GeoSpatial Tools and Analysis David Currie GeoSpatial Tools and Analysis Data for Good Sept 23, 2014 • Focus on open source tools • Bias towards remote sensing topics • Some examples from completed projects David Currie, P.Eng Open Source Geo Tools A good place to start: http://www.osgeo.org/ • GUI tools for GIS and Image Analysis • QGIS, OpenJUMP, Grass • Command line utilities • GDAL/OGR • GEOS • Proj4 • Ossim • Databases and Web Services • PostGIS/PostgreSQL • Mapserver, OpenLayers/GeoMoose, Leaflet 1
35
Embed
GeoSpatial Tools and Analysis - Meetupfiles.meetup.com/11057822/Geospatial-Tools-Analysis_David-Currie... · • Django/GeoDjango • Scriptingtools • R • Python QuantumGIS(QGIS)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
GeoSpatial Tools and AnalysisDavid Currie
GeoSpatial Tools and Analysis
Data for Good Sept 23, 2014
• Focus on open source tools• Bias towards remote sensing topics• Some examples from completed projects
• can also function as a tileserver• GeoDjango• Python based CMS with PostGIS in the background• Client-side tools (javascript)• Openlayers/GeoMoose• Leaflet
PostGIS
• Spatial engine running on PostgreSQL• handles vector and raster data• provides Proj4 and GEOS functionality for projections and topology
library("RPostgreSQL")v<-dbSendQuery(con,"select gid, length, width from feature_polygon where featuretype_id=22")exfeat<-fetch(v,n=-1)hist(exfeat$length,main="Length of Extreme Features",xlab="Length (m)",breaks="FD",col="lightblue")
Length of Extreme Features
Length (m)
Fre
quen
cy
0 10000 20000 30000
050
150
4
Scripting: R
• R-project http://www.r-project.org/• free and open source• powerful statistical capabilities• package system to link external capabilities, including GDAL, GEOS, GRASS, Raster, Spatial, SQL . . .
• free and open source https://www.python.org/• lightweight, useful for web servers• lower level programming interface than R, can give higher performance
The problem: capture dimensions of thousands of sea ice features from hundreds of images in order to quantify the variation in size by season andlocation.
The deliverable: Probability of exceedence statistics for each zone and season
9
Area 2 − Iceberg
Diameter (m) Note Log Scale
Exc
eede
nce
Pro
babi
lity
n:6470 m:0
10 100 1000
0.0
0.4
0.8
Web Data Capture
Tools
• PostGIS/PostgreSQL• Python with GDAL for data loading• Grass for image enhancement• Mapserver with GeoMoose for web based digitizing• Django for project administration and reporting• R and QGIS for generation of results
10
Web Data Capture
11
12
Web Data Capture
13
14
Web Data Capture
15
Web Data Capture
Results: - Selected 408 images from a set of 2073 (Landsat and Aster) - 45,578 icebergs detected and measured - 3,292 other significant featuresdetected - Multiple dates between 1999 - 2011 allowed movement to be detected and mapped
16
Project Examples: Raster Data Mining
17
The problem: Arctic polynyas are ephemeral features that are critical to wildlife and human survival. Where and when they occur is difficult to mapdue to access and weather conditions.
The deliverable: Spatio-temporal analysis of open water occurrence for period July 2002 - June 2013
Tools:- R with spatial and raster packages
18
Raster Data Mining
Starting from MODIS sea surface temperature grids. - 1 km resolution - Two grids produced per day
19
Raster Data Mining
• Day and Night grids give up to 6 observations of each square kilometer• Cloud and lack of sunlight affect available data
20
21
Raster Data Mining
• Over 110K images collected, registered, classified, and stacked
22
23
Raster Data Mining
• Although daily data is available, the coverage was not continuous• Daily data was stacked and filtered to generate a weekly time series• A set of open water composite maps were generated for each week of the 11 year span
24
25
Raster Data Mining
• Landsat imagery was used to tune the analysis and quality control the result.
26
Raster Data Mining
• Summary statistics from the resulting grids were extracted and analyzed• Open water fraction gives a normalized comparison
27
28
Raster Data Mining
• Polynyas are most easily recognizable in the early Spring• Comparing maps from specific week between years gives a good indication of variation
29
30
31
Raster Data Mining
• Specific locations were of interest, including the Last Ice Area