging
• Semi-variogram
γ(h) = 1/2 var [z(s+h) – z(s)]
= 1/2 E{[z(s+h) – z(s)]2} if E[z(s+h)] = E[z(s)] (ie, no trend)
hdji
ji
ij
szszhN )|,(
2))(2
• Ordinary Kriging– Assume:
• z(s) is intrinsically stationary, ie z(s) = m + e(s)• known semi-variogramγ• know values at z(s1) …… z(sn)
– Want to predicted z(s0) using a linear predictor
– With the objectives
Review of Kriging (cont.)
)]()(var[min
)]([
00
^
0
^
szszimum
mszEunbiased
n
iii szwz
10
^
)(*)(s
• Ordinary Kriging– With the objectives
)]()(var[min 00
^
szszimum
n
ii
n
iji
n
i
n
jjiji
wwsswsswwimize
i 110
2
1 1
22
,12)(2)(min
• Differentiating and set the partial derivatives to zero can get wi and
Review of Kriging (cont.)
i
iwmszEunbiased 1)]([ 0
^
Step 1: get data
• Get data– Start arc– Createworkspace workspace (workspace must not already exist)
• createworkspace d:\geog579\lab05
– Copy sample.e00 and test.e00 to your workspace using Windows copy and paste command
– Use the following command to import e00 files• Import cover sample sample• Import cover test test
– Show available coverage in the workspace• lc
– Use dir info to list Arc/Info files• dir info
– Use list to see the information in the files• list sample.pat
Step 2: Examine data
• Examine the data via– ArcMap
From ArcMap add data sample
– Or Arcplot
From arc, use the following commands
to plot the data• Display 9999• Arcplot (Launch Arcplot)• Clear• Mape sample• Points sample noids
Step 3: Examine spatial continuity in the sample data
• In ARC using the following statement to examine the spatial continuity of the sample data– Kriging sample s50_1 vars50_1 value # graph spherical sample 12 50
Step 3: Examine spatial continuity in the sample data (cont.)
• Use list s50_1.svg to examine the model fitted semivariance
• Examine the data using other semi-variogram models (exponential, gaussian, circular, and linear models)– Kriging sample s50_1 vars50_1 value # graph exponential sample 12 50
– Kriging sample s50_1 vars50_1 value # graph gaussian sample 12 50
– Kriging sample s50_1 vars50_1 value # graph circular sample 12 50
– Kriging sample s50_1 vars50_1 value # graph linear sample 12 50
Step 3: Examine spatial continuity in the sample data (cont.)
• Launch arcplot to display the semi-variogram– Clear– Semivariogram s50_1.svg– Use arcplot and the commands outlined in the handout to print/save the image
Step 4: Choose a proper model
• Choose a semi-variogram model based on the shape of the sample semi-variogram
Step 5: Perform interpolation
• Perform interpolation– For example:
• Kriging sample s50_1s vars50_1s value # lattice spherical sample 12 50
– Name conventions:• Differentiate: # of sample points, maximum radius, resolution, semivariogram
model
Step 5: Perform interpolation (cont.)
• Display the interpolation via– ArcMap
– Or Grid• Display 9999• Mape s50_1s• Gridpaint s50_1s # linear # gray
Step 6: Quality evaluation
• Use ArcMap or Arc/Info grid subsystem to print/save the interpolation variance map
Step 6: Quality evaluation (cont.)
• Compare RMSE– Copy rmse.eaf, edriveaml.bat, rmse.aml and aml.bat to your local storage– In Arcplot: use a locally develop program to report RMSE and ME
• &run rmse test value s50_1s
Step 7: Repeat
• Repeat– Change search radius from 50 20 (other parameter unchanged)
– Change spatial resolution from 1 3
– Change search sample points from 12 30
• Do comparison