Floating Sensor Networks for River Studies
Andrew Tinka, Student Member, IEEE, Mohammad Rafiee, and Alexandre M. Bayen, Member, IEEE
Abstract—Free-floating sensor packages that take local measurements
and track flows in water systems, known as drifters,
are a standard tool in oceanography, but are new to estuarial and
riverine studies. A system based on drifters for making estimates
on a hydrodynamic system requires the drifters themselves,
a communication network, and a method for integrating the
gathered data into an estimate of the state of the hydrodynamics.
This paper presents a complete drifter system and documents
a pilot experiment in a controlled channel. The utility of the
system for making measurements in unknown environments
is highlighted by a combined parameter estimation and data
assimilation algorithm using an extended Kalman filter. The
performance of the system is illustrated with field data collected
at the Hydraulic Engineering Research Unit, Stillwater, OK.
Index Terms—Data assimilation, hydrodynamics, Kalman
filters, sensor systems and applications.
I. Introduction
A. Freshwater Systems
The majority of the renewable freshwater available for
human use flows through rivers [1]. Human freshwater demand
will increase significantly in the next 50 years, due mainly to
population increase, urbanization, and increased use of waterintensive
agriculture [2]. Modeling and monitoring the flow of
freshwater, and the mixing and transport of constituents such
as salt, can lead to improvements in water use efficiency and
can help balance supply and demand [3]. Specific examples of
environmental management scenarios requiring understanding
of complex hydrodynamic systems include predicting the
movement of silt disturbed during dredging and underwater
construction operations, planning reservoir release and gate
control policies to affect the intrusion of salt water based on
specific local needs, and assessing vulnerabilities to contaminant
spills or other unforeseen events in critical water resource
regions. In each of these examples, high-quality hydrodynamic
models, based on data gathered from the actual system, can
be crucial for responsible environmental policy and decisionmaking.
Manuscript received February 21, 2011; revised September 14, 2011;
accepted December 2, 2011. Date of publication August 14, 2012; date of
current version February 20, 2013. This work was supported by NSF Awards
CNS-0615299, CNS-0915010, and NSF CAREER Award CNS-0845076. The
work of A. Tinka was supported by NSERC.
A. Tinka is with the Department of Electrical Engineering and Computer
Sciences, University of California, Berkeley, CA 94720 USA (e-mail:
tinka@berkeley.edu).
M. Rafiee is with the Department of Mechanical Engineering, University of
California, Berkeley, CA 94720 USA (e-mail: rafiee@berkeley.edu).
A. M. Bayen is with the Department of Electrical Engineering and Computer
Sciences and the Department of Civil and Environmental Engineering, University
of California, Berkeley, CA 94720 USA (e-mail: bayen@berkeley.edu).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JSYST.2012.2204914
B. Environmental and Mobile Sensing
The physical properties of large water systems can be
measured using several different sensor types and modalities.
Sensors are often categorized as Eulerian or Lagrangian (using
terminology from fluid mechanics) according to whether they
observe the medium as it flows past a fixed location (Eulerian)
or are embedded into the flow itself, measuring the
medium while moving along a trajectory (Lagrangian). The
canonical Lagrangian sensor is a small floating package that
transmits its location, and possibly other sensor measurements,
as it is carried by the water current through the system.
The oceanographic community calls such sensors drifters.
While most infrastructural sensing in rivers and estuaries is
implemented using Eulerian sensors, the evolution of wireless
sensor network technology has increased the interest in
novel Lagrangian sensor systems. The relative benefits of
Lagrangian sensors compared to Eulerian sensors can be classified
into two categories: logistical benefits and information
benefits.
The logistical benefits of a Lagrangian sensor system derive
from its flexibility and redeployable nature; in other words,
intrinsic benefits of self-contained devices designed for autonomous
operation. A fleet of drifters can be deployed, recovered,
and redeployed in response to changing needs or new
information. Their wireless communication allows them to be
used in remote locations where power and communication
infrastructure may not be available. These advantages are not
inherent to the Lagrangian or Eulerian distinction; it would be
possible to build an Eulerian sensor that was battery-powered,
communicated using wireless networks, and could be easily
redeployed. Rather, these logistical advantages are between
Lagrangian systems as they must be implemented compared
to Eulerian sensing as it is practised today.
The information benefits of mobile sensing, however, are
unique to the Lagrangian or Eulerian split. By following the
flow of water, Lagrangian sensors determine the particle outcomes
of water in the system. An Eulerian sensor, observing
the water as it flows past, can (normally) not infer anything
about the water’s history: where it came from, or where
it will end up. Tracking movement of water is particularly
important for studying the movement of contaminants or other
constituents, especially in regions with complex topology, such
as an estuary or a delta. Constituent transport is governed
by processes including advection and diffusion [4]; the Lagrangian
framework helps disambiguate the two, and allows
investigation into the precise location of interfaces or rapid
changes in concentration. One example of a hydrodynamic
phenomenon of interest where Lagrangian drifters are relevant
is tidal trapping, in which phase lags in tidal flow cause “dead
1932-8184/$31.00 c 2012 IEEE
TINKA et al.: FLOATING SENSOR NETWORKS FOR RIVER STUDIES 37
zones” where constituents can be trapped and released after a
delay [5], [6].
Lagrangian sensors do have some disadvantages in river
environments. Not all locations are suitable for deploying
drifting sensors. Rapids and waterfalls have the potential to
damage these devices. Rivers can contain obstacles that can
capture drifters. The drifters must be retrieved at the end of
a deployment, which can be a difficult procedure if they are
scattered over a wide area (or snagged on different obstacles
over a long stretch of river). The suitability of an environment
for drifter studies must be assessed prior to drifter deployment.
C. Data Assimilation
River hydraulics can be modeled with shallow water equations
in one or two dimensions [7]. Shallow water equations
are a standard constitutive model used in the environmental
engineering community and the hydraulics community to
model river flow; they are commonly used for simulation
and control. When dealing with experimental measurements,
algorithms are required to incorporate them into a model.
One such technique is data assimilation, which is the process
of integrating measurements into a flow model, and which
originated in meteorology and oceanography [8].
Most data assimilation methods can be placed into the
historically named categories of variational or sequential
assimilation methods [9]. Variational assimilation methods
perform a single optimization step on all the observed data
to minimize a cost functional. By contrast, sequential assimilation
methods, such as the Kalman filter and its extensions,
perform a series of update and analysis steps, blending the
observed data into the state estimate one step at a time. Several
extensions of the Kalman filter are applicable to nonlinear
systems. Examples include: the extended Kalman filter [10],
which uses the Jacobian of the state update equation to update
the estimate of the mean and covariance of the state, the
ensemble Kalman filter [11], which tracks the evolution of
a number of random samples in order to update the various
estimates, and the unscented Kalman filter [12], which also
tracks an ensemble of samples, but generates those samples
using a deterministic technique in order to accurately track
the mean and covariance with a minimal sample set.
This paper presents a data assimilation method based on
the extended Kalman filter. Sequential assimilation methods
are well suited to real-time assimilation, which is one of the
future goals for this system. The extended Kalman filter is
appropriate for nonlinear systems where the Jacobian is easy
to compute, which will be seen in Section III.
D. Drifters in Oceanography and Hydrology
Although studies of flotsam drift (drawing inferences about
currents from the observed movement of accidentally dropped
material) can be found in antiquity, the first deliberate drifter
study seems to be the work of G. Aim´e circa 1845 [13].
His first drifters were drift bottles: sealed bottles containing
a message asking the eventual recipient to report the date
and location found. Drift bottle studies became a widely used
technique in European oceanography around the beginning of
the 20th century [14].
The first drifter that could actively communicate its position
back to researchers was the “swallow float,” invented by
J. Swallow in 1955 [15]. It was a neutrally buoyant float that
would drift approximately 1000m underwater while transmitting
acoustic pulses that would be received by researchers’
hydrophones. Development of drifters with acoustic communication
capabilities continued in the 1960s and 1970s [16]. In
1978, the introduction of the Argos satellite service [17] gave
oceanographic researchers a global location and data uplink
system, which lead to the development of oceanographic
drifters that could communicate the