How fast is a cow? Cross‐Scale Analysis of Movement Data

Abstract

Data representing the trajectories of moving point objects are becoming increasingly ubiquitous in GIScience, and are the focus of much methodological research aimed at extracting patterns and meaning describing the underlying phenomena. However, current research within GIScience in this area has largely ignored issues related to scale and granularity – in other words how much are the patterns that we see a function of the size of the looking glass that we apply? In this article we investigate the implications of varying the temporal scale at which three movement parameters, speed, sinuosity and turning angle are derived, and explore the relationship between this temporal scale and uncertainty in the individual data points making up a trajectory. A very rich dataset, representing the movement of 10 cows over some two days every 0.25 s is investigated. Our cross‐scale analysis shows firstly, that movement parameters for all 10 cows are broadly similar over a range of scales when the data are segmented to remove quasi‐static subtrajectories. However, by exploring realistic values of GPS uncertainty using Monte Carlo Simulation, it becomes apparent that fine scale measurement of all movement parameters is masked by uncertainties, and that we can only make meaningful statements about movement when we take these uncertainties into account.

The effect of sampling rate on observed statistics in a correlated random walk

“…For simplicity, we neglect the effects of measurement noise; while the effects of noise and sampling frequency are related [4], many open questions remain about the role of sampling frequency alone. We discuss this in further detail in §6.…”

“…While the details of the experimental protocol vary between different studies, in the majority of cases the data obtained are similar, namely an ordered list of position vectors (observations) sampled at discrete time points that approximates the continuous-time underlying motion. The time intervals between observations may have a profound impact on the movement patterns observed and the conclusions drawn [3][4][5]. It is therefore crucial to characterize these effects.…”

Understanding movement data quality

“…Generally, derived thematic attributes depend on the sampling rate in the data, i.e., temporal frequency of recorded positions (Laube and Purves 2011). Thus, due to the triangle inequality, a decrease of the sampling rate causes a decrease of the computed distance between position records and, further on, a decrease of the speed.…”

What is an Appropriate Temporal Sampling Rate to Record Floating Car Data with a GPS?

“…On a first glance, this conclusion contradicts with results obtained by other authors. In [45], a Monte Carlo simulation is used to illustrate that measurement error in trajectories sampled at high frequencies does not allow to calculate realistic movement parameters. However, this simulation assumes that GPS measurement error scatters entirely randomly between each two consecutive position estimates.…”