Antarctic radar data sharpen Southern Hemisphere weather forecasts

Weather forecasting is an uneven science. In the Northern Hemisphere, dense networks of ground stations, weather balloons, satellites, and aircraft reports continuously feed numerical models, keeping forecast errors comparatively small. In the Southern Hemisphere, the situation is starkly different. Antarctica and the vast Southern Ocean that surrounds it remain among the most poorly observed regions on the planet, and that observational gap has long translated into measurably less accurate forecasts for Australia, New Zealand, southern South America, and the polar continent itself. Now, a team of Japanese researchers has shown that a single radar installation on the Antarctic coast, running continuously through the polar winter, can meaningfully close that gap.

The new study, made available online on July 11, 2026, in the journal Scientific Reports, was led by Assistant Professor Kazutoshi Sato of the National Institute of Polar Research (NIPR), together with Professor Jun Inoue and Associate Professor Yoshihiro Tomikawa of NIPR, Dr. Akira Yamazaki of the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), and Professor Kaoru Sato of The University of Tokyo. Their central question was deceptively simple: if you feed hourly wind measurements from an Antarctic radar into a numerical weather prediction system, do the resulting analyses and forecasts improve? The answer, across seven atmospheric river events during the 2022 austral winter, was a clear yes.

The instrument at the heart of the study is the Program of the Antarctic Syowa Mesosphere–Stratosphere–Troposphere/Incoherent Scatter Radar, known as PANSY, located at Japan’s Syowa Station in Antarctica. Unlike radiosondes, the balloon-borne instrument packages that have anchored upper-air observation for decades, the PANSY radar measures wind conditions continuously across the atmosphere, returning observations at a far higher temporal resolution. Radiosondes are routinely launched twice a day at many stations worldwide, but in Antarctica the financial and operational constraints of the extreme environment make frequent launches impractical. A radar that never sleeps, unaffected by blizzards or the logistics of balloon launches, offers precisely the kind of persistent, high-frequency data stream that data assimilation systems can exploit.

“Observations from the Antarctic radar are not currently incorporated into operational numerical weather prediction systems. Our findings show that assimilating these observations can improve forecast accuracy, demonstrating the value of continuous radar measurements for weather forecasting,” says Assistant Professor Sato. That gap between what the radar can provide and what operational centers currently use is the crux of the research. Data assimilation, the mathematical process by which observations are blended with a model’s own forecast to produce the best possible estimate of the current atmospheric state, is exquisitely sensitive to the quantity and quality of input data. In regions where observations are sparse, each additional measurement carries disproportionate weight.

To test the radar’s value rigorously, the team ran parallel experiments using the ALEDAS data assimilation and forecasting system, developed and managed by JAMSTEC on the Earth Simulator, JAMSTEC’s supercomputer. The study period covered the 2022 austral winter, and crucially, both experimental arms included the enhanced radiosonde observations gathered during the Year of Polar Prediction in the Southern Hemisphere (YOPP-SH) campaign. That design choice matters. Rather than comparing radar-assisted forecasts against a starved baseline, the researchers isolated the incremental benefit of the PANSY observations on top of an already observation-rich special observing period. Seven atmospheric river events over the Southern Hemisphere mid-latitudes served as the test cases, chosen because these narrow corridors of intense moisture transport are both meteorologically critical and notoriously sensitive to initial-condition errors.

The results were striking. Assimilating the Antarctic radar observations improved the representation of atmospheric conditions in the analysis, including wind speed, temperature, and geopotential height, over Antarctica and the Southern Ocean. More dramatically, the presence or absence of PANSY radar data produced differences of more than 30 percent in the modeled moisture transport associated with atmospheric rivers. Those improvements did not remain confined to the analysis window; they persisted into the forecast period, enhancing predictions of atmospheric circulation and integrated water vapor linked to atmospheric rivers across Southern Hemisphere mid-latitudes. Across all seven cases examined, forecasts that included the radar observations showed lower prediction errors than those built on conventional observations alone.

Atmospheric rivers are the reason this matters far beyond the ice sheet. These elongated filaments of concentrated water vapor transport can deliver heavy rain, substantial snowfall, flooding, and damaging winds when they make landfall, and they directly affect Australia, New Zealand, South America, and Antarctica. Because they can span thousands of kilometers and originate over the data-sparse Southern Ocean, their forecast skill depends heavily on how accurately the model captures conditions far upstream. An observation over coastal Antarctica may therefore ripple through the entire forecast chain, correcting the initial state from which these moisture plumes evolve. For communities that face atmospheric river impacts, better initialization over the polar margin translates directly into longer warning times and more reliable warnings.

“Although radiosonde observations are routinely performed to improve weather forecasts, high-frequency launches are challenging in extremely cold environments. We therefore proposed a new method for improving forecast accuracy using instruments capable of continuous, high-temporal-resolution observations,” explains Professor Inoue. The method he describes is essentially a substitution strategy: where the traditional observing network cannot be densified, instrument classes with different operational profiles can fill the void. Ground-based radars, by their nature, tolerate conditions that would ground balloon campaigns, and their continuous sampling gives assimilation systems a stream of data that is more evenly distributed in time, reducing the analysis errors that otherwise grow rapidly between sparse observation times.

The implications extend beyond operational forecasting into climate science. Reanalysis datasets, which combine historical observations with models to reconstruct past atmospheric states, carry large uncertainties over Antarctica precisely because of the sparse observational record. These datasets are widely used to estimate the surface mass balance of the Antarctic Ice Sheet under climate change, a quantity central to projecting future sea-level rise. If assimilating PANSY radar observations reduces uncertainties in the analysis over the continent, the same principle could sharpen the reanalyses that underpin ice sheet studies, tightening the link between meteorological observation and climate understanding.

The study ultimately makes a broader argument about where the next gains in forecast skill will come from. As numerical models grow more sophisticated, their accuracy is increasingly limited not by physics but by the observations available to constrain them. Uncertainties in Antarctic atmospheric conditions can spread far beyond the polar continent, which means that strengthening observations in data-sparse regions yields dividends across much of the hemisphere. Integrating continuous measurements from atmospheric radars and other emerging technologies into operational systems, the researchers suggest, could make forecasts more reliable for disaster preparedness, aviation and shipping, scientific operations, and the communities who bear the brunt of extreme weather. A single radar on the Antarctic coast cannot transform global prediction on its own, but as this study demonstrates, even one persistent instrument in one of Earth’s emptiest observing regions can measurably improve the forecasts that millions of people depend on.

Subject of Research: Assimilation of Antarctic MST radar wind observations into numerical weather prediction to improve Southern Hemisphere atmospheric river forecasts

Article Title: Can Antarctic radar improve southern hemisphere weather forecasts?

Article References: Can Antarctic radar improve southern hemisphere weather forecasts?. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: Antarctica, PANSY radar, weather forecasting, atmospheric rivers, data assimilation, Southern Hemisphere, Syowa Station, National Institute of Polar Research, Scientific Reports, reanalysis, radiosondes, YOPP-SH