Spatial Information Technology: Final Project

Goal: To build a model to predict light levels based off of Landsat data, and field data.

Base files used

>UW

FieldOBS.csv (converted to feature class through ARCmap XYFieldOBS.csv

Landsat data (sept022006utm.img)

Studyarea.shp

Base Setup

Basemap= Imagery with Labels

XYFieldOBS symbology changed to

Base image

>Spatial analyst tool>multivariate>Iso Cluster Unsupervised Classification=ISOcluster layer

Created an Cluster classification and then using reclassify reduced the classification from 13 to 10 categories that i felt represented the areas well.

Reclass images timeline

13->12->10

Using Sept0220065utm and Study area made a clip of the study area using image analysis toolbar and created an NDVI>clip_sept022006utm.img>NDVI_slip_sept02006utm.img>and exported the raster data (file: NDVI) which allowed me to create a dat table that I joined with the original XYfieldOBS to take NDVI data and compare it to the original data in excel. Light and NDVI were graphed using a scatter plot with a trendline

Data quite spread out from trendline note: r^2 = 0.10 and the trendline equation y 241.08x+8767.6

The equation allows an individual to predict light intensity using the equation above x=NDVI value and y is predicted light intensity with a low goodness to fit (R value) of .10 percent. I am no sure why it eded up in such a low R value maybe if I had relassified it more times it would have resluted in a higher goodness to fit.

The ovserved NDVI and Predicted (created using raster calc with the values (NDVI * 114.65) + 38957 ) shown below (file name P_lightraster)

Observed

Predicted

Using>Spatial analyst tool>Zonal> Zonal Histogram, and Zonal Statistics as table I was able to view that data

Next I wanted to combine other areas of study from the class so I added the Dem30_uw and using the spatial analyst tools>Surface>Hillsshade to give more depth to the map,