Visualizing surface change

Join along @boyd.xyz/c

Yesterday in the Great Smokies
Webcam archive

Exploring multi-temporal DSMs

Boyd Shearer

• Senior Lecturer, UKy Department of Geography.
• Online Digital Mapping program 🔗
• outrageGIS mapping topographic trail maps. 🔗
• Contact.

Explore heights
using KyFromAbove lidar point clouds

Height model

• Lidar point cloud derivatives.
• DSM from first-returns.
• DEM from KyFromAbove raster layer.
• height = DSM - DEM

Accurate?

• Walk around neighborhood with theodolite? On an iOS app?
• Drop a tape measure off of a bridge?
• Sorta...

15' 2" field height

15' 4" GIS tool

15' 3" DSM - DEM

Well within cartographic tolerances

First use of height model: visualizing floods

2022 floods

• Identify informal infrastructure prone to flooding.
• Notably, private bridges.
• Relative elevation model (REM) for height above river.

25 ft above Troublesome

• Subtract height from REM.
• Jupyter Notebook with ArcPy. 🔗
• GEO509 report. 🔗

Compare DSMs from
Phase 1 and Phase 2

Simple math

• change = DSMP2 - DSMP1
• but trickier symbology.
• Lexington 2010–2019 DSM change over 2019 shaded relief.

Record of change
lest we forget

Caveats

• If the replacement structure is same height, no change.
• No understanding of difference.
• Fuse other data sources?

Code

• Jupyter Notebook with ArcPy. 🔗
• Shows the process of calculating change between DSMs and colorizing point clouds.

```py # Smooth surface to remove artifacts along the edges of tall structures neighborhood = arcpy.sa.NbrRectangle(3, 3, 'CELL') change = arcpy.sa.FocalStatistics(change_uncleaned, neighborhood , 'MEDIAN') # Where has there been increase and decrease in height filled = arcpy.sa.Con((change > 2), change, 0) removed = arcpy.sa.Con((change < -2), change, 0) ```

Measuring tree heights

A tree canopy model

• Use NAIP NDVI to identify trees.
• Extract heights from DSM.
• Visually select NDVI values for trees.

Good enough for cartography?

Caveats

• Temporal resolutions don't match.
• NAIP's profound vertical displacement.

Opportunities

• Fuse NDVI with DSM change
• to find loss and gain of tree canopy.
• loss = (NDVIP1 > x) & (heightP2-P1 < y)

Accurate?

• Again, get into the field.
• But, what about the historical observations?
• Google Maps Street View? 🔗 🔗

Thank you