Now that your data has been collected and stored, you are ready to create outputs from your drone survey. Using the steps below will assist you in creating orthomosaics from your drone imagery and how to run those orthomosaics through a state-of-the-art machine learning tool which automates species-level kelp canopy mapping.
Stage outcomes
At this stage of the workflow the imagery collected by drones is used to derive kelp delineation. The outcomes from this stage include:
A georeferenced orthomosaic (.tif file) generated from the processed drone imagery
A canopy kelp extent polygon (.shp file) from the orthomosaic that contains species-level classification of surface floating canopy kelp present in your drone imagery
Step 1 - Generate an orthomosaic
Outputs from a structure-from-motion process used to generate an orthomosaic.
An orthomosaic is created by combining many overlapping images and removing the distortions to create a map. It is this map from which you can determine kelp canopy extent. Create a high-quality orthomosaic from drone imagery by following these steps:
Selecting appropriate structure-from-motion (SfM) software to use.
Structure-from-motion (SfM) software is used to generate a single orthomosaic from the hundreds of images that a drone takes during a survey. There are SfM softwares that can be purchased as standalone licenses (e.g. Agisoft Metashape, Pix4D Mapper) or as online subscription-based services (e.g. DroneDeploy). Which software you choose to use may be based on:
What SfM software is already available to you or your organization
Cost
Local vs cloud processing needs
Use the structure-from-motion software to generate an orthomosaic.
Review your orthomosaic by loading it in a GIS and ensuring that it meets expected quality standards (e.g. all areas are accurately represented, no unexpected gaps). You may need to recreate the orthomosaic using the SfM software if some images were not included, or gaps exist.
Export your orthomosaic to the Working directory for your project.
Make sure you are using consistent file naming and structure for both your orthomosaics and your outputs. The “Data Storage” section of the Coastal Habitat Mapping Using Remotely Piloted Aerial Systems (RPAS) offers an example of a file naming and organizational structure.
Step 2 - Georeference your orthomosaic
Survey equipment to collect ground control points on a rocky shore.
For time series analysis, it is important that orthomosaics are aligned accurately in space. New orthomosaics should be georeferenced to existing imagery for the area when such data exists. An existing survey with ground control points can be used as reference, or imagery can be aligned to a basemap or other source, using a GIS. Georeferenced orthomosaics should be stored appropriately.
ArcMap and QGIS provide instructions on how to georeference your orthomosaic within that GIS. Once the orthomosaic is georeferenced, it is recommended that the file (.tif) is stored appropriately using consistent file naming conventions and structure. The “Data Storage” section of the Coastal Habitat Mapping Using Remotely Piloted Aerial Systems (RPAS) guide provides an example of a naming structure.
Step 3 - Delineate kelp extent from the orthomosaic using Habitat Mapper
Orthomosaic from a kelp survey with kelp canopy extent detected from Habitat Mapper shown in orange.
The Habitat Mapper is a machine learning tool that automates species-level detection of surface canopy kelp in your orthomosaic. It converts your drone orthomosaic (.tif) into a polygon shapefile data that provides species-level classification of where emergent canopy kelp is present in the orthomosaic. It is open-source and freely available for anyone to use.
The steps to delineate kelp extent from your orthomosaic include:
Installing and running Habitat Mapper.
Install and run the Habitat Mapper using the Beginner Guide. The Habitat Mapper can be run on any computer but consider running it on a computer with a good GPU.
Review Habitat Mapper outputs.
The output of the Habitat Mapper is a polygon shapefile that can be opened in a GIS, where both extent and species accuracy can be modified. This could include manually removing non-kelp polygons, manually digitizing kelp canopy, and fixing any misclassified species. See the Post-Processing section in the Beginner Guide for more information on how to review and edit the outputs from the Habitat Mapper.
Step 4 - Create and store analysis ready data
Folder containing list of analysis ready data from kelp drone surveys.
Finalize the polygon with kelp delineation data, and make it analysis-ready by ensuring that you are comparing the same area over time and store your data. Be sure to store and name your files appropriately.
Create your Area of Interest (AOI) for your study area.
In a GIS, generate a polygon vector feature which you will “clip” all of your data to - we refer to this as an Area of Interest (AOI). This feature should be a region that is consistent across all years of surveys.
Extract kelp area to your AOI.
Clip the kelp extent output from your survey with the AOI.
Copy kelp extent to a data sheet.
Extract the kelp canopy area extent values and put them into a data sheet (.csv) where you are compiling your annual kelp extent values.
Export your final dataset.
Export the data as a shapefile or gdb feature class and give it an appropriate name and attributes.
The file name should follow the same naming convention as the orthomosaic. e.g. YYYYMMDD_Region_Site_kelp.shp
Network storage used by the Hakai Institute (don’t worry - you probably don’t need one of these but good storage is important!).
Maintaining reliable access to your data ensures long-term usability, transparency, and reproducibility. Make sure your processed data, visualizations, and any derived products (e.g. orthomosaics, reports) are securely stored and can be easily downloaded when needed.
Data storage best practices include:
Maintain backups in multiple locations, consider the 3-2-1 backup rule
Preserve original and processed datasets or layers
Include data summaries and rich metadata
Document data dependencies
Resources
Easily access all the resources used by the steps in this stage.
Habitat Mapper
Tools
Automated kelp detection and quantification tool for processing drone imagery and satellite data.
