Assessing The Local Spatial Variation In The Relationships Between Rainfall, Vegetation And Elevation
Rainfall varies spatially ranging from large to local scales. Spatial elements such as vegetation and topography are the contributing factors to local variations of rainfall. However, local spatial variation process in rainfall due to vegetation and topography is unidentified when using a global...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/48261/1/ROHAYU%20HARON%20NARASHID_hj.pdf http://eprints.usm.my/48261/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Sains Malaysia |
| Language: | English |
| Summary: | Rainfall varies spatially ranging from large to local scales. Spatial elements
such as vegetation and topography are the contributing factors to local variations of
rainfall. However, local spatial variation process in rainfall due to vegetation and
topography is unidentified when using a global model. This study aims to assess the
local spatial variation of rainfall in the relationships between rainfall, vegetation and
elevation using a local modelling approach. The main data used consist of rainfall
depths, vegetation index of Normalized Difference Vegetation Index (NDVI) from
Landsat 7 ETM+ satellite images and the elevation data from 174 and 103 locations
of rainfall stations within the Northern and East Coast Region of Peninsular Malaysia
respectively. Based on the availability of NDVI datasets from the years 2000, 2009
and 2011, the local spatial variations of rainfall were determined. The small
clustering patterns in rainfall, vegetation and elevation that were computed in
Moran's Index with the value of 0.1 to 0.5 showed low values of the variables being
clustered in the study areas. Thus, the spatial process in rainfall, vegetation and
elevation demonstrated a potential for local variations. The spatial pattern of these
variables led to the exploration of non-stationary relationships. In order to explore the
local spatial variation of rainfall, the regression techniques of Ordinary Least Square
(OLS) and Geographically Weighted Regression (GWR) were applied to determine
three types of models i.e. : (1) the relationship between rainfall and vegetation; (2)
the relationship between rainfall and elevation; and (3) the relationship between rainfall, vegetation and elevation. The statistical findings for all relationships had
shown significant local variations when Akaike's Information Criterion (AICc)
obtained from GWR were lower. The GWR R-squared (0.146 to 0.770) improved the
OLS r-squared (0 to 0.176). The best GWR model with the highest AICc difference
values ( AICc) for years 2000, 2011 and 2009 were found in Model 1(164.571),
Model 3 (163.946) and Model 2 (147.605), respectively. Land use and vegetation
changes are the possible reasons when the relationship between rainfall-elevation for
year 2011 was found to be more significant. The significant location of local spatial
variations of rainfall due to vegetation and elevation can also be demonstrated based
on the findings. With the detailed capabilities provided in remotely sensed data, the
local variations of the relationships are possible to be carried out. Therefore, the
spatial relationship that exists between rainfall, vegetation and elevation at the local
level are significantly contributing to the local variations in rainfall. |
|---|