Kelp Canopy Extent - Calvert Island - 2006-2016

Kelp extent has been manually digitized in ARC GIS using air photos / UAV imagery. Vector polygons delineate the boundaries of the kelp canopy at the surface of the water. All imagery is sourced from late July to August to reflect peak kelp extent for the season. Image resolution used to digitize the kelp canopy ranges from 25 cm - 10 cm. Attribute information recorded: kelp species, kelp bed size (square meters), kelp species latin name, kelp density, result confidence, and credit. For a full breakdown of attribute information see lower in the abstract. For 2012 and 2014 an additional metric was calculated to indicate the actual density of the kelp bed. A percentage of coverage was developed to determine the true density of kelp in each bed (see full methodology below).

Dates: 2006, 2012, 2014, 2015, & 2016.

This methodology outlines the procedure for obtaining and processing high-resolution imagery of the canopy-forming kelp beds in the Northwest Calvert Island area. It also includes the procedure for data analysis and metrics of interest to be monitored over time.

Aerial Survey Area The survey for annual kelp bed mapping extends south from Starfish Island in Choked Pass to 7th Beach. Northernmost latitude: 51.6819 Southern Most latitude: 51.6363.

Survey Timing, Imaging, and Logistic Considerations: Survey timing protocols are framed by the goal of obtaining imagery that best represents the maximum extent of the canopy-forming kelps. Based on optimizing the timing of maximum canopy development, tide level, and light conditions, we select acceptable “survey windows” at the beginning of each season. Firstly, survey windows occur during the period of maximum canopy extent, generally considered to be between August and October in British Columbia. Secondly, survey windows are determined by the timing of low tide cycles, such that all imagery is captured at tide heights lower than +1.0 m MLW. Lastly, the survey is conducted during the first available window that has acceptable environmental conditions: calm seas (sea state less than 1.5 m swell), clear visibility (no fog or low lying cloud), and low winds (less than 15 knots). If possible, surveys are conducted during time windows that minimize glare. In 2014 and 2015, the aerial surveys have been conducted in the first low tide sequence in August.

Imagery Capture: Surveys are conducted using DJI Phantom 3 professional Unmanned Aerial Vehicles (UAVs or drones) equipped with a commercial grade 4k RGB camera to capture imagery. The UAV and camera are operated by using the DJI Capture software application connected to a tablet or smartphone. Imagery is captured at a flying altitude of between 250 – 300 m, allowing an area of 1.5 x 1.5 km to be covred in one flight. Images are taken to overlap at a minimum of 50% to ensure adequate coverage and the ability to select out images of poor quality. Multiple flights are taken during a single “capture window” (see above for specific conditions) with up to 6 flights possible with one UAV. UAV flights are flown from a zodiac or small research vessel so the UAV can move from coverage areas quickly. Flights range from 12 – 17 minutes depending on conditions and are manually flown in a grid formation to cover the study area. Various resolutions are possible depending on flight height. Low flights ( 10 plants per 10m2) or ‘low’ (< 10 plants/fronds per 10m2) density. The high and low density classification is based upon the B.C. Province’s Kelp Inventory Method (Foreman 1975) and has been used in all Provincial kelp surveys to provide a more accurate estimate of total kelp biomass. While mapping kelp polygons, it may be necessary to split kelp beds if densities change within neighbouring patches. For example, if a kelp bed is relatively dense, but has a larger patch of low density plants at the outer edge, map the ‘high density’ bed and the ‘low density’ bed as separate polygons, despite the fact that they are the same contiguous kelp bed. Once a single kelp polygon is complete, fill in the attribute table with the appropriate data being as consistent as possible. List the kelp species (Nereocystis, Macrocystis, or mixed), mapper initials, the estimated bed density, and the level of confidence (high, medium, low) associated with either the kelp bed extent, species identification, or the density classification. The ‘confidence’ attribute provides an uncertainty measure, as kelp bed mapping can be challenging due to reflectance, debris, glare, and shadows. Once the classification of all kelp beds is complete, the dataset is checked by a second person to ensure quality control. With the kelp polygons finalized, use ArcMap area calculation tool to calculate the area of each kelp polygon and populate the attribute table with this information under the column “bed area.”

Kelp Percent Cover and Density (optional for further analysis) - currently used for 2012 + 2014 data: The aforementioned methods provide a dataset to examine changes in the spatial extent of canopy kelps. The following methods can be applied to obtain estimates of standing stock biomass and are based upon the Provincial Kelp Inventory Methodology KIM-1 (Foreman 1975, Sutherland et al. 2007). We are currently in the process of applying these methods to the Hakai Calvert Island kelp imagery, but work is still underway and methods are not finalized.

Kelp density / species / and extent are recorded - see attributes listed below. Kelp canopy extent represents the maximum growth for the year. Tide level was roughly 2.47 meters (8.1 foot) - a level which could impact kelp extent and the ability for GIS technicians to properly detect and identify species. More research is being done to validate and test tide levels with associated kelp extent findings.

All digitizing occurred using Arc GIS. SFU volunteer crew (Amanda Schrack and Sebastian Mather) along with principal investigators Jenn Burt and Keith Holmes.

Methods based on KIM 1 methodology as described in Sutherland 2007 report:

Areas covered: West Beach – 7th beach to West Beach + Choked Passage – Surf Pass and North Beach to Wolf Beach North to Hakai Land and Sea Beach.


FID - Unique identification number - ARC auto generated during polygon creation

Shape - Polygon - spatial representation Species - "Macro" / "Nereo" / Mixed detailing the two possible kelp canopy species. Macro refers to macrocystis pyrifera or giant kelp. Nereo refers to nereocystis luetkeana or bull kelp. Mixed is a homogenous mix of both Macro and Nereo species. Defined by GIS techs - identified by "credit" field.

Density - density of kelp defined as either high or low. Based on the Sutherland methodology for kelp density. High = high density / based on mean kelp plants found per square meter. Low = low density - still a continuous bed of kelp with large gaps the length of each fond.

Area sq_m - size of each kelp bed in square meters. Defined by arc geometry calculation.

Concat - concatenation of species and density information. Used for cartographic purposes. Defined by concatenating species and density metrics in ARC GIS.

Confidence - Used by GIS technician to identify difficult to detect species / kelp bed extent / density. Used to inform reviewers and quality control spatial information. High = High confidence - no need to check over, Medium = medium confidence - check over, Low = low confidence - unsure of major elements - reviewer check over. A "Checked" may be added to the information to determine quality control has been completed on low confidence polygons. Determined by the last "credit" reviewer.

Credit - Initials of GIS technician to create and give the final quality check on the kelp extent polygon. KH = Keith Holmes, JB = Jenn Burt, AS = Amanda Schrack, SM = Sebastian Mather.

Percent cover: Kelp bed percent cover is not calculated in the B.C. Province kelp inventories, but a similar approach is used in aerial kelp monitoring conducted by the Washington State department (cite ref here). We are exploring the use of this method for future monitoring purposes. The % cover measure is defined as the proportion of a kelp bed that is visible kelp stipe/blade/frond tissue, as opposed to visible sea surface water. For example, a low density bed would have a low % cover and high density bed would have a high % cover. Calculation of % cover for each kelp bed is done using a combination of ArcMap tools and ENVI software. First, the aerial image file is clipped to create an image mask of the kelp bed polygons within a given sample year. The clipped kelp bed imagery is then reclassified using ENVI software to determine the appropriate thresholds to distinguish “kelp tissue” from water surface. This supervised classification is performed by sampling throughout the mapping region. Depending on the quality of the image, classification may occur within one large sample region, or may need to be split into “exposure categories” if there are areas of the imagery with variable exposures or poor spectral resolution (e.g. in the 2014 kelp mapping dataset, imagery was split into 3 exposure categories, which greatly improved the percent cover output). ENVI classification results in a raster dataset with a binary classification of 1 = Kelp, 0 = water. This raster file is then converted to vector file, and if necessary, merged in ESRI ARC desktop with the other classified datasets within the mapping region. Using the new vector file, the total area of “kelp tissue” within a bed (m2) and the % cover for each bed (kelp tissue area divided by total bed area) is calculated using summary statistics tools and a spatial join. The “kelp tissue area” and “% cover” values are added to the attribute table. Visual inspection should confirm that the classification of % cover is successful in most kelp beds. In some beds where image quality and processing caused poor classification, quality control measures are manually corrected by editing the kelp tissue extents in ARC desktop.

Species_la - full Latin species name for the kelp. macrocystis pyrifera / nereocystis luetkeana / mixed. Mixed has a homogeneous mix of both macrocystis pyrifera and nereocystis luetkeana species

Dataset created by Jenn Burt and Keith Holmes. Simon Fraser University Volunteers: Amanda Schrack and Sebastian Mather, and Andrea Chee. Contact:

Access and Use

Licence: Appropriate credit must be given to Hakai Institute and the authors of the dataset.

Data and Resources


Metadata Created October 26, 2018, 22:48 (UTC)
Metadata Updated October 26, 2018, 22:48 (UTC)
Reference Date(s) 2017-01-01 (Creation)
2017-01-01 (Publication)
Frequency of Update
Metadata Date October 16, 2018, 23:10 (UTC)

Graphic Preview

Dataset extent

Map data © OpenStreetMap contributors

Additional Info

Field Value
Contact Email
encoding utf8
metadata-language eng
progress completed
resource-type dataset
Responsible Party Hakai Institute (Point of Contact, Processor)
spatial-reference-system 26909