Potatoes and Drones

Agriculture
1

Hello,
We grow potatoes on 170 hectares of land at our farm. We are starting to use a DJI Mavic 3M drone with Pix4Dfields. I would very much appreciate some practical advice and insights on detecting mold and other diseases in potatoes.

Thank you

2

Hi Josef,

We are glad you chose our solution to monitor your field. Here is practical advice on how to get the most out of your hardware and software to detect these diseases early.

Why Your Multispectral Camera Matters

Potato blight fundamentally alters the physiological structure of the plant’s leaves, specifically attacking chlorophyll and cell walls before causing the visible brown/black necrotic lesions.

The M3M has four multispectral bands (Green, Red, Red Edge, Near-Infrared). For detecting potato blight, the Red Edge and Near-Infrared (NIR) bands are your most valuable tools. The Red Edge band is highly sensitive to subtle changes in chlorophyll content, meaning it can detect stress days before the human eye (or a standard RGB camera) can spot the leaves browning.

1. Flight Best Practices for Potatoes

To get actionable data in Pix4Dfields, your raw imagery needs to be flawless. Potatoes have complex, overlapping canopies that can be difficult for photogrammetry software to stitch together if you don’t fly correctly.

  • Optimal Altitude: Fly between 45 to 60 meters (150-200 feet) above ground level (AGL). Going higher reduces your resolution (GSD), making it harder to spot small pockets of early infection.

  • High Overlap: Potato canopies look very homogeneous from above. Set your flight parameters to at least 75% frontal overlap and 70% side overlap to ensure Pix4Dfields can stitch the map without artifacts.

  • Lighting Conditions: Try to fly around solar noon (when the sun is highest) to minimize shadows cast by the potato ridges. Overcast, diffuse light is actually excellent for multispectral mapping, provided the lighting stays consistent throughout the flight.

  • RTK is Crucial: If your M3M has the RTK module active (via a base station or network), use it. Centimeter-level accuracy ensures that when you generate a spot-spraying map, your tractor or spray drone hits the exact infected rows.

2. Processing in Pix4Dfields: The Right Indices

Pix4Dfields is designed for fast, offline processing directly at the edge of the field. Once your map is stitched, you need to generate vegetation indices. Do not just rely on standard NDVI.

Use NDRE for Early Detection

NDVI (Normalized Difference Vegetation Index) is great for measuring overall biomass, but it “saturates” easily in dense, healthy potato canopies. Instead, use NDRE (Normalized Difference Red Edge). Because it uses the Red Edge band instead of the standard Red band, NDRE penetrates deeper into the canopy and is much more sensitive to early-stage chlorophyll degradation caused by blight.

Use NDVI for Damage Assessment

Once a late blight infection has caused defoliation and visible necrosis, NDVI becomes very useful for assessing the total extent of the damage and mapping areas where the canopy has collapsed.

3. The Golden Rule: Ground Truthing

This is the most critical insight: Your drone cannot diagnose a specific disease; it can only diagnose stress.

When you generate your NDRE map in Pix4Dfields, the software will highlight areas of low vigor (usually in red or yellow). However, a red patch on your map could be:

  • Late Blight or Early Blight.

  • Nematode damage.

  • Waterlogging from a broken irrigation pipe.

  • Nutrient deficiency.

The Workflow:

  1. Process the map in Pix4Dfields on your laptop.

  2. Identify the stressed zones (anomalies) on the map.

  3. Walk out to those specific GPS coordinates in the field.

  4. Inspect the leaves manually. Look for the classic signs of Late Blight: water-soaked spots, purplish-black lesions, and white fuzzy mold on the underside of the leaves.

4. Taking Action: Prescription Maps

If your ground truthing confirms an outbreak of mold/blight, you can use Pix4Dfields to act immediately:

  1. Zonation: Use the Zonation Tool in Pix4Dfields to divide your NDRE map into distinct management zones (e.g., Healthy, Moderate Stress, High Stress).

  2. Magic Tool / Spot Spraying: If the blight is highly localized, use the Magic Tool to select only the infected areas and a buffer zone around them.

  3. Export: Export this as a Variable Rate Application (VRA) prescription map (in shapefile or ISOXML format) directly to your tractor’s terminal or to a spray drone (like an Agras). This allows you to apply heavy fungicides exactly where needed, saving chemicals on the healthy parts of the field.