Ministry of Earth Sciences (MoES)
Browse and strategize across all 30 official problem statements submitted by Ministry of Earth Sciences (MoES) for Smart India Hackathon 2026.
Problem Statements (30)
Match My TeamAI-Powered Automated Underwater Marine Debris and Anomaly Detection System using Side-Scan Sonar Imagery
Marine plastic debris, discarded ghost fishing nets, and submerged pollutants in coastal and deep Indian waters degrade marine biodiversity and threaten submarine infrastructure with virtually zero automated underwater detection systems. Build an AI-powered automated underwater marine debris and anomaly detection platform for Autonomous Underwater Vehicles (AUVs) and ROVs that uses deep learning computer vision with underwater optical restoration (water turbidity dehazing) to identify, classify, and map marine litter on the seafloor.
Development of a Low-Power, Real-Time Adaptive Software-Defined Sonar Transmitter Payload for Autonomous Underwater Vehicles (AUVs)
Autonomous Underwater Vehicles (AUVs) exploring deep Indian waters require acoustic sonar for navigation and seafloor mapping, but commercial hardware sonars are bulky, consume excessive battery power, and have rigid unmodifiable signal processing. Build a low-power, real-time Adaptive Software-Defined Sonar (SDS) platform for AUVs in GNU Radio / C++ that implements reconfigurable acoustic waveforms (LFM chirp, CW pulses), adaptive matched filtering, and side-scan bathymetric image reconstruction.
AI-Enabled Antarctic Sea-Ice, Iceberg Trajectory, and Navigation Decision Support System
Indian Antarctic expeditions (vessels travelling from Cape Town to Bharati/Maitri stations) face catastrophic iceberg collisions, thick pack-ice besetting, and severe polar storms with delayed satellite ice charts. Build an AI-enabled Antarctic sea-ice and iceberg trajectory forecasting decision support platform combining multi-mission SAR satellite imagery (Sentinel-1 / RISAT), optical MODIS feeds, and ocean-wind drift physics to calculate optimal fuel-efficient and safe polar navigation corridors.
Digital Platform for efficient remote management of Indian Antarctic Research Stations
Indian Antarctic Research Stations (Maitri and Bharati) operate thousands of kilometers away in sub-zero isolation, where equipment telemetry (diesel generators, HVAC life support, water purification, scientific instruments) is monitored on fragmented local computers without unified remote digital twin oversight from NCPOR Goa. Build a centralized Digital Platform for remote management of Indian Antarctic Research Stations with satellite IoT telemetry synchronization, predictive life-support maintenance, and emergency response coordination.
AI-Driven Smart Energy Management System for Polar Research Stations
Indian Antarctic stations rely almost exclusively on burning expensive, carbon-intensive Aviation Turbine Fuel (ATF / Jet A-1) transported by ship for diesel generator power and heating, facing severe logistics costs and environmental impact under Antarctic Treaty regulations. Build an AI-driven Smart Microgrid & Energy Management System for Polar Research Stations that dynamically orchestrates wind turbines, bifacial solar PV arrays, battery energy storage (BESS), and diesel cogeneration (CHP) to minimize fuel consumption.
Integrated Polar Expedition Logistics and Asset Management System
Indian Antarctic and Arctic scientific expeditions involve complex international logistics (chartering ice-class ships, cargo flights from Cape Town, shipping containers, specialized polar gear, food rations, heavy snow vehicles) across multi-month planning cycles where lost cargo or expired maintenance parts can jeopardize year-long scientific missions. Build an Integrated Polar Expedition Logistics and Asset Management Platform for NCPOR that tracks multimodal shipments, manages extreme-environment asset lifecycles (PistonBully snow groomers, cranes, skidoos), and optimizes container packing.
Integrated Polar Science Outreach, Knowledge Repository and Media Dissemination Portal
India's landmark scientific discoveries in the Polar regions (Antarctica, Arctic, Himalayas/Third Pole) and Southern Ocean remain hidden in academic journals, limiting public awareness, school STEM engagement, and accessible media dissemination. Build an Integrated Polar Science Outreach, Interactive Knowledge Repository, and Media Dissemination Platform for NCPOR featuring interactive 3D digital twins of polar stations, virtual Antarctic expeditions, open scientific dataset downloads, and school educational gamification.
Low-Cost Deployable Seafloor Metal Detection Sensor for Ocean Resource Exploration
Deep seafloor exploration in the Indian Ocean for Polymetallic Nodules and seafloor massive sulfides (copper, nickel, cobalt, manganese) is hindered by heavy, expensive commercial survey systems. Build a low-cost, lightweight deployable seafloor metal detection sensor and digital signal processing system for National Institute of Ocean Technology (NIOT) that utilizes pulsed eddy-current electromagnetic induction (EMI) to identify and classify buried metal ore deposits on the ocean floor.
Autonomous Low-Cost Ocean Observation Platform for Polar and Southern Oceans
Long-term oceanographic and climate monitoring in the harsh, storm-tossed Southern Ocean and Antarctic waters is severely limited by the high operational cost of research vessels and lack of low-cost autonomous observation platforms. Build an Autonomous Low-Cost Ocean Observation Platform (Wave/Solar powered surface drifter/glider) and digital telemetry portal for NCPOR that measures sea surface temperature, salinity, wave height, dissolved oxygen, and acoustic ambient marine noise with satellite cloud sync.
OceanEmbed - Satellite Embedding-Based Deep Learning Framework for Reconstruction of Subsurface Ocean Temperature from Surface Satellite Observations.
Raw satellite ocean observations (SST, Chlorophyll-a, Sea Surface Height) have missing data gaps from cloud cover and low spatial resolution, limiting marine fisheries and cyclone track models. Build 'OceanEmbed'—a self-supervised foundation model that generates dense spatio-temporal embeddings from multi-satellite rasters to reconstruct missing oceanic fields, super-resolve Sea Surface Temperature (SST) to 1 km resolution, and predict Potential Fishing Zones (PFZ) for INCOIS.
Develop a web-based interactive 3D visualization platform that integrates numerical ocean model outputs and in-situ observations.
INCOIS and oceanographic scientists generate complex multi-dimensional ocean data (3D temperature depth profiles, salinity thermoclines, subsurface currents, bathymetric trenches) that are locked in static 2D plots, preventing intuitive visual analysis. Build an interactive web-based 3D oceanographic visualization and digital twin platform using WebGL / CesiumJS that renders multi-depth oceanic layers, animated particle current vectors, and interactive vertical thermocline slices.
WeatherGPT: Conversational AI for Weather Forecasting, Alerts, and Climate Information
Citizens and farmers in India struggle to understand complex technical weather bulletins (convective available potential energy, isobars, hectopascals) issued by the India Meteorological Department (IMD), while static mobile apps fail to answer specific local questions. Build 'WeatherGPT'—a conversational AI weather intelligence platform combining grounded IMD/NCMRWF forecasts, live Doppler radar feeds, and regional language voice LLMs to provide hyper-localized weather advice, agricultural spray advisories, and disaster safety alerts.
National Weather Big Data Analytics Platform
The Ministry of Earth Sciences (MoES) operates thousands of weather observation platforms (Doppler Weather Radars, Automatic Weather Stations, satellites, lightning sensors, climate models) that generate petabytes of high-velocity data in incompatible formats, creating analytical bottlenecks for forecasters. Build a National Weather Big Data Analytics Platform that provides high-throughput streaming ingestion, distributed raster query engines, and predictive extreme weather anomaly discovery for IMD meteorologists.
To develop an Artificial Intelligence (AI) / Machine Learning (ML) based system for identification, classification, and prediction of different tropical cyclone patterns using multi-source satellite data.
Dense winter radiation fog across the Indo-Gangetic Plains (Delhi, Punjab, UP, Bihar) disrupts hundreds of flights, passenger trains, and highway freight operations daily with low-visibility conditions (<50 meters) that existing numerical models fail to forecast accurately at sub-kilometer scales. Build an AI/ML-based high-resolution fog and visibility nowcasting system combining satellite land surface temperature, boundary layer humidity sensors, and Ceilometer backscatter to forecast Runway Visual Range (RVR) up to 12 hours in advance.
AI/ML-Based Integrated heavy rainfall Early Warning and Inundation Prediction System using Satellite, Radar, observational Weather and numerical weather prediction model data.
Extreme monsoon downpours in major Indian cities (Mumbai, Chennai, Bengaluru, Delhi) cause catastrophic urban flooding within 60 minutes due to outdated stormwater drain capacities and lack of integrated hydraulic modeling. Build an AI-driven integrated heavy rainfall early warning and 2D urban inundation prediction system coupling Doppler weather radar rainfall nowcasts with municipal drainage networks to forecast street-level water logging depth.
AIML based Nowcasting of thunderstorm and lightning using atmospheric observation including multiple radars, satellite, lightning and model data.
Severe thunderstorms, squalls, and cloud-to-ground lightning strikes kill over 2,500 rural farmers and laborers annually in India due to sudden onset and lack of localized nowcasting. Build an AI/ML-based nowcasting and early warning platform for IITM/IMD that ingests Lightning Location Network (LLN) pulses, Doppler radar reflectivity, and INSAT-3D cloud-top temperatures to forecast thunderstorm cell paths and lightning strike hazard zones 30-90 minutes in advance.
AI/ML-Based Intelligent Anomaly Detection for Automatic Weather Stations (AWS)
The India Meteorological Department (IMD) operates a nationwide network of 1,200+ Automatic Weather Stations (AWS), but field sensors suffer from sensor drift, clogged rain gauge funnels, solar battery depletion, and bird roosting, leading to erroneous data entering national numerical weather prediction models. Build an AI/ML-based intelligent anomaly detection and automated quality control platform for IMD that flags stuck sensors, physical drift, and temporal-spatial inconsistencies in real time.
Downscaling of weather forecast from Block level to Panchayat level: Inferring high-resolution plots/ data/ information from low-resolution plot /data /information /variables for agro-meteorological advisory services.
IMD currently produces Numerical Weather Prediction (NWP) forecasts at coarse 12 km grid resolution (Block level), but rural farmers need accurate rain, temperature, and wind forecasts at the 1-3 km Gram Panchayat / village level where localized terrain, microclimates, and water bodies create severe weather variations. Build an AI/ML statistical downscaling platform combining high-resolution topography (DEM), land cover, and localized AWS observations with NWP model outputs to generate Panchayat-level 1 km hyper-local forecasts.
Participants are invited to design and develop **CAPACITY CONNECT A Digital Capacity Building and Learning Management Portal** to support organizational training, competency development, and knowledge sharing through a centralized web-based platform.
Indian researchers, universities, and students struggle to access advanced scientific infrastructure (oceanographic research vessels, supercomputing clusters, polar test facilities, mass spectrometers) across MoES institutes (NCPOR, NIOT, INCOIS, IITM, NCMRWF) due to lack of a unified booking and training ecosystem. Build 'CAPACITY CONNECT'—a centralized digital research capacity building, scientific equipment sharing, and internship management platform for the Ministry of Earth Sciences (MoES).
Development of personalized homepage for 'Mausam' mobile application:
The official IMD 'Mausam' mobile application provides a one-size-fits-all interface that bombards users with static national data rather than prioritizing the exact information relevant to their lifestyle (e.g. farmers need rain & soil moisture, daily commuters need thunderstorm nowcasts and air quality, fishermen need wave heights and wind). Build an AI-driven personalized dynamic homepage and modular widget engine for the 'Mausam' app that adapts its UI, alerts, and content based on user persona, location, and severe weather proximity.
AI-Driven Hyper-Local Early Warning System for Severe Weather Nowcasting
Severe local convective storms (thunderstorms, squalls, microbursts) develop and dissipate within 1 to 2 hours over small 5-10 km areas, evading traditional numerical models. Build an AI-driven hyper-local severe weather nowcasting system combining dual-polarization Doppler weather radar volumetric scans, INSAT-3D rapid-scan infrared imagery, and surface AWS networks to generate automated 0-2 hour convective storm cell tracks and wind gust nowcasts at 500m resolution.
AI-Driven Spatio-Temporal Tracking of Extreme Weather Anomalies in Medium-Range Forecasts
National Centre for Medium Range Weather Forecasting (NCMRWF) runs global and regional Numerical Weather Prediction (NWP) ensemble models (NCUM) producing massive multi-dimensional forecast grids, where detecting emerging extreme weather anomalies (monsoon low-pressure depressions, heat dome patterns, atmospheric rivers) across 10-day forecast horizons requires tedious manual analysis. Build an AI-driven Spatio-Temporal Extreme Weather Anomaly Tracker for NCMRWF that uses 3D Convolutional Neural Networks and extreme value statistics to automatically identify, track, and rank extreme meteorological anomalies across ensemble forecast members.
AI-Based Forecast Bust Detection for Medium-Range Weather Forecasts
Numerical Weather Prediction (NWP) models occasionally suffer from catastrophic 'forecast busts'—where forecasted weather drastically fails in reality (e.g. predicting clear skies when a torrential downpour occurs, or missing the track of a monsoon depression) due to initial condition errors or convection parameterization failures. Build an AI-based Forecast Bust Early Detection and Confidence Scoring platform for NCMRWF that analyzes real-time observation discrepancies, ensemble spread-error relationships, and atmospheric instability indicators to flag low-confidence forecast runs before public release.
Regime-Aware AI Post-Processing of Monsoon Rainfall Forecasts
Indian Summer Monsoon rainfall exhibits distinct large-scale atmospheric regimes (Active vs Break phases, Madden-Julian Oscillation MJO phases, El Niño / IOD states) where raw Numerical Weather Prediction (NWP) models suffer from regime-dependent systematic biases (e.g. overestimating dry spells during break phases or mislocating the monsoon trough). Build a Regime-Aware AI Post-Processing and Bias-Correction platform for NCMRWF that uses unsupervised regime clustering and physics-guided neural networks to dynamically correct monsoon rainfall forecasts based on active large-scale climate states.
Hybrid AINWP Multi-Model Forecast Blending System
NCMRWF and IMD run multiple distinct Numerical Weather Prediction models (NCUM Global, NCUM Regional, GFS, WRF, ECMWF) alongside emerging pure-AI atmospheric models (GraphCast, Pangu-Weather, FourCastNet), each with differing strengths across lead times and geographical regions. Build a Hybrid AI-NWP Multi-Model Forecast Blending System that dynamically computes optimal spatial-temporal blending weights using Bayesian Model Averaging and deep neural meta-learners to generate a single unified consensus forecast.
Air PollutionWeather Coupled Forecasting System (Delhi NCR Focus)
Severe winter air pollution in Delhi-NCR (hazardous PM2.5 / PM10 AQI > 450) is driven by an explosive combination of stubble burning emissions, vehicular exhaust, and stagnant meteorological conditions (nocturnal temperature inversions, calm surface winds <1 m/s, low boundary layer height). Build a coupled Air Pollution-Weather Forecasting platform for IITM/SAFAR combining WRF-Chem atmospheric chemistry models, satellite fire radiative power (FRP), and deep learning to forecast hyper-local AQI and stubble smoke dispersion 72 hours in advance.
Extreme Heatwave Early Warning and Human Thermal Stress Index
Severe summer heatwaves across India claim thousands of lives and cause massive labor productivity loss, while standard temperature forecasts fail to capture the lethal physiological threat of 'Wet-Bulb Temperature' (combined extreme heat and high humidity where the human body cannot cool down through sweating). Build an AI-driven Extreme Heatwave Early Warning and Human Thermal Stress Index platform for IMD that computes localized Wet-Bulb Globe Temperature (WBGT), Universal Thermal Climate Index (UTCI), and neighborhood-level heat vulnerability heatmaps.
Convective scale nowcasting for Thunderstorms, Hail & Cloudbursts (06 hr)
Severe localized convective cloudbursts, severe hailstorms, and squall lines across the Himalayas and central plains develop within 30 minutes, causing sudden flash floods and severe crop destruction before traditional numerical models can react. Build an AI-driven convective scale nowcasting platform for IMD (0-6 hour forecast horizon) that assimilates 3D Doppler radar reflectivity grids, INSAT-3D rapid-scan cloud-top cooling rates, and GNSS Integrated Water Vapor (IWV) to predict cloudburst initiation and hail size with sub-kilometer precision.
Urban Flood Nowcasting System (Drainage and Rainfall Coupling)
Indian metropolitan cities (Bengaluru, Chennai, Mumbai, Hyderabad) suffer chronic urban waterlogging where water accumulates in specific street intersections within 30 minutes of rain, but municipal corporations lack real-time visibility into drainage bottleneck points. Build an Urban Flood Nowcasting and Drainage Coupling System for IMD and Municipal Corporations that couples high-resolution Doppler radar rainfall nowcasts with municipal stormwater pipe GIS topologies to predict sub-catchment waterlogging depths and pump dispatch schedules.
Hyperlocal Monsoon Onset & Break Prediction System (Block/Village Scale)
Indian farmers make critical crop sowing and seed investment decisions based on the onset and dry-spell break periods of the Southwest Monsoon, but current IMD onset forecasts are issued at broad regional scales (e.g. 'Monsoon onset over Kerala'), failing to predict the exact village-level arrival date or prolonged 2-3 week agricultural break phases. Build an AI-driven Hyperlocal Monsoon Onset & Break Prediction System for IMD that predicts village-scale onset dates, dry-spell break durations, and sowing risk windows 15-30 days in advance.