Application of Remote Sensing Technique to Estimate Leaf Area Index in Ayer Hitam Forest Reserve, Puchong, Selangor, Malaysia

Leaf Area Index (LAI), quantifies the total foliage per unit ground surface area. It is the "driving" biophysical variable and is therefore an important input parameter to many hydrological, ecological and climate models. In this study, we develop LAI estimation model for obtaining LAI value using LANDSAT TM data image analysis. The use of vegetation indices acquired h m remotely sensed data considered as mean of estimating LAX. A study was carried out to estimate LA1 using Normalised Difference Vegetation Index (NDVI) using Landsat 'Rvl data. The objective of this study was to estimate LAI in Air Hitam Forest Reserve (AHFR) using NDVI, in which, the empirical relationship between LAI (NDVI) and measured LAI (LICOR 2000) was developed. The AHFR, located in Puchong, Selangor, Malaysia was chosen for this study. Landsat TM was digitally processed and enhanced using ERDAS IMAGINE 8.4 which produced the NDVI map. Based on this NDVI values, 12 sample plots of 20 m x 40 m were established to measure the LAI (using LICOR 2000) along with six stand parameters namely the dbh, height, crown volume, stand density, basal area and light intensity. The measurements were analysed to determine the descriptive statistic of each plot using SPSS Software. LA1 was later regressed against the stand variables as well as the NDVI value, derived earlier from the Landsat TM image, to establish the regression fit. Once the regression fit model between NDVI and LA1 was obtained, the NDVI image data was converted to the mapping of LAI. The accuracy of LAI was evaluated using the accuracy assessment in ERDAS IMAGINE 8.4. Stand characteristic variables such as dbh, height, crown volume, basal area and light intensity showed correlation with LAI except stand density. It also produced correlation with the NDVI which was derived fiom Landsat TM. Linear relationship described the relationship between LA1 and NDVI with the regression equation of LAI = 7.0861 xNDV1+ 1.0929 with high R~ of 0.81 with significance level of 0.05. The range of LAI observed (LICOR 2000) was similar to the overall range in LAI from the NDVI. Through the models of LAI value estimation, LAI mapping on AHFR can be done with each pixel holding its own LAI value. This study suggests that remote sensing can be used to estimate LA1 in forest stands and produce LAI map in AHFR subsequently provide predictive models of LAI specific for forest stands.

Text FH_2005_5.pdf Download (2MB)

Actions (login required)

View Item