Telangana lost 18 m of groundwater in 25 years. Nizamabad, Karimnagar, and parts of Warangal are over-exploited. AgriSense AI predicts borewell sustainability, suggests AWD & SRI for paddy, schedules pump runtime, and recommends rainwater harvesting structures β by farm, by village, by district.
Standing-water paddy uses 5,000 L per kg of rice. AWD and SRI cut this by 22-40% with no yield loss.
Keep field flooded for first 2 weeks. After that, let water drop 15 cm before re-flooding. AI tells you exact re-flood day via PVC tube sensor reading.
-22% water Β· no yield lossSingle seedling spaced 25 cm. No standing water β just moist soil. Yields 18% higher than flood-paddy.
-40% water Β· +18% yieldSub-surface drip lines. 70% less water. PMKSY 55% subsidy. Pre-filed by AgriSense AI.
-70% water Β· βΉ68k subsidyΒ±2 cm flat field = even water distribution. One-time service. 18-25% water saving on flood-paddy.
-22% water Β· βΉ850/ac one-time| District | Block category (CGWA) | Avg depth | Trend (5 yr) | Status | Action |
|---|---|---|---|---|---|
| Nizamabad | Over-exploited | 58 m | βΌ 14 m | Critical | SRI mandatory Β· borewell ban |
| Sangareddy | Over-exploited | 62 m | βΌ 16 m | Critical | AWD + recharge pit |
| Karimnagar | Semi-critical | 48 m | βΌ 9 m | Stressed | AWD + farm pond |
| Warangal Urban | Semi-critical | 42 m | βΌ 8 m | Stressed | AWD recommended |
| Bhadradri Kothagudem | Safe | 22 m | βΌ 3 m | Safe | Maintain current practice |
| Khammam | Safe | 28 m | βΌ 4 m | Safe | Recharge pit recommended |
| Mulugu | Safe | 18 m | βΌ 2 m | Safe | Continue monitoring |
Random Forest + Bayesian uncertainty trained on Central Ground Water Board (CGWB) aquifer maps, Geological Survey of India (GSI) lithology, Sentinel-2 NDWI, 5-year rainfall, electricity consumption per pump, and reported borewell yields. Output: depth probability, sustainability score, draft-recharge ratio, and 12-month forecast per farm. Calibrated against 8,422 borewell observations across Telangana.