Application and Validation of LWCI (Leaf Water Content Index) to Tropical Seasonal Forest Region
3. Methodology and Analysis
(1) Definition of LWCI (Leaf Water content Index)
On the basis of LWCI proposed by Raymond Hunt (1987), we contrived new LWCI intended for practical use, so as to utilize the energy value of remote sensing data instead of measuring reflectance. The index TM4 and TM5 spectral bands in which the difference in water content of leaves appears conspicuously (Fig.1).
The formula of new LWCI is as follows,
A: Maximum value of TM4 through whole images
B: Maximum value of TM5 through whole image
a: a correction coefficient (usually 1).
b: the coefficient for converting the energy value into the reflectance(the reflectance value ratio at the season of the minimum water stress in a leaf = the ratio of TM5 in the theoretical solar radiation spectrum curve: the representative value : 0.2).
TH4ft & TM5ft: TM4 and TM5 at the season of the minimum water stress (full turgor), that is these are the maximum value composite image of TM4 derived from all image and the minimum value composite image of TM5 derived from all images.
-
Here, instead of reflectance, the energy value (transformed from radiance according to sensor's dynamic range) is used for TM4, TM5, TM4ft and TM5ft, respectively.
- Creating of the multi-temporal LWCI Images
By applying LWCI to those geometrically corrected and radimetrically rectified images, 7 LWCI images were created (Fig.2).
- Creation of the multi-temporal LWCI Images
The seasonal change of LWCI and the meterological data of the Mae-Klong watershed research station were compared (Fig.3). The experimental plot data (LAI, Leaf Litter Production) of MWRS were compared with LWCI. Furthermore, GPS photos at the ground were compared with LWCI Image by using 3D perspective view images.
- Water Available Map in Kanchana Buri based on LWCI
The K-means clustering method was applied to the LWCI images to classify the pattern of the LWCI changes in this area. As the result, 21 cluster were identified. These 21 cluster were summarized into 10 categories which show different seasonal change patterns of the LWCI.
The difference of the seasonal changes of LWCI are caused by the water available and also by the vegetation types. Therefore, a filtering is needed to be applied to the clustering result to minimize the influence of the difference of vegetation types and to get the zoning map on water availability. (Fig.4).
- Environmental Gradient Analysis for three watersheds
The study area was topographically divided into three watersheds for analyzing the site characteristics by the environmental gradient analysis.
Integrating the DEM data, the environmental gradient analysis (Fig.6) was conducted to the water availability map for three watersheds, north side, and east side of the study area(Fig.5).
The ratio of the classes on the LWCI changes were summarized in every 100m interval of the elevation.
Result and Discussion
(1) Comparison of LWCI with other observed data.
In Comparison of the seasonal change of LWCI and the meteorological data of the Mae-Klong watershed research station, it seems that the changes of LWCI correspond to precipitation and evaporation with the time lag of 1-2 months(Fig.3). However, it's necessary to collect more data ranging long term (more than 2 years) for analyzing the detail correlation.
(2) Water Available Map and Environmental Gradient Analysis
The results (Fig.6) clearly show as follows,
-
In the north side (where the watershed faces the north), the rate of evergreen trees and deciduous trees that do not shed the green leaves is the highest. This indicates that the water availability condition is the best at the north side.
- In the west side, the relation between the elevation and the rate of evergreen is linear.
This indicates that the west side has the standard trend of the vegetation types according to the elevation.
- Compared with the north side, it is found that the land use of the west and the east sides are much more advanced (agricultural area, irrigating water etc.).
In tropical seasonal forest zone, LWCI is useful index for zoning water availability of vegetation area and for analyzing the phonological dynamics dry season. However, in case of comparison among pixels, it is necessary to consider the crown density and other factors that related to leaf density.
