Spatio-temporal asset management in a database of glacier change

Spatio-temporal asset management in a database of

glacier change

David Percy(1)Geospatial Data ManagerPSU Geology Department

Co-authors: Eric Hanson(2), Darrell Fuhriman, Cris Holm(2), William Garrick(2), Chris LeDoux(1), Matt Hoffman(1) and Andrew Fountain (1)(1)Dept of Geology, Portland State University(2)Academic and Research Computing, Portland State University

Funding by NSF, NASA, USGS and Portland State University

Three main points

• Hierarchical spatial structure: spatial ontology? Initially avoided due to complexity of relations, eventually imposed top-down

• RDF for database

• Temporal topology

Mt Hood, Oregon – Eliot Glacier

The challengesGlaciers may have many sources that describe their state.

Glaciers change over time and these various data sets need to be connected to the currently displayed outline.

Over 8000 polygons showing extent of perrenial ice and snow. No one wants to touch each of these to associate them with assets. Need a solution to link space and time with data base of assets.

Components of Open Source

Web Mapping

The new “LAMP”• L – Linux

• A – Apache

• M – MySQL

• P – PHP

• L – Linux

• A – Apache

• M – MapServer

• P – PostGIS

Metadata

Vector Data

Asset id (PK)

Asset type desc

Metadata id (FK)

Asset name

Shape

File name

File location

Source scale

Year start

Year end

Glacier databaseWeb version

Aggregated data

Glac_num

Centroid lat

Centroid long

Max elev

Min elev

Mean elev

Mean elev area weighted

Mean aspect

Region

State

HUC

Metadata id (FK)

File name

File location

Source scale

Cell size

Year start

Year end

Glac_num (PK)

Asset id (PK)

Glacier asset mm

Various

Asset id (PK)

Asset type desc

Metadata id (FK)

Asset name

File name

File location

Source scale

Cell size

Year start

Year end

Glacier – polygon

Moraine – polyline

HUC – polygon

GNIS – point

Oblique photo

Asset id (PK)

Asset type desc

Metadata id (FK)

Asset name

File name

File location

Year start

Year end

Aster

Asset id (PK)

Asset type desc

Metadata id (FK)

Asset name

File name

File location

Year start

Year end

Cell size

Quality

Img acquired month

Ref id

<fields>

References

Raster Data

DRG

DOQ

DEM

Glac_num

Centroid lat

Centroid long

Max elev

Min elev

Mean elev

Mean elev area weighted

Mean aspect

Region

State

HUC

Metadata id (FK)

File name

File location

Source scale

Cell size

Year start

Year end

GlacierAsset

Asset id (PK)

Region_id

Asset type desc

Metadata id (FK)

Asset name

File name

File location

Source scale

Cell size

Year start

Year end

Asset_Props

Prop_Id

Asset id (PK)

Name

Value

Regions

Region_Id

Geometry

Description

Region_level

Parent_id

Metadata

1946

1936

Temporal topology

1907Past

19241907

1924

1936

19561946

19861956

1936

Future

1924

User says "show me all glaciers from 1940"

By having stored the temporal topology, ie prior year date and

subsequent year date, we can find the closest object in the following three

ways:

Past only

( [Year] <= 1940) and ([Subsqunt_yr] > 1940 )

Closest Temporal buffer

(( [Year] + ([Subsqunt_yr] - [Year]/2)) >= 1940) and

(([Year] - ([Year] - [Prior_yr]/2)) <= 1940)

Gives the temporal mid point between two spatially referenced objects

Conservative Temporal buffer

(( [Year] + ([Subsqunt_yr] - [Year]/1.5)) >= 1940) and

(([Year] - ([Year] - [Prior_yr]/3)) <= 1940)

Gives you 2/3 past, 1/3 future

This can be parameterized!

Conclusions

• Hybrid RDF databases are viable, useful and in production NOW.

• The big question is how granular to get with RDF? For example individual vector data? Performance issues...

• Temporal topology may be the way to solve certain problems WRT spatio-temporal asset management

Future• WMS and WFS services to expose “stovepipe” to

outside access• http://geospatial.research.pdx.edu/cgi-bin/mapserv?map=/disk/mapserver/maps/percy/

glaciers_wms.map&service=wms&request=getcapabilities

• Google Earth?

• Query results dynamically to KML...

• Example:

http://geospatial.research.pdx.edu/~bjpd/usgs/ngm_quicksearch.html