Using UAVs to create digital surface models for northern peatlands research

I just attended a fascinating presentation with the unprepossessing title of "CCRS/NRC/McGill Scientists - TOPICS OF PEATLANDS AND THE MER BLEUE BOG."  This was about a new project called MBASSS initiated by scientists from the Canada Centre for Remote Processing (CCRS), National Research Council (NRC), and McGill University to acquire airborne hyperspectral imagery and ground-based measurements for the purpose of validating satelllite optical earth observation (EO) multi-hyperspectral data products for northern/subarctic peatlands. 

The world has about about 3 million square km of northern boreal peatlands, which are huge carbon stores, estimated to contain 200 to 600 gigatonnes (Gt) of carbon.  About one million square km of these are in Canada.  Peatlands act as both a carbon source and a carbon sink in the terrestrial carbon cycle. The net effect of northern peatlands is a small to moderate carbon sink, removing an estimated 49 billion kg/yr (Tg/yr) of carbon from the atmosphere.

75% of these peatlands are bogs and one of these is Mer Bleue a 28 square km bog very near to Ottawa.  Since it is much easier to study a bog that is 13 km from Ottawa than to travel a thousand km to subarctic Canada, Mer Bleue, which already has a peatland observatory, was chosen for the study. 

Mer Bleue is normally a net sink for carbon, removing on average 60 grams of carbon per square meter per year (g C m-2 yr -1), but this can vary from year to year depending on factors such as water level.  Historically it has ranged from 10 to 128 g C m-2 yr -1.

Conventional overflights for the project are being conducted by the NRC using a fixed wind aircraft (Twin Beaver).  But as an alternative, Matt Maloney of CCRS has conducted trial flights with three different UAVs.  At the low end with a Phantom 2 equipped wiith a GoPro camera.  The other devices are a Spyder PX8 and an Aeromapper EV2 with a Sony camera.  He had to get Transport Canada Flight Certification which requires line of site, daylight only and less than 300m altitude operation.  The area covered by the UAV flights was 500 square meters.

He used a technique called structure from motion which takes the 2D images captured by the UAV and stitches them into point clouds, ortho mosaics, and digital surface models.  By way of comparison with a typical fixed wing overflight equipped with LiDAR which typically generates two points per square meter, the point clouds captured from the UAV have a density of 500 points per square meter.  The ortho mosaics and digital surface models captured with the UAVs (and using ground control points)were quite impressive with 3-5 cm pixels and rmse less than 5 cm.

Comments

[Levin Rodriguez]

Geoff, you should check out Bentley's Acute3D technology for reality modeling.... www.Acute3D.com
It produces amazing 3D surfaces from photographs taken from AUVs.

Levin