Development and evaluation of an open-source, low-cost distributed sensor network for environmental monitoring applications

Abstract

Over the last decade there has been a proliferation of low-cost sensor networks that enable highly distributed sensor deployments in environmental applications. The technology is easily accessible and rapidly advancing due to the use of open-source microcontrollers. While this trend is extremely exciting, and the technology provides unprecedented spatial coverage, these sensors and associated microcontroller systems have not been well evaluated in the literature. Given the large number of new deployments and proposed research efforts using these technologies, it is necessary to quantify the overall instrument and microcontroller performance for specific applications. In this paper, an Arduino-based weather station system is presented in detail. These low-cost energy-budget measurement stations, or LEMS, have now been deployed for continuous measurements as part of several different field campaigns, which are described herein. The LEMS are low-cost, flexible, and simple to maintain. In addition to presenting the technical details of the LEMS, its errors are quantified in laboratory and field settings. A simple artificial neural network-based radiation-error correction scheme is also presented. Finally, challenges and possible improvements to microcontroller-based atmospheric sensing systems are discussed.

Performance of the ATMOS41 All-in-One Weather Station for Weather Monitoring

“…With the increasing use of wireless sensor networks [ 11 ], various non-standard low-cost weather monitoring systems have been developed in the past few years using a wide range of sensor hardware and different microcontroller architectures, such as Arduino [ 12 , 13 , 14 ] or Raspberry Pi [ 7 , 15 , 16 ]. These stations can be very cost effective, with prices of several hundred Euros [ 3 ], but they often lack adequate calibration and testing, raising concerns about the accuracy, precision, and reliability of the collected data [ 17 ]. However, information on data quality in terms of both the accuracy and repeatability of such low-cost weather stations is crucial for modelling applications and decision-making [ 18 , 19 ].…”

Using Neural Networks to Estimate Site-Specific Crop Evapotranspiration with Low-Cost Sensors

“…The wind speed was not always necessary, although the time of day was also used as an input parameter and does not incur additional cost. With open source microcontrollers (such as Arduino), a complete sensor package could be constructed for as little as 500 USD [37]. Truly low-cost sensors could potentially reduce the cost further, although these were not evaluated in this study.…”

“…The ANN method developed in Matlab's Neural Network toolbox requires access to expensive and complex software; porting this approach to various open source computational toolboxes/libraries such as existing Python toolboxes [38,39] would allow for much broader access. Using the ANN method with very low-cost sensors such as the LEMS system [37] would also make this method more affordable for in field monitoring. Finally, using the ANN method to utilize agricultural weather networks would facilitate new uses for this valuable, publicly available data, with specific applications such as validation of remote sensing [40], spatial interpolation of reference ET [41] and determination of crop coefficients for new crops.…”

Enabling Multi-Site Stormwater Environmental Compliance Approvals in Ontario, Canada

“…In order to determine their robustness additional field testing would be required. Efforts to compare traditional and open source sensors have been carried out with variable results with some being able to withstand different clients and report reliable data and others being unable to perform based on their posted specifications (Rabault et al 2017) (Gunawardena et al 2018).…”