Identification of Sustainable Locations in Pigeon Flights

using Flow Simulation Method

Margarita Zaleshina

and Alexander Zaleshin

Moscow Institute of Physics and Technology, Moscow, Russia

Keywords: Visual Perception, Spatial Navigation, Flow Simulation.

Abstract: Navigation behaviour in nature is based on data obtained from perception of the terrain where movement

occurs. The aim of this work is to study the influence of visual factors on the flight of birds over medium

distances (about 10 km). In this study, we propose a method for probabilistic analysis of pigeon flights over

combined countryside and urban terrain, based on surface flow simulation. Z-value – an altitude analogue

that describes the characteristic gradient of the flow – is calculated as a function of "landscape complexity"

based on the density of significant landscape objects. The calculated probabilistic model is compared with

data on GPS-tracks of untrained and trained pigeons. As a result, significant features of terrain that

determine sustainable locations in pigeon flights are identified. In the study, visual characteristics of the

territories over which pigeons flew are calculated using remote sensing data from open sources, and spatial

data are processed using the geographical information system QGIS.

1 INTRODUCTION

Here we studied the properties of pigeon flight

trajectories over combined countryside and urban

terrain. The aim of this study is to identify the

interdependence of the characteristics of the

trajectories of pigeons and the visual properties of

the landscape, based on surface topology.

Study of the typical ways in which pigeons

respond to changes in the landscape over which they

fly shows that pigeons rely on visual perception of

the terrain to determine their routes. Their

perception of the terrain allows them to distinguish

characteristic features that are suitable for guiding

flight above terrain. These features are determined

by such parameters as tone, colour, and density of

detached objects.

Moreover, the way in which a pigeon orients

itself based on visual data is directly influenced by

the degree to which the landscape is filled with

separate stimuli. Mann et al. (2014) studied the

influence of “landscape complexity” on pigeon

navigational behaviour. The authors concluded that

pigeons orient themselves better when flying above

territory where “landscape complexity” is neither

too high nor too low.

To assess the probabilistic characteristics of a

flight over surface, the visual perceptibility factor is

used here, depending on saturation of the terrain

with visual objects. It can be represented as z-value,

an analogue of height, which describes the

characteristic directions of flight. To find potential

flight trajectories, the flow paths over the surface

caused z-value differences are calculated. The

computational model is compared with pigeon GPS-

track data over this area.

Visual features in the landscape that are

important for long-range navigation can be

identified by Kano and colleagues (Kano et al.,

2018). These authors discuss the training of a

particular route by repeated flights on the same

terrain.

The properties of trained and untrained birds can

be reflected in perception preferences for objects on

the ground and in flight paths, which is also shown

in this work.

This paper is structured as follows. In Section 2,

we provide a brief review of the following topics: i)

visual perception of the terrain during flight; ii)

formation of flight trajectories based on visual data

perceived by a pigeon; iii) analysis of surface

properties to determine the possibility of external

dynamic processes; and iv) representing the density

of objects as a surface on which flow can potentially

Zaleshina, M. and Zaleshin, A.

Identiﬁcation of Sustainable Locations in Pigeon Flights using Flow Simulation Method.

DOI: 10.5220/0009000305350541

In Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods (ICPRAM 2020), pages 535-541

ISBN: 978-989-758-397-1; ISSN: 2184-4313

535

occur. In Section 3, we describe data and data

processing methods for calculations used for surface

flow simulation. In Section 4, we compare data on

calculating flows over a landscape and real flight

paths for untrained and weakly trained pigeons. In

Section 5 we discuss the applicability of

identification of sustainable locations in pigeon

flights using flow simulation method.

2 BACKGROUND AND RELATED

WORKS

2.1 Visual Perception of the Terrain

during Flight

Navigation mental maps can be formed by animals

based on the results of perception of the terrain.

Birds can orient themselves based on both

“landscape complexity” and individual reference

objects. In general, during medium-distance flights,

bird trajectories are determined by visual perception

of the terrain and reflect the visible characteristics of

the surface.

Figure 1 shows adjustment of pigeon trajectories

caused by perception of the terrain.

Figure 1: The consecutive places of "attraction" caused by

terrain perception: (A) Satellite view, (B) map view, (C)

pigeon flight near places of "attraction".

The navigational behaviour of pigeons in

different situations is described in many

publications. Mann et al. (2014) showed that pigeons

can make use of some form of navigational map,

combined with time-compensated solar compass

information, to orient homeward from distant

unfamiliar places. Blaser et al. (2013) showed that

birds knew their geographical position in relation to

targets, and chose a flight direction according to

their needs – clearly the essence of a cognitive

navigational map. Also, in paper (Blaser et al., 2013)

the authors proved that pigeons are able to

remember routes and fly to the objects which are

important to them, such as home or feeding spots,

and that they can also choose where to turn

depending on the degree of their satiety.

The significance of visual stimuli is different for

investigatory flights, when a pigeon surveys

unknown terrain, than for purposeful flights, when a

pigeon flies along the known path to the known

goal. In an unknown place untrained pigeons first try

to explore the area, and perform survey flights in

different directions, while trained pigeons head

straight to the goal (Blaser et al., 2013; Wiltschko

and Wiltschko, 2015; Pettit et al., 2012).

The path of the investigatory survey flight may

cover a large territory. It is during the survey flight

that a pigeon actively reacts to the terrain features in

a way that is noticeable based on its flight. A trained

pigeon mostly flies almost directly to the goal with

insignificant deviations from a set route. Biro et al.

(2007) state that when orienting itself while flying

above known terrain, a pigeon may combine

purposeful movement in a chosen direction with

landmark guidance. Even over previously unfamiliar

terrain pigeons demonstrate fairly stable sets of

behaviour. For example, they prefer not to fly over a

wide water surface. At the same time, pigeons have

a tendency to use linear structures, such as roads,

rivers, or boundaries between dissimilar surfaces

(Kano et al., 2018; Lipp et al., 2004; Vyssotski et al.,

2009).

To simulate the flight of untrained and trained

pigeons, the following assumptions can be adopted:

 When choosing a route, untrained pigeons

are guided by the visual perceptibility of

the terrain; their flight route is directed

from the places that are least visually

attractive to more attractive places.

 Trained pigeons are guided by routes that

they learned earlier, and to which they try

to return when they have deviated from

their accustomed route. As a result, they

may stop responding to many visual

stimuli, except for the main reference

points.

2.2 Formation of Pigeons’ Routes

based on Visual Data Perception

In unfamiliar conditions, birds look for objects that

they have previously encountered, stable options for

the location of these objects, and the usual sequence

in which these objects appear as a basis for

recognition of other information.

A bird's behaviour over previously familiar

terrain changes during flights. Thus, repeated

viewing of the same point of the terrain (or localized

site) forms changes in behaviour over this point.

ICPRAM 2020 - 9th International Conference on Pattern Recognition Applications and Methods

536

Visually, terrain is perceived from different

heights and viewing angles, at different scales, and

with different degrees of detail. Thus, when flying

over a forest, only the boundary line of this forest

can be tracked, but smaller details – such as

commonly viewed margins and clearings – can also

be observed, although transitions between them are

less noticeable and significant.

A bird flying along a certain trajectory forms a

panoramic perception of its environment,

complementing, if necessary, the general visual

scene with elements that are not always observable

at some specific instant. At the same time, it is

possible to fully observe the borders and extended

areal interrelations of the terrain elements, including

those outside the current viewing angle of the bird.

Moreover, in the process of panoramic viewing

during flight, such extended areal interrelations can

be established or not established for a short time.

In addition, with repeated observation of the

same mixed and erratic data, birds can form unstable

interrelations, while maintaining a stable

composition of the data. This differs from rare

observations of the same mixed and erratic data,

when perception of different data occurs separately

and the interrelations are not formed.

Depending on the scale and the flight route,

selection of individual terrain elements – which

differ from each other in some ways – may not

occur. Similarly, selection of boundaries and

extended area objects may not occur (Figure 2).

Figure 2: Adjustment of the pigeon flight route, taking into

account perception of elements of the terrain that are not

visually connected and preference for flying over linear

objects. (A) schematic view of water objects, (B) satellite

view.

2.3 Analysis of Surface Properties to

Determine the Possibility of

External Dynamic Processes

The properties of a surface affect the dynamics of

external processes that occur over it. Thus,

landscape have a significant impact on urban

development: interpretation of time series processing

is represented in urban growth monitoring (Sexton et

al., 2013). Fluid dynamics simulation methods are

also applied in biomedical analysis (Ferrari et al.,

2018; Rispoli et al., 2015). Shape of a river bank,

and formation of its bends, inflow and inner islands

are determined by relief over which it flows.

Interpretation of complex flow patterns is

represented in geosciences (Gallien et al., 2018;

Graser et al., 2019) and geological survey (Essaid,

1990; Essaid et al., 2015) in generalized analysis of

topography induced stream subsurface exchange

(Stonedahl et al., 2010) and in considering ground

water and surface water as a single resource (Winter

et al., 1998).

Probable pigeon flight routes can be calculated

using surface flow simulation, on different parts of

the routes. To accomplish this, the isolines with the

same “landscape complexity” are used to construct

typical cases of pigeon “flow”. In particular, surface

flow can be constructed for places with a sharp

change in “landscape complexity” and/or for places

with dense accumulations of isolines. Such a

method, in general, resembles typical methods for

calculating for water stream on hilly terrain.

Formation of the flow path, taking into account

the steep slope, is limited on the sides in watershed.

(Figure 3). Similarly, the location of arc-shaped

thresholds is formed where pigeon route cross the

texture border isolines. When flying over texture

boundaries, perception of a pigeon can change

abruptly. Differences in density of perceived

textures can either be presented in the form of a

“sharp change” or be smoothed. A considerable

height differential, as a rule, is localized in a small

area.

Figure 3: Formation of the flow path in watershed.

Identiﬁcation of Sustainable Locations in Pigeon Flights using Flow Simulation Method

537

2.4 Representing the Density of Objects

as a Surface on Which Flow Can

Potentially Occur

Computational methods for modelling flows can be

applied in cases where there is a certain “height”

analogue (z-value). Density of stand-alone

distinguishable objects, variation in density and

direction makes it possible to create a flow model

for the terrain, where the density of objects is used to

determine the height, taking into account the

perceptibility of the terrain. Surface characteristics

which are attributed to all units of the surface,

providing sets of direction gradients "top-down".

Based on the fact that pigeons prefer a certain

density of separately standing distinguishable

objects for flight (Mann et al., 2014), it is possible to

snap such points by lines in cases where the points

are close to each other and the density drop is

insignificant.

Figure 4 shows an example of representing the

density of objects as a surface on which flow can

potentially occur: layer contains “voronoi polygons”

corresponding to input data about density.

Figure 4: Flow path calculation using data on the density

of objects as a “height” analogue. Red arrows indicate

direction of gradient, leading to flow formation (green

line).

3 MATERIALS AND METHODS

3.1 Materials

The calculated flow paths over the surface can be

compared with observed flight paths of pigeons. In

this work, data on the flights of pigeons and remote

sensing data for terrain over which these flights took

place were used as primary source materials. Data

packages with GPS-tracks of pigeons were collected

from Dryad Digital Repository

(https://datadryad.org, dataset (Pettit et al., 2012),

publication (Pettit et al., 2012)) in the form of CSV

files.

The pigeons flew over two types of

heterogeneous terrains: over countryside terrain

covered with forests and fields (site 1, where the

distance between start and finish points was 11.5

km) and over urban terrain with buildings and roads

(site 2, where the distance between start and finish

points was 12.5 km). Measurements of coordinates

between individual points of GPS-tracks were taken

one time per second. The characteristic distance

between separate coordinate values of pigeons’

GPS-tracks is in the range of 20-40 meters. The

areas at a distance of 150 meters from flying up and

landing of pigeons were not considered (in these

areas paths crossed themselves more often and

movement direction was constantly changing).

The study was performed for flights of untrained

pigeons (the first flights over the previously

unknown terrain) and weakly trained pigeons (the

second and third flights over the previously

unknown terrain): 21 flights over site 1, 27 flights

over site 2.

Remote sensing data (satellite images) in the

form of OpenLayers (http:// openlayers.org) was

used for ground surface information about the

surface of terrain. The coordinate system for the

project was WGS 84/Pseudo-Mercator

(EPSG:3857).

3.2 Methods

Z-value and flight characteristics calculation was

accomplished for the untrained and trained pigeons.

Calculations were performed in the following steps:

 Creation of primary data layers:

- Point and line vector layers with pigeon

flight data based on GPS-tracks.

- Raster layers with satellite image materials

for the area of pigeon flights.

 Mapping “landscape complexity”

“Landscape complexity” map was built based on

the remote sensing data. Visual features of the

landscape (“landscape complexity”) were identified

according to the density of visually observed

objects. Firstly, boundaries of individual

homogeneous surfaces were identified by

constructing isolines. A density map of the existing

terrain inhomogeneities was then constructed based

on the resulting clusters of isolines for characteristic

inhomogeneity dimensions of 50 and more meters.

The specified accuracy for the inhomogeneity map

corresponds to the typical distance between two

neighbouring points of the GPS pigeon tracks.

ICPRAM 2020 - 9th International Conference on Pattern Recognition Applications and Methods

538

Figure 5: Construction of the terrain inhomogeneity map:

(A) the analyzed area (site1); (B) “landscape complexity”

map; the range shows the density of perceived elements in

the territory.

Figure 5 shows an example of the analysed area

(a) and the corresponding “landscape complexity”

map in the form of a “heat map” (b).

 Z-value definition and flow mapping using

z-value

Z-value and flight characteristics were calculated

for pigeons. Based on the “landscape complexity”

map, z-value was calculated (as Invert Grid) for all

parts of the terrain; this is the equivalent of the

height used in relief maps. Z-value determines flow

direction for pigeon flight (“top-down”) in the

absence of additional stimuli. Flows were calculated

on the basis of the constructed heat map, using the

plugin GRASS r.watershed (Figure 6).

Thereafter, base centroids of calculated flow

paths points were constructed, and then applied to

analyse the results. This calculation takes into

account the fact that the distance between

neighbouring z-value isolines is significantly less

than distance between start and finish points (about

1%).

Figure 6: The calculated flows in the z-value map, built

from data on "landscape of difficulty".

3.3 QGIS Plugins

The data were processed using the open source

software program QGIS (http://qgis.org), including

additional analysis plugins: QGIS geoalgorithms and

GDAL tools (http://www.gdal.org) integrated into

QGIS. The source data layers were added using the

OpenLayers Plugin in QGIS, which allows to obtain

Google Maps, Bing Maps and another open layers.

In addition, analogues of "hydrological

parameters" were calculated in our model and a set

of maps was formed indicating the accumulation of

runoff towards the most saturated complex

landscape.

The applied QGIS tools and plugins are

presented in Table 1.

Table 1: Applied QGIS tools and external plugins.

Plugin Description

OpenLayers Plugin:

https://github.com/sourcepol

e/qgis-openlayers-plugin

QGIS plugin embeds

OpenLayers

(http://openlayers.org)

functionality.

It allows to obtain Google

Maps, Bing Maps,

OpenStreetMap and anothe

open source layers.

Points2One:

http://plugins.qgis.org/plugin

s/points2one

Create lines and polygons

from vertices. Connects

oints in a layer to form lines

and polygons.

Heatmap Plugin:

http://www.qgistutorials.com

/en/docs/creating_heatmaps.h

tml

Create a density raster of an

input point vector laye

ased on the number o

oints in a location, with

larger numbers of clustered

oints resulting in large

values.

GRASS r.watershed

https://grass.osgeo.org/grass7

6/manuals/r.watershed.html

Calculates hydrological

arameters and generates a

set of maps indicating flow

accumulation, drainage

direction, the location o

streams and watershe

basins.

SAGA Invert Grid

http://sagagis.org/saga_tool_

doc/7.1.1/grid_tools_34.html

Inverts a grid, i.e. the highest

value becomes the lowest

and vice versa.

4 RESULTS

4.1 Calculated Flow Paths and Real

Flight Routes for Pigeons

We compared the flow paths calculated based on the

surface flow simulation and the real GPS-trajectories

of untrained pigeons, using QGIS.

After simulation was completed, calculated flow

paths and real flight routes were compared using

pigeon GPS-tracks.

Identiﬁcation of Sustainable Locations in Pigeon Flights using Flow Simulation Method

539

As the result of calculations, it was obtained that

for the countryside terrain (site 1), pigeons prefer to

fly over the calculated flow paths with a probability

of 16% greater than to not fly (p-value < .05). For

the urban area (site 2), there is no significant

dependence of flight over the calculated flow paths.

The samples of calculated flow paths and real

trajectories are shown in Figure 7.

Figure 7: The flow paths calculated based on the surface

flow simulation and real flight routes of pigeons. Yellow

dots shows calculated flow paths, blue and magenta lines

shows real pigeons’ flight routes. A heat map with z-value

is shown as an additional layer (the legend for z-value is

shown in the upper left corner). Red arrows indicate places

of “adjustment” in the flight routes of pigeons, depending

on the surface properties.

Based on comparison of untrained and weakly

trained pigeons’ flights, it is apparent that training

leads to cases of reducing in the perceptibility of

terrain parameters (Figure 8).

Figure 8: Difference between the real route parameters for

one pigeon in first, second and third flights. Magenta line

– Flight1, green line – Flight2, blue line - Flight3, yellow

dots – centroids of calculated flow paths.

It was obtained that pigeons prefer to fly over

places with noticeable extended variations on

countryside landscape, with a probability of 16%

greater than to fly not over them.

Table 2 shows the ratio of the length of the flight

path within the flow area to the total length of the

flight path for untrained (1st flight) and weakly

trained (2nd and 3rd flights) pigeons.

Table 2: Comparison of results for untrained and weakly

trained pigeons.

Pigeons Untrained

1st flight

Weakly

trained

2nd flight

Weakly

trained

3rd flight

Pigeon 1 15,70% 29,80% 28,00%

Pigeon 2 37,30% 17,60% 19,70%

Pigeon 3 20,30% 32,20% 37,00%

Pigeon 4 39,00% 36,50% 32,50%

Pigeon 5 36,60% 33,00% 30,30%

Pigeon 6 30,40% 20,20% 18,00%

Pigeon 7 37,00% 25,80% 27,60%

The results of the comparison of calculated flow

paths and real flight routes were different for

untrained and weakly trained pigeons.

5 CONCLUSIONS

When pigeons fly across medium distances (about

10 km), visual features of the surface significantly

affect the probability of pigeon flights over this area.

The paper explored the flights of different

pigeons over mixed terrain. The trajectories of flight

of untrained and weakly trained pigeons are guided

not only by reference points or extended landmarks,

but also by the general structure of the terrain. After

perception, terrain’s surface determines the

sequential flow forms and influences the choice of

the direction of movement.

The influence of external visual information

causes birds to change their trajectories, which are

partially "attracted" not only to specific points of

interest, but also to areas with the most saturated

landscape.

In the present work it is shown that the attention

of untrained pigeons was most often diverted, they

deviated from the usual path selection algorithm,

ICPRAM 2020 - 9th International Conference on Pattern Recognition Applications and Methods

540

and began to shift to the areas of "saturated

landscape complexity".

The results of this work can be used to

understand the selection algorithms for the

navigational behavior of birds, other animals, or

humans.

REFERENCES

Biro, D. et al. (2007) ‘Pigeons combine compass and

landmark guidance in familiar route navigation’,

Proceedings of the National Academy of Sciences,

104(18), pp. 7471–7476. doi:

10.1073/pnas.0701575104.

Blaser, N. et al. (2013) ‘Testing cognitive navigation in

unknown territories: homing pigeons choose different

targets’, Journal of Experimental Biology, 216(16),

pp. 3123–3131. doi: 10.1242/jeb.083246.

Essaid, H. I. (1990) ‘A multilayered sharp interface model

of coupled freshwater and saltwater flow in coastal

systems: Model development and application’, Water

Resources Research. doi:

10.1029/WR026i007p01431.

Essaid, H. I., Bekins, B. A. and Cozzarelli, I. M. (2015)

‘Organic contaminant transport and fate in the

subsurface: Evolution of knowledge and

understanding’, Water Resources Research. doi:

10.1002/2015WR017121.

Ferrari, S. et al. (2018) ‘The Ring Vortex: A Candidate for

a Liquid-Based Complex Flow Phantom for Medical

Imaging’, in Tavares, J. M. R. S. and Natal Jorge, R.

M. (eds) VipIMAGE 2017. Cham: Springer

International Publishing, pp. 893–902. doi:

10.1007/978-3-319-68195-5_97.

Gallien, T. et al. (2018) ‘Coastal Flood Modeling

Challenges in Defended Urban Backshores’,

Geosciences, 8(12), p. 450. doi:

10.3390/geosciences8120450.

Graser, A. et al. (2019) ‘Untangling origin-destination

flows in geographic information systems’, Information

Visualization, 18(1), pp. 153–172. doi:

10.1177/1473871617738122.

Kano, F. et al. (2018) ‘Head-mounted sensors reveal

visual attention of free-flying homing pigeons’, The

Journal of Experimental Biology, 221(17), p.

jeb183475. doi: 10.1242/jeb.183475.

Lipp, H.-P. et al. (2004) ‘Pigeon Homing along Highways

and Exits’, Current Biology. England, 14(14), pp.

1239–1249. doi: 10.1016/j.cub.2004.07.024.

Mann, R. P. et al. (2014) ‘Landscape complexity

influences route-memory formation in navigating

pigeons’, Biology Letters, 10(1), pp. 20130885–

20130885. doi: 10.1098/rsbl.2013.0885.

Pettit, B et al. (2012) ‘Data from: Not just passengers:

Pigeons, Columba livia, can learn homing routes while

flying with a more experienced conspecific’,

Proceedings of the Royal Society B. Dryad Digital

Repository. doi: 10.5061/dryad.53f4b.

Pettit, Benjamin et al. (2012) ‘Not just passengers:

pigeons, Columba livia, can learn homing routes while

flying with a more experienced conspecific’,

Proceedings of the Royal Society B: Biological

Sciences, 280(1750), pp. 20122160–20122160. doi:

10.1098/rspb.2012.2160.

Rispoli, V. C. et al. (2015) ‘Computational fluid dynamics

simulations of blood flow regularized by 3D phase

contrast MRI.’, Biomedical engineering online.

BioMed Central, 14(1), p. 110. doi: 10.1186/s12938-

015-0104-7.

Sexton, J. O. et al. (2013) ‘Urban growth of the

Washington, D.C.–Baltimore, MD metropolitan region

from 1984 to 2010 by annual, Landsat-based estimates

of impervious cover’, Remote Sensing of Environment,

129, pp. 42–53. doi: 10.1016/j.rse.2012.10.025.

Stonedahl, S. H. et al. (2010) ‘A multiscale model for

integrating hyporheic exchange from ripples to

meanders’, Water Resources Research, 46(12). doi:

10.1029/2009WR008865.

Vyssotski, A. L. et al. (2009) ‘EEG Responses to Visual

Landmarks in Flying Pigeons’, Current Biology.

Elsevier Ltd, 19(14), pp. 1159–1166. doi:

10.1016/j.cub.2009.05.070.

Wiltschko, R. and Wiltschko, W. (2015) ‘Avian

Navigation: A Combination of Innate and Learned

Mechanisms’, in Advances in the Study of Behavior.

Elsevier Ltd, pp. 229–310. doi:

10.1016/bs.asb.2014.12.002.

Winter, T. C. et al. (1998) ‘Ground Water and Surface

Water - A single Resource - U.S. Geological Survey

Circular 1139’, USGS Publications. doi:

10.3133/CIR1139.

Identiﬁcation of Sustainable Locations in Pigeon Flights using Flow Simulation Method

541