Tips To Improve Your Vegetation Mapping Using Drones
Vegetation mapping formulates one of the key pieces of information within environmental and land management planning. Imagery for vegetation maps typically come from two sources, satellites and manned aircraft, with additional data gathered by qualified, on ground engineers, botanists or ecologists. The development of more efficient, cost effective and precise methods of imagery and data capture is an important technical issue in environmental and land management planning.
Drones are employed in a wide range on information gathering activities and currently service local to landscape scale areas where satellites and manned aircraft experience severe limitations. Delivery of this service comes at a lower operational cost, compared with the classical manned aerial photogrammetry (Berni et. al. 2009, Dunford et. al. 2013, Eisenbeiss 2004). Sensors placed on drones are capable of combining high spatial and temporal resolutions with quick turnaround times providing low-cost approaches to meet the critical requirement of spatial, spectral, and temporal resolution (Berni et. al. 2009, Korehisa & Seiich (2014).
Identification, analysis and mapping of vegetation by drone imagery can generate detailed maps of vegetation assemblages at the species level through the identification of individual plant species. This requires roughly resolutions of <1cm per pixel to be able to identify with confidence the identity of plants, however with will change depending on project objectives. Satellites and manned aircraft are incapable of capturing data at these resolutions, leaving a niche where drones are positioned well.
The identification process does have limitations. When supported by ground truthing activities, confidence in identifications can improve, but examples of lookalike species with similar growth habit and spectral reflectance signatures do occur.
Here are some useful tips on how to reduce the occurrence of misidentifications and improve your vegetation surveys/maps when using drones.
1. Know your objective.
Mapping ecological vegetation classes (EVC) or vegetation assemblages at the species level have two very different survey methods and are conducted on different spatial scales. Clearly understand what data you are trying to capture and plan missions accordingly.
2. Get a qualified and experienced professional to review the data.
Having a professional who understands and has experience with the assemblage of plants in the EVC of the survey area will help improve confidence in the identification process. They should have the knowledge of the seasonal colour changes, flouring systems, growth habits and other ecological parameters essential in distinguishing between species.
3. Conduct thorough ground truthing activities.
Ground truthing will verify the spectral signature of individual species. A quick way to do this is to take geolocated images of the dominant and target plant species. This will ensure you have the location of know plant to refer against the vegetation map. Make sure you have multiple samples of each species to ensure you have a representative sample.
4. Capture at appropriate resolutions
Make sure you are capturing your data at a resolution that meets the needs of the objective. Understand that the generation of orthomosaics can often reduce the end output resolution. You may need to compensate for this. Capturing test images at a range of altitudes can assist in finding the minimum resolution threshold for your application.
5. Choose your season of survey
Flowering and seasonal growth cycles can be very useful in identifying plants. Choosing a time of year when your target species or their lookalikes are in flower or losing its leaves is an easy way of improving the confidence in your vegetation maps. Use the ecology of the plants to your advantage. This may not always be possible as synchronising project, weather and ecological cycles can be difficult.