Archives for March 9, 2015

A group of researchers in Utah, home of the finest powder snow on earth (and I speak from direct experience) has developed a system to report on the snowfall conditions during storms. The research originally supported by the National Science Foundation (NSF) is now being commercialized with the intent of selling the camera – based system to the DOTs.

Using fundamental precipitation research from atmospheric scientist Timothy Garrett, who developed the original multi-angle snowflake camera with Cale Fallgatter, the technology can resolve falling particles down to the diameter of a human hair and also measure the speed at which they fall.

“The funding to Fallgatter Technologies will allow them to demonstrate an important new tool to better understand weather conditions in real-time,” says Ben Schrag, NSF Small Business Innovation Research program director. “And will hopefully help local authorities and meteorologists to make better decisions with regards to severe weather.”

– See more at: http://blog.lidarnews.com/#sthash.uAlMARnL.dpuf

Source: http://blog.lidarnews.com/

The points clouds obtained from Airborne LiDAR data can be used at certain extent for direct measurement of tree attributes for example diameter at breast height. The filtering process is a must to remove understorey vegetation.

The histogram-based individual tree filtering has been introduced for this purpose.

Non-destructive Laser-based Individual Tree Aboveground Biomass Estimation in Tropical Rainforest

Introduction

Research community mainly in forestry related applications have been exposed with numerous biophysical datasets from satellite imagery. However, these kind of datasets are not supported by accurate field measurements data that causes uncertainty in derived datasets. Terrestrial laser scanning (TLS) is a ground-based LiDAR technology that can be used to retrieve highly detailed 3-dimensional vegetation structure. For plot level forest measurement, TLS could assist the estimation of several useful tree parameters for example tree number and position, tree height, diameter at breast height (DBH), tree volume, above ground biomass and so on. Furthermore, Terrestrial TLS has huge potential to provide accurate tree structure including how the foliage and stems are arranged above the ground of a specific tree. This study aims at estimating individual tree aboveground biomass from high density point cloud obtained by terrestrial laser scanner (TLS). The point clouds were obtained together with field measurements data in 30 forest plots covering 604 trees over tropical rainforest landscape in Royal Belum State Park, Perak, Malaysia. Tree parameters such as Diameter at breast height (DBH), tree height and height to crown base were estimated from TLS for assessment of individual tree measurements. Tree parameters such as stem, branches and leaves volume were estimated from TLS to derive aboveground biomass through volume to biomass conversion factors.  Cylinder fitting was applied on the point cloud of the tree stem for DBH and stem volume measurements. Tree height and height to crown base are computed using histogram analysis of the point clouds elevation. Tree branches and leaves volume is estimated by fitting a concave-hull. The results were assessed using other allometric-derived aboveground biomass.

 

Research objective

The main objectives of this research is to estimate the above-ground biomass using non-destructive laser scanning approach for selective tree species in Malaysia’s tropical rainforest.

 

Study Area

This study is carried out at the northern part of Peninsular Malaysia in Royal Belum State Park, Gerik in the state of Perak. The coordinate of the area is around 5º 33’ 25.68” N and 101º38’29.41” E, located at 230km away from Ipoh and 430.5km from Kuala Lumpur. This area receiving 1998 to 2300mm of mean annual rainfall, varies throughout the year. According to (WWF-Malaysia 2013), Royal Belum State Park (RBSP) was gazetted as a protected area on 3 May 2007 under the Perak State Parks Corporation Enactment 2001. The park covers a total area of 117,500ha in the most northerly region of the State of Perak in northern Peninsular Malaysia. RBSP lies between border of Thailand on the north, the state of Kelantan to the east and Sungai Gadong in the west. The East-West highway is on its southern border separating the RBSP with Temenggor Forest Reserve to the south. Royal Belum State Park consists of forest, grassland, abandoned agricultural plots, and a large man-made lake, Tasik Temenggor. Forest types found here are mainly lowland dipterocarp, hill dipterocarp and upper dipterocarp. The majority of species are characteristic of tropical rainforest in Peninsular Malaysia, Sumatra and Bornea such as Meranti and Keruing.

 

Data collection

The primary data for this study is high density point clouds data generated from Riegl VZ-400 and the secondary data is the field measured data. These two datasets were taken by UTM@RoyalBelum Scientific expedition on 19^th of September 2014 till 30^th of September 2014.

 

   

 

Raw TLS point cloud data

Images captured from digital camera mounted on TLS during scanning were used to apply color on the registered point clouds. Scan positions and signage for tree identification can be seen clearly from the colorized point clouds to assist in individual tree extraction.

 

 

Individual Tree Extraction

Individual trees were extracted from the registered point clouds. Point clouds from neighbouring trees, understorey and vicinity branches were removed. This process was done manually to get detailed structure of an individual trees.

               

Separation of tree parts

Extracted trees were partitioned into stem, branches and leaves for biomass estimation of every compartment.

 

Reconstructed Trees

Individual trees were reconstructed from cylinders of tree stem and wrapped crown structure as shown below:

Raw point clouds obtained using Terrestrial Laser Scanner. The point clouds were collected using Leica Scanstation C10 laser scanner that operates with the green wavelength at 532nm. This scanner is capable to scan at most 300m distance object with the nominal accuracy of 4.5mm

After registration process, the unnecessary or noise point clouds of those originate from neighbouring trees, understory trees and ground surface are removed to avoid confusion and complexity in data processing. The point clouds are then separated into individual trees for further processing to estimate tree attributes. Further cleaning process of leaf and small branches are required to separate the trunk and branches prior to diameter estimation at different height intervals. The noise removal procedure has been done manually by a careful inspection of every single trees marked in the field as well as in the point clouds.

The geometric reconstruction of individual tree trunk was done by fitting the point clouds with cylinder model at different tree height.

The point clouds of tree crown is wrapped with convex-hull surface. The complete individual tree reconstruction is shown below: