## Geostatistical Analyst

Atmospheric ozone monitoring for California. Measured at multiple stations representing overall whole area's concentration.

### Data Import

- Import all; though mostly be needed `OzoeLeevel_O3_Sep06_3pm`, and `ca_outline`
- `ca_hillshade` would help understand flat surface and where population mostly live
- Change `OzoeLeevel_O3_Sep06_3pm` symbology to: Quantities > Graduated colors; value = OZONE_LEVEL_VALUE (optional)

### Geostatistical Analyst

- Ensure `Geostatistical Analyst` extension is checked
- Customize > Toolbox > Geostatistical Analyst
- Geostatistical Analyst > Geostatistical Wizard
- Geostatistical methods > Kriging
	- Dataset > Source Dataset: OzoeLeevel_O3_Sep06_3pm
	- Date Field: OZONE
	- Next
	- Kringing Type: Ordinary
	- Output Surface Type: Prediction
	- Keep default
	- Next
	- Keep default for now
	- Next
	- Keep default
	- Next
	- Understand each tab
	- Finish
#### Fix Extends

- Rename produced "Kringing" to "Default Kringing"
- Open properties > Extend:
	- Set the extent to: rectangular extent of ca_outline
- Open "Layers" properties:
	- Data Frame > Clip Options:
		- Clip to shape
		- Specify Shape: select Outline of Features to ca_outline

#### Explore > Hostogram

- Geostatistical Analyst > Explore Data > Histogram
- Set attribute to "OZONE_LEVEL_VALUE"
- Values are multiplied by 10 (automatic)
- Select by draging into the plot; check map
- It can be realized that ozone level is high on where population mostly live (i.e., flat or low elevation area)

#### Explore > QQPlot

- Geostatistical Analyst > Explore Data > Normal QQPlot (Quantile Quantile Plot)
- Set attribute to "OZONE_LEVEL_VALUE"
- Understand that data is not normally distributed ($y = x$)

#### Explore > Trend

- Geostatistical Analyst > Explore Data > Trend Analysis
- Set attribute to "OZONE_LEVEL_VALUE"
- Curve representating trend

#### Explore > Semivariogram

- Geostatistical Analyst > Explore Data > Semivariogram/Covariance Cloud
- Set attribute to "OZONE_LEVEL_VALUE"

#### Mapping Ozone Conc.

- Geostatistical Analyst > Geostatistical Wizard > Kringing
	- Kringing Type: Ordinary
	- Set order of trend removal to second
	- Next
	- Keep default
	- Next:
		- Under View Settings, set search direction to true
		- May reduce lag size (e.g., to $15000$)
		- Under Model #1, set Anistropy to true
	- Next:
		- Under "Search Neighborhood", may change sector type, copy from variogram > angle (e.g., to 90)
		- Ellipse may be moved by clicking on points on the map, or
		- Under "Predicted Value", may change X (e.g., to 66000), Y (-220000)
	- Next:
		- Observe "Normal QQPlot"
	- Finish
- Fix extend

#### Prediction Standard Error

- Right click into "Kringing"
- Select "Change Output to Prediction Standard Error"
- Compare with <...>
- Revert

#### Validation Comparison

- Right click into "Kringing"
- Open "Compare"
- Compare with "Default Kringing"