Training activities on ranges can adversely impact the environment. Vehicle traffic on unpaved roads and tank trails causes extensive erosion resulting in reduced water quality because of increased sediment loads. Currently, the Army does not have a way to continuously and directly monitor suspended solids concentration (SSC) in streams. In this demonstration project, twelve sensors designed to simultaneously measure SSC and flow velocity were deployed at three military installations—Fort Riley, Kansas; Fort Benning, Georgia; and Aberdeen Proving Ground (APG), Maryland—through a three-tier wireless sensor network (WSN) to realize remote, Internet-based, continuous, long-term monitoring of sediment loads. The objectives were to improve the sensor and the WSN; validate the accuracy, repeatability, and operability of the sensor in measuring SSC and flow velocity; and validate the functionality of the WSN.

The technology consists of two parts: (1) an optical sensor that continuously measures SSC in streams using visible and infrared (IR) lights and flow velocity using the cross-correlation method; and (2) a three-tier WSN to remotely transmit the SSC data from the sensor site(s) to the Internet. For the demonstration, twelve SSC/velocity sensors and a three-tier WSN were deployed at Fort Riley, Fort Benning, and APG.

Performance objectives of this project included accuracy and repeatability of the sensor in measuring sediment concentration and flow velocity; the operability of the sensor and its lens cleaning mechanism; and reliability of the components of a three-tier WSN examined over a long demonstration period.

The performance objectives for SSC measurement accuracy, repeatability, and operability were generally not achieved. The objective for SSC measurement accuracy was set to be ±10% or ±50 milligrams per liter (mg/L) of actual SSC, whichever is greater. Within the 95% confidence interval, the highest prediction error for the validation data set was found to be -46.2% for SSCs larger than 500 mg/L, and 290.9 mg/L for SSCs lower than 500 mg/L. The objectives for repeatability and operability were set to be one half of that for accuracy. The actually achieved repeatability was lower than 12.9% for SSCs larger than 500 mg/L, and lower than 292 mg/L for SSCs lower than 500 mg/L. Data post-processing applied to 6 months of SSC data showed that the actually achieved operability was 23.1% for SSCs greater than 500 mg/L, and 234.5 mg/L for SSCs lower than 500 mg/L.

The performance objectives for flow velocity measurement were generally not achieved. The objective for measurement accuracy was set to be ±10% or ±0.01 meters per second (m/s) of actual flow velocity, whichever is greater. The objective for repeatability was set to be one half of that for accuracy. The error found in velocity measurement was less than 27.8%. The repeatability for velocity measurement was lower than 0.37 m/s.

The performance objectives for various tiers and components of the three-tier WSN were generally not achieved. The objectives for percentage of normal operation (PNO) and data loss rates (DLR) for individual components in the network were set to be 90% and 0.5%, respectively. The lowest PNO and highest DLR recorded were 55% and 8.83%, respectively.

During the demonstration, one of the velocity sensors was deployed at a U.S. Geological Survey (USGS) stream-gaging station in Pine Knot Creek at Fort Benning. Through continuous velocity measurement over a 1-year period, the measured point velocities were used to generate an index-rating curve that can be used to estimate the mean velocity from measured point velocity. The stage measurement provided by USGS and the estimated mean velocity were then used to estimate discharge using the “index-velocity method.” This experiment demonstrated the possibility of using both stage and point velocity measurements to provide better discharge estimation.

Throughout the demonstration, SSC sensor calibration was found to be the most difficult issue for implementation. A two-stage procedure was used for the calibration. The second stage of this procedure requires grab samples. In order to allow the sensor to measure SSC accurately within a wide range, a large amount of water samples with SSCs distributing within the desirable range need to be collected at the sensor site. This requires water samples to be taken during various rain events. The cost related to labor and transportation is very large.

In order to alleviate this concern, an alternative approach for the second stage of sensor calibration was developed. The approach used a field sampler that continuously took water samples at various sediment concentrations and completes the sampling process within one to two hours.

Other implementation issues are related to deployment of the sensors in natural waters, including streams, lakes, rivers, and reservoirs. Securing the sensor in the water is always a challenge, especially during the high-flow season. Adding mechanical reinforcement usually alleviates the problem. However, for streams with sand/stone bottoms this may become extremely difficult.

The size of the stream needs to be considered when deploying the sensor. The general recommendation is that the sensor be deployed near a bank, perhaps within a distance of 20 feet. The maximum measurable SSC is 5000 mg/L, and the velocity sensor has a maximum measurable velocity of 5 m/s. These limits should not be exceeded.