Spatiotemporal Context in Analysis and Visualization of Movement - - PowerPoint PPT Presentation

Spatiotemporal Context in Analysis and Visualization of Movement

Somayeh Dodge Assistant Professor University of Colorado, Colorado Springs sdodge3@uccs.edu Glenn Xavier (UCCS) Maike Buchin (Ruhr-University Bochum) Sean C Ahearn (Hunter College - CUNY)

Gain insights into the behavior of dynamic objects and spatiotemporal processes

Image courtesy of Sebastian Cruz

Introduction | Context | Analytics | Visualization | Final Remarks

Movement and Context

Bird migrations in Movebank: Euroasia

https://www.youtube.com/watch?v=y4JJgyTncCA

1,654 individual birds, tracked between 1992 and 2012. The data represent 58 species, over 2 million locations, and 276,800 tracking days.

Introduction | Context | Analytics | Visualization | Final Remarks

What is happening and why?

This research is led by Prof. James L.D. Smith, Department of Fisheries, Wildlife and Conservation Biology, University of Minnesota Introduction | Context | Analytics | Visualization | Final Remarks

Somayeh Dodge, AAG 2015

What is happening and why?

This research is led by Prof. James L.D. Smith, Department of Fisheries, Wildlife and Conservation Biology, University of Minnesota Introduction | Context | Analytics | Visualization | Final Remarks

extracting behavior movement analytics

From Observation to Behavior

Introduction | Context | Analytics | Visualization | Final Remarks

Geographic and dynamic visualization movement

• bservation

relation to context

explore data & generate hypothesis visualize patterns explore relationships & correlations infer & interpret behavior

Spatiotemporal Context

๏ External factors that influence a dynamic process at a specific time scale ๏ Geography/Physiography ๏ Land cover ๏ Characteristics of terrain ๏ Network ๏ Obstacles (river, lake) ๏ Environment ๏ Ambient condition ๏ Weather ๏ Presence of other agents (interactions) ๏ Time ๏ Season ๏ Time of day ๏ Weekdays/weekends Introduction | Context | Analytics | Visualization | Final Remarks

Modeling Context

๏ Discrete ๏ Network ๏ Polygons ๏ Continuous ๏ raste grid Introduction | Context | Analytics | Visualization | Final Remarks

t

time

Somayeh Dodge, AAG 2015

Movement Track Annotation

Introduction | Context | Analytics | Visualization | Final Remarks

interpolation in space and time

te

t i m e

๏ Attributing context information to each tracking location (in space and

time) along a movement path.

๏ Using spatiotemporal interpolation

t1 t2

Movement Track Annotation

๏ Env-DATA: The Environmental-Data Automated Track

Annotation

๏ to link animal tracking data to a diverse range of context

variables

๏ to examine relationships between animal movement and

environmental conditions

๏ 16 large global datasets (~2500 variables): NASA, NOAA,

USGS, NCEP/NCAR and ECMWF weather reanalysis datasets

Introduction | Context | Analytics | Visualization | Final Remarks

Source: Dodge, et. al. (2013), Movement Ecology More info at: https://www.movebank.org/node/6607

Analytics Visualization

Using annotated trajectories.

Trajectory Similarity with Context

Introduction | Context | Analytics | Visualization | Final Remarks

Trajectory T = <(p1,t1),…,(pn,tn)> C = <c1,…,cn>

for matched points (p,t,c) and (p’,t’,c’)

dist(p,p’) + α dist(c,c’)

C1 C2 C2 C3 p p’

Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS ๏ integrate context & spatial similarity

Somayeh Dodge, AAG 2015

Trajectory Similarity with Context

๏ dissimilarity is computed as the

weighted sum of spatial and context distance


๏ spatial distance: ๏ context distance: ๏ context weight:

Introduction | Context | Analytics | Visualization | Final Remarks

for matched points (p,t,c) and (p’,t’,c

dist(p,p’) + α dist(c,c’)

Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS

C1

C2 C2 C3 p p’

for matched poin

dist(p,p’) +

nts (p,t,c) and (p’,t’,c

+ α dist(c,c’)

points (p,t,c

’) + α d Trajectory Similarity with Context

๏ dissimilarity is computed as the

weighted sum of spatial and context distance


๏ spatial distance: Fréchet distance

Introduction | Context | Analytics | Visualization | Final Remarks

Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS

Fréchet for matched poin

dist(p,p’) +

for matched points (p,t,c) and (p’,t’,c

dist(p,p’) + α dist(c,c’) Context Distance

water grass forest grass C2 C1 C4 C3 Distance matrix: dual graph C1 C2 C3 C4 C1 c1 c1 c3 C2 c1 c∗ c2 C3 c1 c∗ c2 C4 c3 c2 c2 where c1 = c(forest, grass), c2 = c(grass, water), c3 = c(forest, water) and c∗ = min(c1 + c3, c2 + c4)

Source: Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS ๏ Shortest path on graph ๏ unit cost ๏ cost depending on the labels

Galapagos Albatross (Phoebastria irrorata)

Image courtesy of Sebastian Cruz

Tracking data: ๏ 9 adult albatrosses ๏ Breeding season (Jun to Sep 2008) ๏ Temporal resolution 90 minutes

Introduction | Context | Analytics | Visualization | Final Remarks

Env-DATA Annotation

๏ Wind speed (m/s) and wind direction (degrees from North) ๏ Source: the NCEP Reanalysis 2 ๏ 6-hour, 2.5°, U/V-wind components ๏ http://www.esrl.noaa.gov/

Image courtesy of Sebastian Cruz

Introduction | Context | Analytics | Visualization | Final Remarks

Wind Flow Assistance

Longitude Latitude Duration (days)

Galapagos Island 20 40 60 80

• 90
• 88
• 86
• 84
• 82
• 80
• 78 0 -2 -4 -6
• 8
• 10
• 12

tail-wind (m/s)

• 5

5 10

Space-time path representation

2D path representation Source: Dodge, et. al. (2013), Movement Ecology

Peru Introduction | Context | Analytics | Visualization | Final Remarks

Track Similarity

Source: Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS

Track Similarity

Source: Buchin, M., Dodge, S., Speckmann, B. (2014). JOSIS

DYNAMO: Dynamic Visualization of Animal Movement and the Environment

Source: Xavier, Dodge (2014), MapInteract Proceedings

Introduction | Context | Analytics | Visualization | Final Remarks

Multivariate Visualization of Movement

๏ Visual variables ๏ line/point width, color ๏ vector size and direction

Source: Xavier, Dodge (2014), MapInteract Proceedings

Introduction | Context | Analytics | Visualization | Final Remarks

Method 1 (movement speed and wind speed) Dynamic Multivariate Visualization of Movement

Source: Dodge, et. al. (2013), Movement Ecology Xavier, Dodge (2014), MapInteract Proceedings Introduction | Context | Analytics | Visualization | Final Remarks

Method 2 (movement speed and wind direction/speed) Dynamic Multivariate Visualization of Movement

Source: Dodge, et. al. (2013), Movement Ecology Xavier, Dodge (2014), MapInteract Proceedings Introduction | Context | Analytics | Visualization | Final Remarks

Shadowing Effect (spatiotemporal footprint)

Introduction | Context | Analytics | Visualization | Final Remarks

The importance of context in movement research

๏ Extracting behavior from

• bservations

๏ Interpreting and

understanding movement patterns

๏ Study the impacts of

environmental change on behavior

Introduction | Context | Analytics | Visualization | Final Remarks Introduction | Context | Analytics | Visualization | Final Remarks

Bird migrations in Movebank: Euroasia https://www.youtube.com/watch?v=y4JJgyTncCA

Thank you!

๏ UCCS LAS Faculty-Student research award ๏ Tiger Research

๏

James L.D. Smith, (University of Minnesota)

๏

Achara Simcharoen (Conservation Ecology Program, King Mongkut’s University of Technology, Thailand )

๏ Movebank Env-DATA Project

๏

Gil Bohrer (Env-DATA Project PI, The Ohio State University)

๏

Max Planck Institute for Ornithology : Rolf Weinzierl (system admin), Sarah Davidson (data curator), Martin Wikelski (Movebank PI), Sebastian Cruz (Albatross data)