Table of Contents
What multispectral drone surveys actually measure
A multispectral drone carries a specialised camera that records reflectance in several discrete wavelength bands at once, rather than the three broad bands (red, green, blue) of a normal camera. Typical sensors capture five or six bands covering green, red, red-edge, and near-infrared (NIR), sometimes with an additional blue band and a downwelling light sensor for irradiance correction. Because healthy plant tissue absorbs red light strongly and reflects near-infrared strongly, the ratio between those bands is a direct proxy for chlorophyll activity and canopy vigour.
The most widely quoted output from this data is the Normalised Difference Vegetation Index, or NDVI, calculated as (NIR − Red) / (NIR + Red). NDVI values range from −1 to +1, with bare soil and water clustering near zero, stressed vegetation sitting around 0.2 to 0.4, and dense, healthy canopy reaching 0.7 to 0.9. NDVI is the most common vegetation index because it is simple, robust, and well understood, but it is not the only one we produce. For specific problems we also calculate GNDVI, NDRE, SAVI, LCI and MCARI, each of which emphasises a different aspect of canopy health. A red-edge-based index like NDRE, for example, becomes more useful than NDVI once a crop closes canopy and NDVI saturates.
Common use cases in Ireland
Precision agriculture and tillage
For tillage farms and progressive dairy operations, we use multispectral imagery to identify within-field variability before visible symptoms appear. A typical deliverable is a variable-rate nitrogen application map handed to the contractor as a shape file compatible with John Deere Operations Center, Trimble Ag, or CNH AFS Connect. Early-season NDVI surveys highlight poor establishment caused by compaction or drainage issues; mid-season NDRE surveys guide top-dressing; pre-harvest surveys support yield forecasting and lodging assessment.
Forestry health and disease mapping
Multispectral data is increasingly used by foresters and ecological consultants to map Phytophthora, Dothistroma needle blight, and wind-throw recovery in commercial plantations. Paired with a drone LiDAR survey for canopy height, multispectral imagery pinpoints individual stressed trees long before a walking inspection would catch them. For more on how this integrates into wider woodland mapping, see our drone forestry survey service.
Environmental monitoring and EIA
Multispectral imagery feeds into Environmental Impact Assessments, Natura 2000 habitat mapping, peatland restoration monitoring, and riparian buffer condition scoring. Because the data is quantitative and repeatable, it gives ecologists a defensible baseline that can be surveyed again at the same spatial resolution in future years. We have flown multispectral over raised bogs, reed beds, coastal dune systems, and constructed wetlands, and the same data supports both the habitat map and the water-quality narrative in one deliverable.
Construction and civils
On civil engineering projects the most common use is vegetation clearance verification and reinstatement monitoring after works. A pre-clearance NDVI survey records what was there; a post-works survey records what has been replaced or reseeded; and a follow-up survey at 12 and 24 months supports any reinstatement bond release. This is particularly useful on linear infrastructure projects such as gas mains, power lines, and greenway construction where the reinstatement area is long and narrow.
The hardware we use
For multispectral work we fly a MicaSense RedEdge-P and a MicaSense Altum-PT, both mounted on our DJI Matrice 300 RTK airframe. The RedEdge-P records five narrow bands (blue, green, red, red-edge, NIR) plus a 5 megapixel panchromatic band for high-resolution sharpening. The Altum-PT adds a radiometric thermal band, which lets us capture both multispectral reflectance and canopy temperature in a single flight – useful for irrigation stress work and roof-survey thermal anomalies. Both cameras use an integrated downwelling light sensor that corrects for changing cloud conditions during the flight, and both are individually calibrated against a known reflectance panel before each mission.
Ground resolution depends on flight altitude. At 60 metres above ground level we typically achieve a ground sample distance (GSD) of around 4 cm per pixel with the RedEdge-P. For standard agronomic work we fly higher (90 to 120 m AGL) to cover more ground per flight and deliver 6 to 8 cm GSD, which is more than sufficient for vigour mapping. For disease detection in orchards or intensive horticulture we fly lower to push the GSD below 3 cm.
The workflow from flight to deliverable
Stage 1: Mission planning and calibration
Every multispectral mission starts with a reflectance panel capture at the launch point and, for long flights, a second capture at the landing point. We plan the flight in either DJI Pilot 2 or DroneDeploy, set overlap at 80 percent forward and 75 percent side, and lock the flight speed to the sensor trigger rate so we get consistent image spacing. For repeat-visit monitoring work we save the mission file and reuse the identical flight plan on every visit to ensure temporal comparisons are genuine, not an artefact of different lines.
Stage 2: Radiometric calibration and mosaicking
Back in the office, images are processed in Pix4Dfields or Agisoft Metashape Professional. Radiometric calibration converts raw digital numbers to reflectance values using the panel captures, the downwelling light sensor logs, and the per-camera calibration coefficients. Once calibrated, the imagery is mosaicked into band-separated orthomosaics, and vegetation indices are calculated on a per-pixel basis.
Stage 3: Index calculation and zoning
From the calibrated orthomosaics we generate NDVI, NDRE, GNDVI and any other index the client has specified. We then create management zones either by equal-area percentile splits (typically 3 or 5 zones) or by natural breaks on the NDVI histogram. Zones are exported as shape files in Irish Transverse Mercator (ITM, EPSG:2157) and, where the end user is a farmer or contractor, also in WGS84 for direct import into the tractor terminal.
Stage 4: Reporting and interpretation
The raw index maps are useful, but what clients actually want is an interpretation – a short written note explaining what the map shows, where the anomalies are, and what ground-truthing we recommend. We provide a concise PDF report alongside the GIS data, and for ongoing monitoring contracts we include a time-series chart showing how index values for each zone have evolved between visits.
Accuracy, limits, and what not to expect
Multispectral drone surveys are quantitative, but they are not laboratory measurements. Three limits are worth understanding before commissioning a survey.
First, reflectance calibration is only as good as the panel capture. If the panel is dirty, shaded, or captured under very different illumination than the flight, absolute reflectance values will drift. This matters for cross-site or cross-year comparisons. For within-flight comparisons (which zone of this field is greener than that zone) the impact is negligible.
Second, NDVI saturates. Once a crop or canopy closes and LAI (leaf area index) exceeds about 3, NDVI plateaus around 0.85 and stops discriminating between healthy and very healthy vegetation. Red-edge indices (NDRE) and chlorophyll indices (LCI, MCARI) continue to respond in this range and should be used instead.
Third, multispectral data is not a substitute for ground truthing. We always recommend that important management decisions – especially disease treatment or variable-rate nutrient application – are calibrated against at least a handful of ground observations within the first survey season. The drone data then accelerates and scales what the agronomist or forester is already doing on the ground.
What we deliver
A standard multispectral survey deliverable includes: calibrated reflectance orthomosaics for each band (GeoTIFF, 16-bit), NDVI and NDRE orthomosaics (GeoTIFF, float), RGB orthomosaic for visual reference, management zone polygons (shape file, ITM and WGS84), a short interpretation PDF, and the raw imagery archived on our secure storage for a minimum of twelve months. Where the project requires specific formats – ISO-XML for precision agriculture controllers, for example, or a file geodatabase for ArcGIS Pro – we produce those on request at no extra cost. The turnaround from flight day to delivered data is typically 3 to 5 working days for a single site.
For more on how multispectral fits alongside our other surveying and mapping services, see the multispectral mapping service page, our drone surveying and mapping overview, and aerial environmental monitoring. If you need canopy height data alongside the spectral information, our drone LiDAR surveys are normally flown on the same visit to avoid mobilising twice.
Frequently Asked Questions
What is the difference between a normal drone photo and multispectral imagery?
A standard drone camera records three broad bands (red, green, blue) that are optimised for looking natural to a human eye. A multispectral camera records several narrow bands – typically including red-edge and near-infrared – that are invisible to the eye but are diagnostic of plant health. Normal imagery tells you what a field looks like; multispectral imagery tells you how the plants in it are performing.
Can you fly multispectral surveys in Irish weather?
Yes, within sensible limits. Multispectral drones need consistent lighting, so we avoid flying in moving cloud shadow where possible. Light overcast is actually excellent because the illumination is even and diffuse. Rain, fog, and strong winds above 10 m/s will stop a flight. Over the course of a year we typically have no difficulty booking suitable flight windows in Ireland, though we plan jobs with a 48 to 72 hour weather buffer.
How often should I survey a site?
It depends on what you are monitoring. Precision agriculture typically wants three to five surveys per growing season – establishment, tillering or early growth, main crop, and pre-harvest. Forestry health mapping is usually annual, timed to peak canopy. Environmental baseline and reinstatement work is usually annual or bi-annual. We are happy to design a survey schedule around your objectives rather than selling a fixed package.
Is drone NDVI as accurate as satellite NDVI?
Drone NDVI has roughly 100 to 1,000 times higher spatial resolution than the best free satellite sources (Sentinel-2 at 10 m, Landsat at 30 m), so individual plants, rows, and small anomalies become visible that satellite data simply cannot see. In exchange, satellite NDVI covers the whole country every few days and is free. The two data sources complement each other: we often use Sentinel-2 for field selection and trend monitoring, then deploy the drone for diagnosis when the satellite flags a problem.
Can multispectral data be used for yield prediction?
Yes, with caveats. Mid-to-late-season NDVI, NDRE, and GNDVI have well-documented correlations with final yield for most arable crops. The correlation coefficient depends strongly on the crop, the season, and the accuracy of the calibration points. For a single-field or single-season forecast, drone NDVI alone is a useful relative indicator. For an absolute yield estimate in tonnes per hectare we recommend calibrating against yield-monitor data from previous seasons on the same field.
Are you a DAA-approved drone operator?
Yes. Drone Services Ireland is a fully licensed and insured Specific Category drone operator under the Irish Aviation Authority, holding the relevant authorisations for the altitudes, airframes, and payloads we fly, including the MicaSense RedEdge-P and Altum-PT multispectral sensors on the DJI Matrice 300 RTK. Every flight is logged and insured to €5 million public liability as standard.
Related guides
For survey teams: For wider context on combining multispectral with traditional survey methods, see our survey-supplementation guide.