David Napier, Dr Vernon Manville
NZ Aerial Mapping Limited
Hastings, New Zealand
The accuracy and repeatability of LiDAR surveys is now entering the legendary arena; can this help us in the world of volcanic hazards?
At Mt Ruapehu, in the central North Island of New Zealand, a series of eruptions in 1995 and 1996 expelled the summit Crater Lake, generating a sequence of eruption and later raintriggered lahars. Early in 1996, the potential for a significant Crater Lake break-out lahar was identified as the Crater Lake began to refill behind a fragile barrier of volcanic material, or tephra, deposited on the hard rock rim. Lahar is a fast flowing mixture of rock debris, sand, silt and water (other than normal stream flow) originating from a volcano, along a river valley. It has the consistency of concrete: fluid when moving, then solid when stopped and can be extremely dangerous.
The threat of such a lahar was taken very seriously as the last time this situation occurred, in December 1953, a major lahar destroyed the Tangiwai Rail Bridge crossing the Whangaehu River at the very moment a train was crossing it. Since then, lahar monitoring equipment has been installed 14km upstream of the bridge to ensure a repeat does not happen.
The Whangaehu valley is one of the most active lahar channels in the world. In addition, eruption-triggered lahars have also passed through New Zealand’s largest and biggest ski area on the northern slopes of the volcano. As the lake rose above the base of the tephra dam in early 2007 it became a matter of when, not if, the lahar would occur. The worst possible scenario in these circumstances was sudden collapse of the new tephra dam causing a lahar as large as or bigger than the 1953 Tangiwai lahar.
Nowadays, Mt Ruapehu and the sister peaks of Ngauruhoe and Tongariro are constantly monitored by GNS Science, based at the Wairakei Research Centre, just north of Taupo in an area of abundant geothermal activity. The GeoNet team has installed a comprehensive suite of sensors around the mountains to monitor volcanic activity.
One component of installing a dedicated a lahar-monitoring system was the need to produce a highly accurate 3D map of the predicted lahar path along the upper Whangaehu River both before the lahar occurred, and most critically immediately afterwards. Specific requirements of the data needed led the researchers to consider the use of LiDAR equipment. The pre lahar data was flown on 18th February 2006 as a joint operation between Fugro Spatial and NZ Aerial Mapping Limited using Fugros Leica ALS50 LiDAR unit mounted in NZ Aerial Mappings Cessna 402b. Data from this survey was processed by Fugro and supplied to GNS.
Just over one year later, the tephra dam failed on the morning of 18th March 2007, releasing 1.3 million cubic metres of warm acidic lake water. The lahar passed Tangiwai, 40 km downstream within 2 hours and reached the coast 155 km away in the early hours of the following morning. Thanks to a detailed inter-agency response plan and the ERLAWS lahar warning system installed by the Department of Conservation no lives were lost and infrastructural damage was minimal. Within less than three weeks of the lahar the post-lahar survey was performed, on 6th April 2007. The post lahar data was captured with an Optech ALTM 3100EA again mounted in Cessna 402b, both owned and operated by NZ Aerial Mapping Limited.
Operationally, the project was rather difficult from two quite diverse angles. Firstly, the area of survey spanned more than 2,000m of vertical relief requiring very accurate flight planning and flying to ensure correct overlaps and point density were achieved.
Additionally, as the area is situated within an Military Restricted zone, formal clearances were required to fly within it. To complicate matters further, at this time a major international exercise was taking place which required the highest levels of cooperation and coordination between the Military, Weather Office and Aircrew. In addition to the LiDAR data, digital imagery was also captured by the cosited medium format camera. This imagery provided both checking imagery for the LiDAR classification and also a set of orthophoto images for eventual supply. For the processing, the LiDAR sensor positioning and orientation (POS) was determined using the collected GPS/IMU datasets and Applanix POSPac software. This work was all undertaken using NZGD2000 coordinate system.
The POS data was combined with the LiDAR range files and used to generate LiDAR point clouds in NZTM map projection but NZGD2000 ellipsoidal heights. This process was undertaken using Optech REALM LiDAR processing software. The data was checked for completeness of coverage then the relative fit of data in the overlap between strips was checked. The point cloud data was then classified into ground, first and intermediate returns using automated routines tailored to the project landcover and terrain. This and subsequent steps were undertaken using TerraSolid LiDAR processing software modules TerraScan, TerraPhoto and TerraModeler. The data was converted from NZGD2000 ellipsoidal heights into Moturiki 1953 vertical datum using the LINZ NZGeiod05 separation and offset model. Comprehensive manual editing of the LiDAR point cloud data was undertaken to increase the quality of the automatically classified ground point dataset. Independent of the aerial acquisition work, Opus International Consultants field surveyed a series of check sites in open ground, to be later used to verify the accuracy of the processed datasets. Sites were chosen outside of the lahar flow path. The height accuracy of the processed data was checked using the provided check site data. This was done by calculating height difference statistics between a TIN of the LiDAR ground points and the checkpoints. The positional accuracy of the processed data was checked visually by overlaying the check site data over the LiDAR dataset displayed with its intensity values. The data was found to fit well in position.
LiDAR data was processed for delivery into two main sets of data: thinned and unthinned. The thinned dataset contained ground classified points only of the entire area and was made up of approx. 45 million points. The unthinned dataset was divided into: First of Many, Intermediate, Only & Last of Many and Water Point Cloud and consisted of approx. 87 million points.
All the LiDAR data was supplied to GNS as ASCII XYZ files whilst orthophotography tiles at 0.21m GSD were supplied as TIFF/ESRI TFW files. Analysis of the pre and post lahar data sets has provided a never before insight into the behaviour and outcome of this lahar. Most significantly for this particular event was the realisation that the lahar was approximately 25% larger than the 1953 Tangiwai lahar.
For further information please visit