RAIN: Rainfall & Flood Prediction Project
Based on the success of the undergraduate projects in developing prototype rainfall sensors, a multidisciplinary team of Rice faculty developed a plan for networking such sensors and using the data for flood prediction and warning.
Team Members
James Young: Professor in the Department of Electrical and
Computer Engineering, with research experience in lasers, optical devices,
optical communications, and fiber optics.
Ashutosh Sabharwal: Faculty Fellow in the
Department of Electrical and Computer Engineering and also a Research
Director of the Center for Multimedia Communications (CMC). His research
focus is access protocols for wireless networks and information
theoretical aspects of wireless communications.
Phil Bedient: The Herman Brown Professor of Engineering in the
Department of Civil and Environmental Engineering. His major research
interests surface water problems including major floodplain studies,
hydrologic modeling for flood control, water quality studies, and GIS
linkages to hydrology.
J. Patrick Frantz: Executive and Technical Director of CMC and a
Lecturer in the Department of Electrical and Computer Engineering. His
interests are building wireless communication system testbeds, DSP-
based hardware systems, mobile wireless devices and communications
algorithm implementations.
Primary Project Tasks
- Development of sensor nodes
We will develop optical rainfall sensors using the temporal covariance of the signals from two separated sensors. This function measures the velocity of raindrops in the beam, and thus their size). This task also includes the engineering of the basic node platform: control, communication, and signal processing electronics, mechanics, and renewable power systems.
- Network and communications
An optically-connected network offers special advantages and challenges. Laser communication exploits the xtremely low power consumption of optical detectors in standby mode, and the directivity eliminates several bandwidth hungry protocol layers. We will develop design rules for adaptive joint sensing-communication links that achieve optimum tradeoff between sensing accuracy and communication capacity-latency, subject to the available power and bandwidth. We will design a new load-balancing protocol between sensing and communication to maximize the network data throughput. We will develop network routing schemes to ensure robust network performance even with node or link failures. Finally, we will implement, test, and characterize these methods.
- Deployment and testing
We will develop and deploy our proof-of-concept testbed in Harris Gully, Houston, a flood prone area containing Rice University and the Texas Medical Center (TMC). The TMC suffered flood damage exceeding \$1.5~Billion in 2001 due to Tropical Storm Allison. We also need to deploy nodes at distant sites, both to test the system and to collect spatially-diverse meteorological data. We propose to locate some of these remote sensors at Houston-area high schools, and to actively involve teachers and students in the program. This portion of the program, the RAIN Teacher Education Connection (RAINTEC), will center around an intensive professional development workshop for teachers, to create a curriculum that will involve researchers, teachers, and students in collaborative problem-solving and learning. The goals of this task, in addition to testing all aspects of the network operation, include determining the time-sampling necessary to characterize weather events: how often the environmental parameters need to be measured. Matching the sampling rate to the need will minimize power consumption.
- Data collection, integration, and applications
The RAIN Project will include a central computer resource to collect (using the Internet) and process the physical data, and deliver it effectively to users. The data will be incorporated into several unique applications. Hydrological models rely on spatially and temporally detailed rainfall data to predict flooding. RAIN data will be used to develop improved physics-based distributed models that combine digital terrain data obtained from LIDAR with the high resolution RAIN data. A distributed flow analysis can predict local flooding at areas of critical importance using RAIN sensors precisely where more coverage, accuracy, and resolution are needed. The current flood alert systems rely on weather radar estimates of rainfall distribution. For additional information on current systems, see Flood Alert System(approve sites) web page. RAIN sensors will be used to provide critical measurements at ground level.
