| description |
PDXOSGeo connects again this week, Wednesday February 25, 6:30p Hot Lips Pizza @ the Natural Capital Center 721 NW 9th Ave, Portland OR. Join us for some show and tell and good open geo discussion!
We're striving to post our speaker a couple weeks in advance of the meeting, so watch this space! |
→ |
PDXOSGeo connects again this week, Wednesday February 25, 6:30p Hot Lips Pizza @ the Natural Capital Center 721 NW 9th Ave, Portland OR. Join us for some show and tell and good open geo discussion!
This month we will be discussing Landslide Susceptibility Modeling With Information Theory: The Terroir of Landslides, with David Percy
David Percy “Percy” has been doing research on applying Reconstructability Analysis (RA) to categorical raster GIS datasets since 2012, and is preparing to defend his dissertation on Spatial Reconstructability Analysis soon. RA is a form of machine learning algorithm (MLA) that works exclusively with discrete data, so categorical data such as the variables in the SSURGO soils database, the geologic database (DOGAMI), and land cover (NLCD) can be used in their native format without conversion to dummy variables, as is typically done in other MLAs.
These variables are used along with topographic data to model landslide susceptibility in two of the counties ranked highest by FEMA for most landslide hazards: Lincoln and Tillamook. We find that there is evidence to suggest that we can map a Terroir of Landslides, using combinations of soil type, geology, elevation,and terrane group, similar to the combinations of environmental factors that are used to characterize wines, chocolate, and coffee. Results will be presented that show how these data are useful in making hazard maps. All of the extraction for analysis was done in QGIS and Python using open source GIS libraries Shapely and Rasterio. Code will be shared and discussed.
Join us Wednesday for a lively discussion! |