AWS Builder Center

Builder Center

JalFaasla: Tackling Punjab’s Groundwater Depletion with AWS Serverless and Geospatial GIS

JalFaasla: Tackling Punjab’s Groundwater Depletion with AWS Serverless and Geospatial GIS

How we built JalFaasla on AWS Lambda, API Gateway, and S3 to translate Punjab's groundwater crisis into actionable farmer economics—calculating exact tube-well hours and ₹ electricity saved through statutory transplanting alignment, Alternate Wetting-and-Drying (AWD), and omnichannel voice/SMS advisory.

JalFaasla: Tackling Punjab’s Groundwater Depletion with AWS Serverless and Geospatial GIS

"Don't just say 'save water.' Tell a farmer: on your plot, delaying transplanting to the notified date and using alternate wetting-and-drying saves tube-well hours and ₹ electricity/diesel costs this season."


📌 Live Deployments & Endpoints

Resource
URL
Infrastructure & Provider
🌐 Production Web App
 https://jalfaasla.abhinavkumarsingh.tech/ 
Cloudflare Edge CDN +
 Amazon S3 
Static Website Hosting
⚡ Serverless Backend API
 https://ofkasl4fri.execute-api.ap-southeast-2.amazonaws.com  Amazon API Gateway 
HTTP API +
 AWS Lambda 
(Python 3.12)
🩺 Live API Health Probe
 https://ofkasl4fri.execute-api.ap-southeast-2.amazonaws.com/health 
Sub-millisecond health status probe & uptime
💻 Open Source Codebase
 https://github.com/abhinav29102005/JalFaasla 
Full GitHub repository with CI/CD and tests
🪣 AWS Storage Bucket
 s3://jalfaasla.abhinavkumarsingh.tech 
Amazon S3 Storage Bucket ( ap-southeast-2 )


🌾 1. The Context: A Crisis Beneath the Soil

Punjab is the grain basket of India, contributing over 12% of the nation's rice and 18% of its wheat. However, this intensive agricultural output has come at a severe ecological cost: critical groundwater exhaustion.
According to official assessments by the  Central Ground Water Board (CGWB) :
  • 114 out of 150 administrative blocks (76%) across Punjab are categorized as Over-Exploited, withdrawing far more water annually than natural monsoon recharge replenishes.
  • In central agricultural districts like Sangrur, Patiala, Ludhiana, and Moga, the groundwater table drops by 50 to 90 cm every single year.
  • Tube-well bore depths have plunged from 30–40 feet during the 1970s Green Revolution to 300–450 feet today, forcing farmers into recurring debt to install deep submersible pumps rated at 10–25 Horsepower (HP).

The Pre-Monsoon Evaporative Trap

When farmers transplant puddled paddy in scorching May or early June, the atmospheric vapor pressure deficit is severe. Reference crop evapotranspiration ($ET_0$) peaks at 8–10 mm/day. Puddling fields under extreme heat demands 4,500–5,000 liters of water per kg of grain. Conversely, crops transplanted in late June (aligned with the onset of the monsoon) need only 2,800–3,200 liters/kg.
To arrest this decline, the state enacted the  Punjab Preservation of Subsoil Water Act, 2008 , which legally prohibits nursery sowing before May 15/20 and transplanting before June 15/20.
Yet, abstract appeals to "save water" routinely fail on the ground. Farmers need to see the tangible operational math: how many pumping hours, kilowatt-hours, and hard cash (₹) will they save on their specific plot of land?
This is why we built  JalFaasla (ਜਲ ਫ਼ਾਸਲਾ) .


💡 2. What JalFaasla Delivers

JalFaasla is a groundwater-aware decision support engine that converts agricultural hydrological physics into practical farm-level unit economics:
  • ⏱️ Quantified Tube-Well Hours: Computes the exact pumping hours eliminated per season based on plot acreage, pump horsepower (HP), dynamic lift depth, and soil category.
  • 💰 Financial & Energy Savings: Shows farmers the monetary savings in electricity tariffs (or equivalent diesel backup generator costs).
  • 🌱 Multi-Pathway Agronomic Evaluation:
  • Timely Statutory Transplanting: Aligning transplanting with the notified date to bypass early-summer peak evapotranspiration.
  • Alternate Wetting and Drying (AWD): Using  Punjab Agricultural University (PAU)  recommended field water tubes (pani pipe) to safely reduce irrigation water consumption by 25–30% with zero yield reduction.
  • Direct-Seeded Rice (Tar-Wattar DSR): Eliminating puddling entirely to reduce seasonal groundwater draft by 15–20%.
  • Kharif Crop Diversification: Comparing high-value, lower-water crops such as Basmati 1509/1121, Maize (Makki), and Summer Moong.
  • 🗺️ Interactive Geospatial GIS: Built with  Leaflet  mapping all 23 Punjab districts with official GeoJSON administrative polygons, OpenStreetMap Road & Topographic terrain layers, and key river networks (Sutlej, Beas, Ravi).
  • 📲 Omnichannel Last-Mile Delivery:
  • Voice Advisory: Browser-native  Web Speech API  speech synthesis in Punjabi and Hindi for low-literacy farmers.
  • WhatsApp Community Dispatch: Generates formatted Gurmukhi & English advisory cards ready for village kisan groups.
  • Kisan SMS Gateway: Formatted 160-character cellular SMS text for feature phone delivery.


🏛️ 3. Architecture: Why AWS Serverless?

Agricultural advisory tools have distinct traffic dynamics: demand spikes heavily during the May–July Kharif sowing and transplanting window, while seeing lower utilization during off-season months.
Running traditional virtual servers ( Amazon EC2 ) would create idle costs and maintenance overhead. We architected JalFaasla as a 100% serverless solution on AWS using the  AWS Serverless Application Model (AWS SAM) :
Plain text
[ Farmer Client (React 19 + Vite) ]
                 │
                 ▼
      [ Cloudflare Edge CDN ]
      ┌──────────┴──────────┐
      ▼                     ▼
 [ Amazon S3 Bucket ]   [ Amazon API Gateway ]
(Static Web Hosting)     (HTTP API v2)
                             │
                             ▼
                        [ AWS Lambda ]
                     (Python 3.12 + FastAPI)
                             │
               ┌─────────────┼─────────────┐
               ▼             ▼             ▼
          [ CGWB Data ] [ PAU AWD ] [ 2008 Act Engine ]

Key Architectural Components:

  1. Frontend Hosting on  Amazon S3 : The client bundle is hosted on an Amazon S3 static website bucket in the  ap-southeast-2  region and served via edge CDN for fast sub-second loading on mobile networks.
  1.  Amazon API Gateway (HTTP API v2) : Provides low-latency, secure API routing with native CORS handling and direct integration to compute.
  1.  AWS Lambda  (Python 3.12 +  FastAPI ): Executes our hydrological calculations via  Mangum  ASGI adapter. Execution time averages under 40ms with minimal memory footprint.
  1. Infrastructure as Code (IaC) with  AWS SAM : The full serverless stack is codified in  template.yaml , enabling deterministic automated deployments ( sam build && sam deploy ).


🧮 4. Scientific Agronomic Models & Mathematical Formulas

Our backend advisory engine implements the mathematical models developed by the  Food and Agriculture Organization (FAO)  and calibrated by  Punjab Agricultural University (PAU) :

Formula 1: Crop Evapotranspiration (ET_c)

Water consumption is modeled using the FAO-56 Penman-Monteith dual crop coefficient method:
text
ET_c = ET₀ × K_c
  • ET₀ (Reference Evapotranspiration): Daily baseline evaporative demand driven by solar radiation, temperature, wind, and relative humidity:
  • May–June (Pre-Monsoon heat peak): 8.5 mm/day
  • July–August (Monsoon season): 4.8 mm/day
  • September (Maturity stage): 3.8 mm/day
  • K_c (Crop Coefficient): Crop-specific physiological factor:
  • Initial vegetative stage: 1.05
  • Mid-season reproductive stage: 1.20
  • Late maturity stage: 0.90
By shifting transplanting from May 25 to the state-notified date of June 20, the crop bypasses 26 days of extreme May heat and atmospheric dryness:
text
ΔW_evaporation = Σ (ET_early - ET_fallow) ≈ 180 to 220 mm
(This translates to saving 1.8 to 2.2 million liters of extracted groundwater per hectare, or 730,000 to 890,000 liters per acre).


Formula 2: Tube-Well Pumping Energy & Dynamic Head

The electrical and mechanical power required to extract groundwater from deep aquifers is governed by hydraulic head physics:
text
                     ρ · g · Q · H_dynamic
P_hydraulic (kW)  =  ───────────────────────
                         1000 · η_pump
Where:
  • ρ (Water Density): 1,000 kg/m³
  • g (Gravitational Acceleration): 9.81 m/s²
  • Q (Discharge Rate): Water discharge rate in m³/s (calibrated to ~25–35 L/s for 10–12.5 HP submersible pumps)
  • H_dynamic (Total Dynamic Head in meters): Static Water Table Depth + Drawdown during pumping + Frictional Head Loss in delivery pipes
  • η_pump (Pump Operating Efficiency): Overall motor-pump efficiency (calibrated to 0.62 / 62% for agricultural pumpsets)
From this hydraulic equation, the seasonal operational savings are computed as:
text
Operating Hours Saved = (ΔWater Volume Saved in m³) / (Discharge Rate Q × 3600)

Cost Saved (₹) = Hours Saved × Motor kW Rating × Effective Electricity Tariff (₹/kWh)


Backend Python Implementation in AWS Lambda

In our  AWS Lambda  handler, these formulas execute in real-time on every evaluation:
Python
# From backend/app/main.py
@app.post("/advisor/evaluate")
def evaluate_farm_plan(request: FarmEvaluationRequest):
    district = DISTRICT_PROFILES.get(request.district)
    
    # Baseline evapotranspiration penalty if transplanted before statutory cutoff
    evap_penalty_mm = calculate_pre_monsoon_evaporation(
        district=request.district,
        transplant_date=request.transplant_date,
        notified_date=district.notified_cutoff_date
    )
    
    # Discharge rate based on pump horsepower and bore depth
    discharge_lps = calibrate_discharge(request.pump_hp, request.bore_depth_feet)
    
    # Water volume savings across plot acreage (m^3)
    water_saved_m3 = (evap_penalty_mm / 1000.0) * (request.plot_size_acres * 4046.86)
    
    # Pumping hours saved
    hours_saved = water_saved_m3 / (discharge_lps * 0.001 * 3600)
    
    # Financial electricity / diesel savings
    cost_saved_inr = hours_saved * (request.pump_hp * 0.746) * ELECTRICITY_TARIFF_INR
    
    return {
        "district": request.district,
        "hours_saved": round(hours_saved, 1),
        "water_saved_liters": int(water_saved_m3 * 1000),
        "cost_saved_inr": round(cost_saved_inr, 2),
        "statutory_compliant": request.transplant_date >= district.notified_cutoff_date
    }


📱 5. Designing for Rural Usability

Field technology must overcome digital literacy barriers to achieve real-world adoption:
  1. Bilingual Gurmukhi & English UI: Key metrics and advisories are rendered simultaneously in Punjabi (ਗੁਰਮੁਖੀ) and English.
  1. Text-to-Speech Voice Synthesizer: Farmers can listen to their customized field advisory spoken aloud in natural Punjabi or Hindi by pressing a single button.
  1. Village WhatsApp Sharing: Pre-formatted text cards can be forwarded with one click into local farmer groups.
  1. Feature Phone SMS Format: 160-character cellular SMS summaries for areas with limited mobile data connectivity.


🚀 6. Verification & Test Suite

The system includes automated tests validating statutory compliance logic, AWD irrigation reduction thresholds, and numerical edge cases:
Bash
cd backend
uv run pytest -v
  • 8 out of 8 tests passing (health probe, district extraction, statutory act compliance, AWD calculations, and boundary edge cases).


👥 7. Team Contributions

  • Member 2 (Frontend & GIS Engineering Lead): Developed the  React 19  single-page interface, built the  Leaflet  geospatial district mapping system with official GeoJSON boundaries, and engineered the omnichannel delivery systems (Punjabi voice synthesis, WhatsApp, and SMS).


🌾 8. Conclusion

By connecting open agronomic science from  Punjab Agricultural University (PAU)  with scalable, cost-effective serverless architecture on  Amazon Web Services (AWS) , JalFaasla proves that complex groundwater conservation policies can be translated into clear, actionable, and financially compelling insights for farmers.
# serverless# aws-lambda# python# web-development# architecture
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article