Climatologists rely on a wide range of tools, from ground-level weather stations and ocean-roaming robots to orbiting satellites and supercomputers running complex simulations. These instruments work together to measure, record, and predict how Earth’s climate behaves over time. Some capture what’s happening right now; others reconstruct conditions from thousands of years ago. Here’s a practical look at each category.
Automated Weather Stations
The most fundamental tools in climate science are the sensors embedded in automated weather stations scattered across the globe. A standard station includes four core sensor types: wind sensors (anemometers), temperature and humidity sensors, barometric pressure sensors, and rainfall sensors. Rainfall is typically captured by a tipping bucket gauge, a simple mechanical device that tips each time it collects 0.1 to 1 millimeter of rain, logging each tip as a data point. These stations run continuously, feeding measurements into long-term climate records that allow scientists to spot trends over decades.
Weather Balloons and Radiosondes
To understand the atmosphere above ground level, climatologists launch weather balloons carrying small instrument packages called radiosondes. These devices measure temperature, humidity, and pressure as they rise through the atmosphere, reaching altitudes of around 35 kilometers (roughly 22 miles) before the balloon bursts. The data they collect builds a vertical profile of the atmosphere, showing how conditions change with altitude. Hundreds of these balloons launch worldwide every day, and the readings they provide are essential for calibrating satellite data and feeding into climate models.
Satellites and Remote Sensing
Orbiting satellites give climatologists a global view that no ground-based network could match. Most climate satellites carry passive instruments that detect energy emitted or reflected by Earth’s surface and atmosphere. These fall into several categories. Radiometers measure the intensity of electromagnetic radiation in specific wavelength bands. Spectrometers analyze the spectral content of that radiation, which reveals atmospheric composition, including greenhouse gas concentrations. Sounders measure vertical profiles of temperature, pressure, and chemical composition from space, doing from orbit what radiosondes do from a balloon.
More advanced versions include imaging radiometers, which scan wide swaths of the planet to produce two-dimensional images, and hyperspectral radiometers, which detect hundreds of narrow spectral bands across visible and infrared light. Together, these instruments track sea surface temperatures, ice sheet extent, vegetation cover, and atmospheric chemistry on a continuous basis.
The SWOT Satellite
One of the newest additions to the satellite toolkit is the Surface Water and Ocean Topography (SWOT) mission, launched in December 2022. Orbiting 554 miles above Earth, SWOT measures the height of water in oceans, lakes, rivers, and reservoirs with 10 times better resolution than previous technologies. Its key instrument, the Ka-band Radar Interferometer (KaRIn), uses synthetic aperture radar to capture data from two 30-mile-wide swaths on either side of the spacecraft. The satellite covers more than 90% of Earth’s surface water every 21 days, tracking river reaches wider than 330 feet and lakes larger than 15 acres. By measuring water height and extent, scientists can calculate slope, flow rates, and volume changes across entire continents.
Ocean Monitoring With Argo Floats
The ocean absorbs the vast majority of the extra heat trapped by greenhouse gases, so tracking ocean conditions is critical. The Argo Program, developed in 1999, now maintains a global array of almost 4,000 robotic profiling floats. Each float drifts through the upper ocean on a 10-day cycle: it sinks to a depth of 2,000 meters (about 1.2 miles), then rises slowly to the surface while recording temperature and salinity at every depth along the way. Once at the surface, it transmits its data to satellites before sinking again.
This constant profiling lets scientists calculate how and where ocean heat content is changing. Newer floats currently being tested can dive to 6,000 meters (3.7 miles) and carry additional sensors to measure oxygen levels, pH, nitrate concentrations, suspended particles, and light penetration. These expanded measurements will give climatologists a much more complete picture of the ocean’s role in regulating climate.
Eddy Covariance Flux Towers
To measure how much carbon dioxide and methane move between the land surface and the atmosphere, climatologists use eddy covariance flux towers. These ground-based towers are equipped with gas sensors that continuously measure trace gas fluxes, capturing the real-time exchange of CO₂, methane, and water vapor. Networks of these towers are positioned in forests, grasslands, wetlands, and agricultural areas worldwide. The data they produce helps scientists understand which ecosystems absorb carbon and which release it, and how those patterns shift with temperature, drought, or land-use changes.
Ice Cores and Paleoclimate Proxies
Understanding past climates requires tools that can reach back thousands or even hundreds of thousands of years. The most powerful of these is ice core drilling. Scientists extract long cylinders of ice from glaciers and ice sheets in places like Greenland and Antarctica, then analyze the trapped air bubbles and chemical layers. Proxy climate indicators within the ice include oxygen isotope ratios (which reflect past temperatures), methane concentrations, and dust content. Each layer corresponds to a specific period, creating a detailed timeline of atmospheric conditions stretching back over 800,000 years. Cores are stored and studied at facilities like the U.S. National Ice Core Laboratory, which curates samples from drilling sites worldwide.
Tree-Ring Analysis
For climate records spanning the last several centuries to several thousand years, climatologists turn to tree rings. Each ring reflects one year of growth, and its width reveals information about temperature and moisture conditions during that growing season. Extracting samples involves an increment borer, a hollow drill that pulls a thin cylindrical core from a living tree without killing it.
Measuring ring widths has become increasingly sophisticated. Traditional methods use a stereoscope, a moving measurement table, and a data recorder. More modern approaches combine slider devices with linear encoders, couple digital cameras to microscopes, or detect ring boundaries automatically from scanned images by identifying abrupt changes in light intensity. Software packages built for the R programming environment, such as measuRing for ring-width detection and dendroTools for correlating ring data with daily environmental records, let researchers integrate measurements with powerful statistical analysis. These tools allow climatologists to build continuous temperature and precipitation records for regions where no instrumental data exists.
Supercomputers and Climate Models
All of the data collected by weather stations, satellites, ocean floats, and paleoclimate proxies ultimately feeds into climate models, which run on some of the world’s most powerful supercomputers. These models simulate the interactions between the atmosphere, oceans, land surfaces, and ice sheets to project how climate will change under different emissions scenarios. NASA’s Discover supercomputer, the centerpiece of its Center for Climate Simulation, contains over 213,000 processing cores capable of 8.28 petaflops, or 8,280 trillion calculations per second. The system generates weather forecasts, climate projections, reanalyses of past conditions, and land information datasets.
Running a full global climate simulation at high resolution is enormously expensive in computing time. A single projection can take weeks even on hardware this powerful. That computational bottleneck is one reason machine learning has become an increasingly important tool in climate science.
Machine Learning and AI
Machine learning is now used at nearly every stage of climate analysis. ML models can act as surrogates for full-scale Earth system simulations, approximating complex physical processes quickly enough to generate rapid climate projections under various greenhouse gas scenarios. They also post-process the output of traditional physics-based forecast models to produce higher-resolution, more accurate results. NASA’s Prism GPU cluster, equipped with dozens of high-performance graphics processors, is specifically designed to support this kind of AI-driven research, drawing on petabytes of curated climate and Earth observation data to train models. The combination of traditional physics-based modeling with machine learning is making climate projections both faster and more detailed than either approach could achieve alone.