Questions & Answers
1. What is this new technology on tree health monitoring?
It is based on spectral reflectance of tree foliage produced by subtle changes in chlorophyll content and leaf cellular structure generated with the NIR, Red Edge and Red bands collected by WorldView-2 and 3 satellites of 50cm to 30cm panchromatic resolution respectively. AComp (Atmospheric compensation) data is used for analysis. AComp is a proprietary DigitalGlobe algorithm to process WorldView-2 and 3 satellite data. It normalizes digital number (DN) value to a surface reflectance value. It also removes imagery variation due to illumination, viewing geometries, and atmospheric effects, such as light-wave scattering from haze, water vapour and particulates in the atmosphere. Hence spectral reflectance analysis derived from AComp data will produce consistent and reliable results.
2. What does this new technology reveal?
This new technology is an early detection system to reveal internal health condition of a tree not discernible externally. With 3 years data, it will show 3 internal warning indicators derived from drop in spectral reflectance of the NIR and Red Edge bands and reduced absorption in the Red band. It will also identify 4 categories of tree health, namely (a) improving, (b) stable, (d) fluctuating, and (d) declining. The objective and quantitative information provided will assist arborists to carry out onsite investigation to implement appropriate precautionary and remedial measures.
3. What this technology does not do?
This new technology is not designed to foretell when or whether a stressed tree will collapse. It does not show changes in the inclination of a tree. Instead, it reveals the changes of health condition of a tree which is of paramount importance for tree management and maintenance.
4. Has this new technology been verified?
We have undertaken many case studies of stressed trees in Asia, Australia, Europe and North America to verify this new technology. Details are presented in this website: Monitoring projects are also supported by consulting arborists in other countries. Our two articles, entitled “A New Technology Warns of Stressed Trees” and “An Integrated Approach to Tree Stress Monitoring” have been accepted by the International Society of Arboriculture (ISA) for publication in their bi-monthly magazine, Arborist News, April 2017 and August 2018 issues. A white paper, entitled “Detecting Stressed Trees with Satellite Imagery” was published by DigitalGlobe in June 2017. They are available online at
5. What are the advantages of this new technology?
This new technology is based on relative comparison of the changes of spectral reflectance of the same tree on different dates in about the same month/season, with about the same collection specifications and under the same environmental conditions. The following are among the main advantages:
(1) It is a straightforward, time-saving, cost-effective and non-invasive approach without the use of specialized instruments and computation.
(2) It is an indoor computer-based technology without field visits, unless there is a need to verify the tree location or to carry out diagnosis and treatment.
(3) There is no need to identify the species and age of a tree, because our emphasis in on relative comparison of the same tree on different dates.
(4) All trees with green foliage can be monitored as long as they can be imaged by the satellites.
(5) There is archive data dating back to 2010 for retrospective study of the changes in health condition of a tree.
(6) One of the two satellites can collect high resolution data covering the whole Hong Kong Special Administrative Region of 4,000 sq.km in a single pass on one date.
(7) The multispectral data can be pan-sharpened to 50cm - 30cm resolution with a “lossless” programme to enhance tree identification and analysis. This process will retain the fidelity of the original data without losing the spectral information content of each pixel.
6. What are some misconceptions about the applications of this new technology?
It has been erroneously assumed that this new technology may not be practicable in urban Hong Kong due to reflection interference from tall buildings. So long as a tree can be imaged by the satellites, any interfering reflection from a nearby building will not affect the result, because such reflection will also exist in all the images collected. Our analysis is based on relative comparison of the spectral reflectance of the same tree. Absolute value is not required. For example, we have analyzed the stone-wall trees in Forbes Street, Kennedy Town using WorldView-2 data collected in July 2010 and August 2016, although the street is narrow with tall buildings on one side.
Another misconception is that if a healthy tree is pruned, its spectral reflectance will drop correspondingly, because there is less foliage. This is not the case. It will drop only when it is severely or improperly pruned, i.e. its health has been impaired. On the other hand, if pruning is done properly, it is beneficial to the tree and the spectral reflectance will not drop. Instead it may rise if the health condition is ameliorating. Our approach to generate spectral reflectance is not polygon-based covering the canopy. Instead, It is pixel-based. Several pixels are selected from the foliage and the spectral reflectance is the average value of the selected pixels, not an aggregate value of the total foliage covered. It has been assumed that satellite data does not have the high resolution to monitor an individual tree. This is true in the past, before the launch of WorldView-2 and 3 satellites on October 8, 2009 and August 13, 2014 respectively. The data collected by these two satellites is now high enough for monitoring an individual tree. The resolution of the panchromatic data is 50cm to 30 cm, while the multispectral resolution is 2m to 1.2m. Using a “lossless” pan-sharpening programme, we can produce 50cm to 30cm resolution data to generate the spectral reflectance curves. They are the highest resolutions available for commercial application at present.
7. How much does it cost to implement this tree monitoring technology?
The archive satellite data comprising panchromatic band of 50cm resolution and multispectral bands of 2m resolution is US$19.00 per sq.km with a minimum order area of 3km x 3km = 9 sq.km. In comparison, Kowloon Park covers an area of about 0.135 sq.km. At least 3 dates of data are required. So the total data cost is US$19.00 x 9 sq.km x 3 dates = US$513.00. The processing fee is US$6.50 per tree for one date. So the total processsing fee for 3 dates is US$6.50 x 3 = US$19.50.