Precision Agriculture, Sensors & Drones

Managing each part of the field according to its needs. Learn the precision agriculture cycle (sense–analyse–decide–act), GNSS/RTK positioning and auto-steering, yield monitoring and mapping, soil sampling and management zones, remote sensing and vegetation indices (NDVI, NDRE), drones for mapping and spraying, IoT soil and weather sensors, variable-rate technology for fertiliser, seed and water, farm management information systems, AI for disease detection and yield prediction, and data economics — with Python for NDVI zoning and variable-rate nitrogen.

Fields are not uniform: soil, water and crop vigour vary within metres. Precision agriculture uses positioning, sensing and data to apply inputs where and when needed — saving fertiliser, chemicals, seed and water while raising yields.

Drones, satellites (Sentinel-2, Landsat), smartphones and cheap IoT sensors are bringing precision tools within reach of smallholders, often via service providers and FPOs.

An analogy

Traditional farming treats the whole field like one patient getting the same medicine. Precision agriculture is like personalised medicine: diagnose each zone (sensing), then prescribe the right dose (variable-rate application).

The precision agriculture cycle

Sense (soil sampling, yield monitors, imagery, sensors) → analyse (maps, zones, models) → decide (prescriptions) → act (variable-rate machines, targeted spraying) → evaluate (yield maps, profit). Positioning by GNSS ties all data to locations; RTK corrections give centimetre accuracy for auto-steer and controlled traffic.

Remote sensing and drones

Platforms and uses:

Sensors and IoT

In-field sensors: soil moisture (capacitance, tensiometers), soil EC mapping (Veris, EM38), weather stations, leaf wetness (disease risk), canopy sensors (GreenSeeker, Crop Circle) for on-the-go N application. LoRaWAN/NB-IoT networks and solar power make low-cost deployments possible; data flow into farm management systems and advisory apps.

Data science and decision-making

Machine learning detects diseases from leaf images, predicts yield from imagery and weather, and optimises irrigation. Management zones from multi-year yield, EC and imagery (k-means clustering). Challenges: data ownership, interoperability (ISOBUS, ADAPT), connectivity, cost for small farms, and proving ROI. Digital public infrastructure (e.g. AgriStack, soil health cards) aims to scale services.

Formulas

Worked example: NDVI

Given: NIR 0.42, Red 0.07

Find: NDVI.

Answer: 0.71

Worked example: Drone spraying

Given: Swath 6 m, 4 m/s, efficiency 0.7

Find: ha/h.

Answer: 6.0 ha/h

Worked example: Input saving

Given: VR saves 100 kg N on 10 ha, urea (46% N) ₹5.4/kg

Find: Urea saving.

Answer: ≈ ₹1170 (subsidised price — larger at unsubsidised cost)

NDVI zoning and variable-rate nitrogen

# Precision agriculture: NDVI and variable-rate nitrogen
pixels = [(0.45, 0.08), (0.38, 0.10), (0.25, 0.12), (0.50, 0.06)]    # (NIR, Red) reflectance
for nir, red in pixels:
    ndvi = (nir - red) / (nir + red)
    zone = "high vigour" if ndvi > 0.6 else "medium" if ndvi > 0.4 else "low vigour"
    print(f"NIR {nir:.2f}, Red {red:.2f} -> NDVI {ndvi:.2f} ({zone})")
# Variable-rate N: base 120 kg/ha scaled by vigour (more N to medium zones)
rates = {"high vigour": 100, "medium": 130, "low vigour": 110}
areas = {"high vigour": 6.0, "medium": 3.0, "low vigour": 1.0}
vr = sum(rates[z] * areas[z] for z in rates)
print(f"variable-rate N {vr:.0f} kg vs uniform {120*sum(areas.values()):.0f} kg for 10 ha")
# Drone spraying coverage
swath, speed_ms = 5.0, 5.0
print(f"drone coverage {swath*speed_ms*3600*0.7/10000:.1f} ha/h (70% efficiency)")
# NIR 0.45, Red 0.08 -> NDVI 0.70 (high vigour)
# NIR 0.38, Red 0.10 -> NDVI 0.58 (medium)
# NIR 0.25, Red 0.12 -> NDVI 0.35 (low vigour)
# NIR 0.50, Red 0.06 -> NDVI 0.79 (high vigour)
# variable-rate N 1100 kg vs uniform 1200 kg for 10 ha
# drone coverage 6.3 ha/h (70% efficiency)

Healthy leaves reflect strongly in near-infrared and absorb red light, so NDVI rises with vigour. Zoning the field lets the applicator give more N where the crop can respond and less where it's already vigorous (or limited by other factors) — here saving 100 kg N on 10 ha. Low-vigour zones should be ground-truthed: the cause may be water, pests or soil, not nitrogen.

Common mistakes

Applications

Key takeaways

Going further