Tree health monitoring with remote sensing
Tree health condition from the near-inferred and red bands ratioing
With the combinations of photogrammetry and remote sensing, we are able to detect the tree health and condition with both active (LiDAR) and passive (aerial images) remote sensing. Indexing (such as NDVI) make use of the band nature (near-inferred and red bands) can pinpoint some trees are unhealthy and stressed with less reflections and it changes over time (Figure 1).
#Temporal image differentiating has a distorted raster, due to the images were not rectified, so there are perspective effect among the raster differentiating process
Collapse accident in Oct 2022, Perth Street
Government responds and actions
The government inspects the trees in the vicinity, removing those with shallow roots while preserving those with deep roots. They also install pest control boxes to minimize the impact of pests and termites on the structural integrity of the trees.
Additionally, the TMO’s tree care and management scheme, PlansD, can use aerial photogrammetry and remote sensing techniques to detect changes in tree health, allowing them to see what is not visible to the naked eye using geomatics.
Methodologies to enhance the tree health checking with Remote sensing
The RGB (visible) bands composition, Figure 2 (left) cannot reflect the tree health, due to the band ratioing will be affected by the topographical effects and shadows. On the other hand, false colour images can showing the unseen tree hazard in another way we can readable with near-inferred bands. In Figure 2 (right), this shows the different of the dead, stressed and healthy leafs band dynamics and composition.
Base on the reflection and plant nature itself, the dead leaf has very low chlorophyll, anthocyanin content and reduced photosynthetic activity, stressed leaf compare has a relative higher content than the deadwood, the healthy leaf has the highest reflectance to the sensing platforms.
Then, we can do analysis the dead and stressed leafs by the band ratioing (NDVI).
Band ratioing is a widely used technique in remote sensing to evaluate the health of vegetation. LiDAR can help us locate the midpoints of trees, while aerial images can provide us with the health index profile of trees. This data can then be utilized to detect changes and analyze the health of potentially problematic trees. Figure 3 displays the band ratioing formula for computing NDVI, which can be used to identify stressed vegetation and monitor its health condition.
Using the collapsed tree (Delonix regia) as sample case study, we can first pinpoint health index of that tree by assigning 22 sample points from it canopy layer (Figure 4 and Table 1).
The tree height of the collapsed tree is 13.147 meters tall and crown spread is around (10.45+12.31)/2 = 11.380 meters long [2].
From the LiDAR point cloud figure, it is evident that the trees are leaning towards the North, and the canopy layer on the North (left) side is denser than on the South (right) side.
Figure 1 shows a clear change detection between Spring 2019 and Spring 2022, indicating that the tree has decayed in terms of NDVI and reflectance. Pinpoint analysis reveals that the median NDVI change is -0.17, indicating gradual decay compared to surrounding trees (Figure 5), whose median changes are -0.10.
In addition, there is a significant difference in NDVI change range between stressed and normal trees. For the normal trees selected (max: 0.12) and the stressed wood (max: -0.08), it is evident that NDVI can clearly indicate stressed vegetation.
For the relationship between the NDVI changes and the tree height, from Figure 4 (right), the NDVI changes in the lower part of the tree more than the canopy part of it, it is because of shading and competition for resources.
Shading and competition for resources can affect the NDVI changes in the lower part of a tree. When a tree is shaded, it receives less sunlight, which can lead to a decrease in NDVI values in the lower branches and leaves.
Trees also compete with each other for resources such as water and nutrients, and those that are taller or have more extensive root systems may have an advantage over shorter or less established trees. This can result in reduced growth and vitality in the lower parts of trees that are not able to access sufficient resources, leading to decreased NDVI values.
Suggestion and future outlook
In my opinion, tree monitoring should not be limited to arboriculture professionals alone. Other industries such as geomatics and remote sensing can also play a significant role in tree management.
NDVI, which is a basic technique in remote sensing, can be used to monitor vegetation health and growth over time. Additionally, other techniques such as tree reflectance fingerprinting can be developed to provide more detailed information about the health and condition of individual trees.
By combining these techniques with open data analysis, we can work towards developing a smart city tree management system that promotes safety and sustainability in Hong Kong.
Reference:
Agriculture, Fisheries and Conservation Department. (2006). Measurement of Diameter at Breast Height (DBH). Retrieved from https://www.afcd.gov.hk/english/conservation/con_tech/files/common/NCPN_No.02_measurement_of_DBH_ver.2006.pdf
Government of the Hong Kong Special Administrative Region. (2022, September 16). Government’s response to tree collapse incident at Perth Street, Ho Man Tin . Retrieved from https://www.info.gov.hk/gia/general/202209/16/P2022091600622.htm
Green Aero Tech. (n.d.). What is NDVI? Retrieved from https://www.greenaerotech.com/what-is-ndvi/
The Standard. (2021, September 28). (Video) Close call for school bus driver as 10-meter-tall tree collapses onto vehicle’s roof. Retrieved from https://www.thestandard.com.hk/breaking-news/section/4/194770/(Video)-Close-call-for-school-bus-driver-as-10-meter-tall-tree-collapses-onto-vehicle’s-roof
