Preliminary Screening of a Large Number of Trees
(Up To Tens of Thousands)
To Detect Those with Continuous Deterioration
Regardless of Species
There is no need to acquire data by special airplanes and UAVs
The use of hyperspectral and LiDAR data is not required
This new technology is designed to detect those suspected trees of declining health, so that further detailed analysis of them can be undertaken to confirm the state of health.
It does not foretell when or whether a stressed tree will collapse, but it reveals changes in internal health condition objectively and quantitatively, before external signs and symptoms are observable, so that arborists can carry out diagnosis and treatment expeditiously.
Detection is based on subtle changes in chlorophyll content and leaf cellular structure which produce variation in spectral reflectance from foliage caused by biotic and/or abiotic factors. The spectral reflectance is generated from AComp (atmospheric compensation) pan-sharpened WorldView-2/3 satellite data of 50cm to 30cm resolution. The interpretation is derived from 3 spectral bands of (1) NIR, (2) Red Edge, and (3) Red bands. They constitute the internal warning indicators.
If the spectral reflectance value of (1) and (2) drops and that of (3) rises (less absorption) over a period, the indication is declining health.
As the spectral reflectance of different species will vary, the list of declining trees in descending order from high to low does not imply any ranking of the internal health condition. The indication is that all are in declining health. Our objective is to find out a tree which has continuous declining value rather than making comparison of the health condition among different trees.
Sample of tree screening
The preliminary screening procedure includes three levels of approach. They are as follows:
Level 1 Declining value in NIR (one band)
Level 2 Declining value in NIR + Red Edge (two bands)
Level 3 Declining value in NIR + Red Edge + Red (three bands)
All these three levels of screening are based on the spectral reflectance value of each tree derived from the respective bands in three recent satellite images.
The following two tables present the results of preliminary screening of a large number of trees.