Capability of the CANAPI algorithm to derive shrub structural parameters from satellite imagery in the Alaskan Arctic

• Published in: Polar Record
• Main Authors: Duchesne, Rocio R., Chopping, Mark J., Tape, Ken D.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Cambridge University Press (CUP) 2015
• Subjects: General Earth and Planetary Sciences; Ecology; Geography, Planning and Development; Arctic; Polar Record; Tundra
• Online Access: http://dx.doi.org/10.1017/s0032247415000509; https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0032247415000509

ABSTRACT The observed greening of Arctic vegetation and the expansion of shrubs in the last few decades probably have profound implications for the tundra ecosystem, including feedbacks to climate. Uncertainty surrounding this vegetation shift and its implications calls for monitoring of vegetation structural parameters, such as fractional cover of shrubs. In this study, CANAPI, a semi-automated image interpretation algorithm that identifies and traces crowns by locating its crescent-shaped sunlit portion, was evaluated for its ability to derive structural data for tall (> 0.5 m) shrubs in the Arctic. CANAPI estimates of shrub canopy parameters were obtained from high-resolution imagery at 26 sites (250 m x 250 m each) by adjusting the algorithm's parameters and filter settings for each site, such that the number of crowns delineated by CANAPI roughly matched those observed in the high-resolution imagery. The CANAPI estimates were then compared with field measurements to evaluate the algorithm's performance. CANAPI successfully retrieved fractional cover (R 2 = 0.83, P < 0.001), mean crown radius (R 2 = 0.81, P < 0.001), and total number of shrubs (R 2 = 0.54, P < 0.001). CANAPI performed best in sparse vegetation where shrub canopies were distinct, while it tended to underestimate shrub cover where shrubs were clustered. The CANAPI algorithm and the regression equations presented here can be used in Arctic tundra environments to derive vegetation parameters from any sub-meter panchromatic imagery.