Weather forecasting is an inaccurate science, yet one that has crucial implications for a number of domains. It has played pivotal moments in world history and contributes to shaping the course of the future: millions of decisions are made every day based on the weather, from food and clothing choices to crop management, to flight or shipping planning. Meteorology is also instrumental in the prevention of climate change or in the control of natural disasters or adverse atmospheric phenomena such as pollution.

According to a study conducted in 2005, climate change could directly cost the world economy $7.9 trillion by the middle of the century as increased drought, flooding and crop failures hamper economic growth and threaten infrastructure. Another study published in 2021 involving UCL researchers states that economic models of climate change may have substantially underestimated the costs of continued warming. The international team of scientists found that the economic damage could be six times higher by the end of this century than previously estimated. Presently, most models focus on short-term damage, assuming that climate change has no lasting effect on economic growth, despite growing evidence to the contrary. Extreme events like droughts, fires, heatwaves and storms are likely to cause long-term economic harm because of their impact on health, savings and labor productivity.

With the prospect of severe weather fronts occurring more frequently, governments and businesses are seeking supplemental and more accurate weather data to better prepare for impending disasters. That’s where IoT comes into the picture…

This article was originally published in August 2020 and updated in February 2022 to reflect the recent innovations in the field. 

Making weather forecasts right as rain: How IoT can help predict the weather more accurately

Although Mother Nature remains somewhat inscrutable, the capabilities of weather forecasting technologies have been bolstered with IoT-enabled devices.

Essential components of a weather forecast service are observation, communication, analysis, prediction, and dissemination, all of which can be helped by new technologies.

Weather sensors are being used to adjust metrics at ground and atmospheric levels, making the gathering of information more accurate. Taking many different shapes, sensors can be stationary or mounted on vehicles, or on drones to monitor meteorological trends and live changes in precise geographical areas. Cellular systems, sail drones, weather balloons and ocean-going robots take atmospheric and ocean measurements.

One such example of an application is the IOT Weather Drone Airship For Weather Forecasting developed by Nevon Projects. The smart weather monitoring station uses a zeppelin mechanism to send an array of weather sensors into the upper atmosphere for live atmospheric data transfer. This system offers a more reliable way to forecast weather. 

Side view of Nevon Projects’ IOT Weather Drone Airship For Weather Forecasting. 
 

Instruments mounted aboard satellites monitor and record global atmospheric conditions and cloud data by collecting numerous weather parameters that affect weather patterns. Satellite imagery has allowed scientists and meteorologists to track and understand large-scale weather patterns like never before. Satellites can provide near-real-time hydrologic, oceanic, climatic, solar, and space data enabling weather forecasting for areas both regionally and globally. Combined with weather data from other data sources they can predict weather patterns many days in advance. 

Long distance telecommunication protocols allow data to be transferred faster and from distant places to the cloud.

AI is helping making sense of the billions of datapoints gathered every day by private and public weather stations, upper-air stations, Voluntary Observing Ships (VOS), moored and drifting buoys, weather radars, specially equipped commercial aircraft and meteorological and research satellites that currently measure key parameters of the atmosphere, land and ocean surface every day. It also helps in generating a prediction by applying equations to data like temperature, wind direction and humidity, and presents that data in the light of historical data to predict future weather patterns. 

The focus of recent innovations —beyond delivering more accurate weather reports— has been to improve longer-timescale forecasts, as they can play a critical role for many sectors, including water conservation, energy demand, and disaster preparedness. This is how we have seen the advent of machine learning weather prediction systems called Deep Learning Weather Prediction (DLWP). Using a convolutional neural network, the system analyses past weather data, which differs from standard numerical weather prediction models that create mathematical representations of physical laws. These standard numerical forecasts perform better for short lead times (i.e 2–3 weeks out) but the DLWP model is more accurate when projecting weather patterns 4–6 weeks down the line.
 

Furthermore, sensors can also be connected to various destinations that may benefit from insights into meteorological trends such as news stations, airliners, logistics companies or event companies.  

Weather station on Mount Vesuvius
 

The IoT in the eye of the storm

Global warming has played a massive role in the development of meteorological and ecological IoT devices. The impact of weather on the world shouldn’t be underestimated, with 1/3 of our global economy being weather sensitive. Global warming impacts economic growth due to damage of property and infrastructure, lost productivity, mass migration and threats to the wellbeing of humankind. The 2019 Amazon rainforest wildfire that destroyed most the Amazon canopy, affecting humans and animals at various levels, and Hurricane Sandy, which flooded much of New York in 2012, are prime examples of how extreme weather events can financially impact countries. To anticipate and better manage these catastrophic events, it is essential to have more accurate and more actionable weather forecasts. To that end we have seen a growing number of weather start-ups, supported by insurance companies, public government organizations, commercial enterprises and homeowners, developing new innovative IoT and Artificial Intelligence (AI) to revolutionize weather forecasting and improve responses to natural catastrophes.
 

Detecting tornadoes thanks to the IoT

One such example is The Raspberry Boom. This low cost device is intended for personal use and is an atmospheric monitor that measures more than just the weather. It detects infrasound waves which are completely inaudible to humans. These sounds emanate from man-made and natural occurrences such as tornadoes, avalanches, meteors, nuclear explosions, sonic booms etc. Each device connects to one another on a live world map called Station View, creating a large citizen science infrasound array on a worldwide level.
 

Detecting avalanches thanks to the IoT

Wyssen, a Swiss company based at the foot of the Alps, has designed an IoT solution to detect and prevent avalanches using various monitoring systems (radars, infrasound sensors, geophones) and artificial triggering techniques with explosive charges. Sensors provide an early warning of possible avalanche activity in a given area based on detection of infrasound emissions. Radar installations, sensors, and cameras monitor activity, including feedback from weather stations, and send the data to a central platform where it is analyzed. The towers are linked to explosive charges that can be detonated if necessary once it is confirmed no humans are in the danger area.  

Photo @ wyssenavalanche.com

Detecting Tsunamis thanks to the IoT

The Indian NIOT (National Institute of Ocean Technology) was created in 1993 to develop technologies and their applications for the management of ocean resources & the environment. The organization designed a network of connected tsunami buoys in the Indian Ocean, that acts as a vital early warning system against devastating tsunamis for coastal communities such as the 2004 Indian Ocean tsunami that killed approximately 228 000 people in 14 countries and left two million homeless. The network of buoys is deployed along the deep and unstable fault-line responsible for the 2004 tsunami. The system, comprised of two units - a surface buoy and Bottom Pressure Recorders (BPRs), detects the sudden increases in pressure deep under the sea that indicate the formation of a tsunami. Communication between the BPRs and the surface is established through acoustic modems and the surface buoys use satellite communications to transmit the real-time information to the shore station, that can in turn inform local inhabitants. The surface buoys also collect a variety of other relevant sea data such as currents, conductivity and temperature. The surface buoys rely on Saft’s LS primary lithium batteries for their sole source of power. Battery reliability and long life are crucial factors, especially as the buoys are installed in the deep ocean, far from the shore, making maintenance visits extremely challenging and cost prohibitive.
 

Detecting volcanos activity thanks to the IoT

Similarly, the Internet of Things (IoT) has contributed to revolutionizing volcanology. GE partnered with Qwake, Libelium and the Nicaraguan Government to bring a volcano’s data online and attempt to better predict its deadly and random activity. Sensors, installed inside craters, allow experts to access information collected in real time about temperature, humidity and atmospheric pressure. Calorie and gas detectors measure changes in the magma and rate of flow and seismographs measure vibrations. Geophysical and geochemical techniques help predict volcanic eruptions and avoid major disasters. The cost and difficulty to maintain instrumentation in volcanic environments is tremendous. Sensors need to be encapsulated and hermetically sealed to protect them from high temperatures (up to a thousand degrees!). They need to have low consumption and long-range capabilities with month-long batteries.  

Predicting floods thanks to the IoT

Another interesting community IoT project is a flood sensor network developed by AB Open in the region of Calderdale in West Yorkshire. The region has been affected by severe and consistent flooding in the last few years which led AB Open to build IoT gateways using LoRaWAN and open source software to measure rising water levels and provide early warnings of incoming flood conditions and key insights to help prevent flooding in the future. The project relies on the local community whereby people can have a sensor installed if they live by the river and become a flood watcher. Other water levels such as groundwater levels can also be monitored in such flood watch application.  

Photo @ Chris Gallagher - Unsplash

AI can advise and automate responses for better reactivity to meteorological events

Modern weather forecasting uses sensors’ data sources to create fully aggregated computational forecasting models to forecast the weather accurately, but the development of AI has allowed the creation of more actionable reports. Rather than just predict the weather, systems are now able to analyze data and patterns to give indication to local communities about what to do or not to do. Some of them, coupled to control systems are even able to automate actions in response to meteorological events. 

The port of Rotterdam is often subjected to floods. To add value to maritime traffic control, the port Authority is conducting a test, using i4sea, an IoT application that develops accurate models that forecast the status of tides, currents, waves, wind and precipitation at a hyper-local level and advises about the predicted impact of weather on the operations of terminals, pilots, etc. 
 

In Rwanda, a chatbot driven by artificial intelligence and used via an app called Line has been developed to promote quick and immediate responses before, during and after natural disasters. This new tool strengthens data collection and processing, improves risk modeling, and ensures effective emergency communications to enable authorities and local populations to better anticipate and overcome crises.

Even more impressive is Glasgow’s Smart Canal, which is the first of its kind in Europe. It has been constructed using the old Forth & Clyde Canal and 21st-century technology to mitigate flood risk and reclaim land. This pioneering digital surface water drainage system uses predictive meteorological technology to provides advance warning of heavy rainfall. Using an autonomous and proactive technology, the system can control the canal’s water level in real time, thus addressing climate-related issues.

Glasgow's Smart Canal from Scottish Canals on Vimeo.

These new, efficient warning systems are using a combination of IoT sensor technologies; ultrasonic level sensors, infrasound sensors, pressure transducers, radar level sensors, atmospheric pressure, calorie and gas detectors, etc. These sensors and new low powered telecommunication networks are becoming more affordable, more robust and more reliable. Combined with trustworthy, long life batteries that can withstand extreme conditions, they are enabling non-governmental organizations and private owners to contribute actively to the safety plan by installing measurement and alert systems in their homes.  

As natural disasters are likely to become more frequent and more deadly, IoT technologies will help prevent natural catastrophes arriving as a bolt from the blue, and react quicker to them, thus saving lives and protecting local communities from economic disaster.  Ultimately, these applications will also help promote climate change management initiatives.