We are creating some awesome events for you. Kindly bear with us.

Artificial Intelligence to Identify Aquatic Plants in New Zealand

Water pollution has deteriorated due to rapid economic growth and increased urbanisation. Understanding water quality issues and patterns are also critical for water pollution reduction and regulation. To truly understand the quality of the marine ecosystem, most countries around the world have begun to develop environmental water management schemes.

Agriculture consumes 70% of freshwater today, with a utilisation efficiency of less than 50%. By reducing the amount of water used in irrigation, AI can help to streamline the use of water in agriculture. At the most basic level, AI systems can understand the water in the soil and estimate the water demand by considering crop growth-stage and managing sprinklers and drips. Advanced systems will be able to predict weather conditions, rain, humidity in the atmosphere etc and guide the sprinkler systems accordingly.

To meet the growing demand for IoT solutions from water utilities, telcos are expanding their capabilities by partnering with IoT platform vendors with water sector expertise. The next step in protecting Aotearoa New Zealand’s lakes and rivers from invasive aquatic weeds appears to be a combination of artificial intelligence and scientific ingenuity.

Management and detection of invasive submerged weeds cost millions of dollars annually, but National Institute of Water and Atmospheric Research (NIWA) researchers have developed a way to detect and identify submerged weeds. This technology will enable agencies to survey far larger areas more efficiently than is currently possible, and potentially lead to much faster responses to new incursions.

Submerged invasive weeds can degrade water quality, exacerbate silt and flooding, reduce the number of native animals and plants, and disrupt irrigation water delivery and hydroelectric power schemes. The National Institute of Water and Atmospheric Research (NIWA) developed a portable invasive species detector module that can be strapped to survey boats. The prototype is housed in a small waterproof case, complete with an underwater video camera. Inside is a computer with an artificial intelligence-based detector that has been trained to identify and log the locations of targeted invasive weed species in real-time.

Furthermore, NIWA’s principal technician has utilised a deep learning neural network – an artificial intelligence function – to train a computer model to recognise and record the GPS locations of two of New Zealand’s most invasive weeds, ‘lagarosiphon and hornwort’. These data can then be exported to a mapping programme to enable the implementation of control or eradication strategies.

“The deep learning process enables us to replace the human eyes and brain with a video camera and a computer by running a detection application that has been taught what to look for,” he said.

Training a detector requires a significant amount of computing power and can take days or even weeks depending on the complexity of the search environment and ‘target species.’ However, once trained, the trained detector is efficient and can be embedded in the computer inside the detector module for real-time detection.

The study is still in its early stages, and more fieldwork, data collection, and software development are needed to determine its true potential. However, according to a NIWA freshwater ecologist, early detection and prevention are critical for achieving effective freshwater biosecurity outcomes. The development of these detector modules will allow for rapid and cost-effective detection and mapping of large areas.

Currently, the majority of invasive species surveillance work is done by specialised divers. He also stated that the new technology has the potential to shift diver expertise from detection efforts to control strategy implementation.

Water quality modelling and prediction have played a pivotal and significant role in reducing time and consumption in lab analysis. Artificial intelligence algorithms were investigated as a potential alternative method for estimating and forecasting water quality.

Send this to a friend