AI-Enabled Antarctic Sea-Ice, Iceberg Trajectory, and Navigation Decision Support System
Ministry of Earth Sciences (MoES) · Smart Education · Software
The free data here is genuinely exceptional and the polar map demo is memorable, but the statement is one sentence covering three separate problems, so decide early that the ice forecast is the submission and let drift and routing be the application of it.
What it actually is
India's Antarctic research vessels have to find a way through sea ice that shifts week to week, while avoiding icebergs that drift on their own paths, and every extra mile through ice costs fuel and time. The ask is a system that forecasts where the ice will be, predicts where the icebergs are heading, and plans a route that is both safe and economical.
What to build
Three connected components behind one map: a sea-ice concentration forecast over a chosen horizon trained on the long satellite passive-microwave record with atmospheric and oceanographic drivers as inputs; an iceberg drift model propagating tracked bergs forward under wind, current and Coriolis forcing, validated against the historical tracking record; and a routing engine over the forecast ice field that costs each grid cell by transit difficulty rather than distance alone, returning a route that trades passage time against ice exposure and fuel, presented on a polar map with the forecast ice field, projected berg positions and the chosen track, plus the alternatives it rejected and why.
Smallest thing that wins the room
Show a forecast ice field for a date you held out, overlay the actual observed ice for that date to prove the forecast held, and then run the router to a station and show the chosen track threading the opening the forecast predicted.
How crowded this one gets
A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.
Quieter than 31% of the 226 · #157 of 226 by expected field
A normal-sized field. Your idea has to be good, not miraculous.
Why: central ministry statements sat below the average.
This is a guess, not a fact
Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.
The scores
The number is the shorthand. The line under it is the reason.
Acceptance potential
4/5Genuinely underrated — the satellite ice record and the iceberg tracking database are exceptional free resources with decades of history, sea-ice forecasting has published benchmarks you can position against, the visual is memorable, and polar navigation is a domain essentially no team will enter.
Feasibility
4/5The data situation here is unusually good — daily passive-microwave sea ice concentration reaches back to 1979, Antarctic iceberg positions are tracked and published, and reanalysis wind, current and temperature fields are free, so both the forecast and the drift model can be trained and back-tested against genuine long records rather than simulated ones.
Innovation scope
5/5The statement is a single sentence naming three tasks with no method, no horizon, no resolution, no vessel and no architecture specified, so every meaningful decision about what the system actually is belongs to you.
Clarity
2/5At 277 characters it names three substantial modelling problems and stops — there is no forecast horizon, no spatial resolution, no definition of a safe route, no vessel class and no success criterion, so you are writing the specification as well as building to it.
Effort
HeavyA spatiotemporal ice forecast, a physically grounded drift model and an ice-aware routing engine are three separate problems in three different disciplines, each with its own data pipeline and validation story, plus a map interface tying them together.
Demo-ability
EasyA polar map with a forecast ice field, drifting bergs and a ship's track threading between them is striking and self-explanatory, and holding out a real date so the forecast can be checked against what actually happened turns the visual into evidence.
In its favour
- Green flag: The satellite sea ice record runs continuously since 1979 at daily resolution, which is one of the best free training datasets available for any statement on this portal
- Green flag: Antarctic iceberg positions are tracked and published historically, so your drift model can be validated against real trajectories rather than assumed physics
- Green flag: Published sea-ice forecasting benchmarks exist, so you can position your skill against a documented baseline instead of defining your own standard
- Green flag: Holding out a real date and overlaying the observed ice field turns your demo into verifiable evidence rather than a rendering, which is rare for a forecasting submission
Against it
- Red flag: Three distinct modelling problems in a single sentence is a scope trap — a team that splits evenly across forecasting, drift and routing will do all three shallowly, and the forecast is the one the other two depend on
- Red flag: With 277 characters of specification you are defining the deliverable yourself, and a panel that reads the sentence differently will measure you against a system you did not build
- Red flag: Ice-aware routing needs a cost model for how hard a vessel finds a given ice concentration, and without a vessel ice class and resistance model your route optimises a difficulty you invented
- Red flag: Sea ice forecasting skill degrades sharply beyond a couple of weeks, so pick a horizon you can actually demonstrate skill at and report against persistence, which is a surprisingly strong baseline
What you will be writing
- passive microwave sea ice concentration time series
- convolutional LSTM or U-Net spatiotemporal ice forecasting
- iceberg drift model with wind, current and Coriolis forcing
- ERA5 and ocean reanalysis forcing fields
- A* or Dijkstra routing over ice-difficulty cost field
- polar stereographic map visualisation
- Polar and cryosphere science
- Spatiotemporal forecasting
- Maritime route optimisation
Prior art to read before you start
sea ice concentration forecasting · iceberg drift trajectory prediction · ice-aware vessel route optimisation
Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.