See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/7104711

Environmental monitoring by wireless

Article in Science · June 2006

DOI: 10.1126/science.1120034 · Source: PubMed

CITATIONS READS
211 975

3 authors:

Hagit Messer Artem Zinevich

Tel Aviv University Lab126
239 PUBLICATIONS 2,952 CITATIONS 10 PUBLICATIONS 572 CITATIONS

SEE PROFILE SEE PROFILE

Pinhas Alpert
Tel Aviv University
353 PUBLICATIONS 10,210 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Urban monitoring View project

cellular monitoring & remote sensing of urban climate View project

All content following this page was uploaded by Pinhas Alpert on 26 May 2014.

The user has requested enhancement of the downloaded file.

 BREVIA
Given measurements of the received signal
level (RSL), we estimated the rain-induced
attenuation A, and then the average rainfall rate
Environmental Monitoring by Wireless over each time frame (4).
Based on the data collected every 15 min
Communication Networks from a few DFRS cellular backhaul links during
a rain event in Israel in January 2005, we have
estimated local rainfall. Figure 1 compares the
Hagit Messer,1* Artem Zinevich,2 Pinhas Alpert3 rainfall estimates from the cellular backhauls,
from radar, and from rain gauges. The skill of
igh-resolution, continuous, accurate from a cellular network, and show its im- our method (correlation with rain gauges) is

H monitoring of the environment is of great

importance for many applications—
from weather forecasting to pollution regulation.
proved accuracy compared with radar-based
estimates.
There are a variety of wireless communica-
0.86 for a 15-min-interval rain intensity and 0.9
for an hourly interval, versus 0.81 and 0.85,
respectively, for radar, when evaluated from the
We propose to use measurements from existing tion systems, and we focus on the digital fixed maximal value over a 3  7 km2 area. How-
wireless communication networks for environ- radio systems (DFRS) that have a number of ever, the corresponding correlation values from
mental studies, complementing existing moni- useful properties: They work at up to a few tens the literature at 3-km gauge-separation distance
toring systems such as weather radars. Weather, of meters above the ground, they operate at with radar are 0.59 and 0.71, respectively (5).
atmospheric conditions, and constituents cause frequencies of tens of GHz, and they are The density of the DFRS cellular backhaul
propagation impairments on radio links. As impaired mainly by near-surface precipitation. links, according to the data from one Israeli
such, similar to global positioning systems DFRS have already established themselves as cellular provider, varies in average from 3 links
(GPS) (1), cellular networks provide built-in the premier cellular backhaul technology in per km2 to 0.3 links per km2 in urban areas, and
monitoring facilities and can be considered as a Europe and Asia, accounting for the majority of is less in suburban and rural areas. That allows,
widely distributed, high-resolution atmospheric cellular base-station connections. for example, the creation of rainfall intensity
observation network, operating in real time The years of research by telecommunication maps at an average spatial resolution of 1 to 3
with minimum supervision and without addi- specialists (3) generated tools for modeling and km, using tomographic reconstruction.
tional cost. interpretation of atmosphere-induced impair- Our results suggest that cellular rainfall
Meteorological monitoring of rainfall by ments on radio links. The rain attenuation de- measurements have features in between those
radar is less accurate at surface levels (2), pends on the size and distribution of the water of gauges and of radar. Cellular measurement
whereas rain gauges, although quite accurate, droplets. There are several models relating the can either replace existing techniques or
are expensive and do not provide sufficient attenuation rate A (given in dB/km) with the colaborate with them to achieve better per-
spatial resolution. We demonstrate the feasibil- rain intensity. The common approach is a formance. However, the potential of the cel-
ity of environmental monitoring with wireless power law model for the attenuation, A 0 aRb, lular environmental monitoring is not limited
communication networks by estimating the sur- where R is the rain rate and the constants a and to rainfall measurements. Solid particles, fog,
face rainfall using standard data collected b are functions of frequency and polarization. snow, sleet, and hail can be detected by mi-
crowaves. In particular, the capabilities of
cellular networks (measuring of the refraction
index of the atmosphere) may provide valu-
able facilities for studying water vapor, which
plays a key role in weather and the global
climate system.

References and Notes

1. M. Bevis et al., J. Geophys. Res. 97, 15787 (1992).
2. S. Raghavan, Radar Meteorology (Kluwer, London, 2003).
3. ITU-R P. 530-7 (International Telecommunication Union
Radiocommunication Bureau Propagation Recommendation).
4. Materials and methods are available as supporting
material on Science Online.
5. E. Morin, J. Hydrometeorology 4, 782 (2003).
6. We deeply thank S. Mouallem and H. Pistiner (Cellcom)
for their cooperation and for providing the cellular data.
We also thank Y. Levy (Shacham/Mekorot) for radar data
and A. Arie (Meteo-Tech), A. Stupp (Tel-Aviv University),
and H. Kutiel (Haifa University) for rain gauge data.
Supporting Online Material
www.sciencemag.org/cgi/content/full/312/5774/713/DC1
Materials and Methods
Table S1
Fig. 1. Comparison of the time series of rainfall intensity measured by cellular links, rain gauges, and References
Movie S1
a weather radar, in two areas in Israel: (A) Tel-Aviv and (B) Haifa. The rainfall event was observed on
19 to 20 January 2005. The location of the radar is given in Movie S1. The rain gauges work at 12 September 2005; accepted 27 March 2006
10.1126/science.1120034
temporal resolutions of 30 min (A) and 10 min (B), whereas the wireless links provide measurements
every 15 min. Temporal lags between the cellular data and the rain gauges are partly due to
differences in locations of the links and the rain gauges (they are separated in space by about 2 km). 1
School of Electrical Engineering, 2The Porter School of En-
Disparities, such as time lags, are also caused by the different nature of observations, i.e., line- vironmental Studies, 3Department of Geophysics and Plane-
integrated data in the cellular links versus point measurements in the rain gauges. tary Sciences, Tel-Aviv University, 69978 Tel-Aviv, Israel.

www.sciencemag.org SCIENCE VOL 312 5 MAY 2006 713

View publication stats