You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

The 3rd ISPRS Workshop on Dynamic and Multi-Dimensional GIS & the 10th Annual Conference of CPGIS on Geoinformatics
Chen, Jun

ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001
Yinchai WANG
Faculty of Information Technology, Universiti Malaysia Sarawak,
94300 Kota Samarahan, Sarawak, MALAYSIA
Tel:+60-82-671000 ex377, Fax:+60-82-672301, E-mail:ycwang@fit.unimas.my
Teck Kiong SIEW
Faculty of Information Technology, Universiti Malaysia Sarawak,
94300 Kota Samarahan, Sarawak, MALAYSIA
High-quality rainfall measurements on a variety of temporal and spatial scales are desirable for a wide range of applications. Reliable
values of rainfall are not yet available for most areas over Sarawak because of scarcity of gauge measurements. This research
developed the optimal rainfall estimation model to estimate monthly rainfall from the surrounding rainfall stations. The technique
pursued was a geostatistical approach by taking into account the effect of topography factor on spatial variation of rainfall. But, there
are problems with the topography factor especially in dealing with geographical data and spatial analysis. These problems, that can not
be solved manually, were overcome with the aid of GIS which has full capabilities in generating geographical data and performing
spatial analysis. Thus, the whole system was implemented using GIS. The procedure involved was the interpolation method applied
over the determined best area of interpolation in which all the surrounding rainfall stations were significantly correlated to the point
which rainfall was to be estimated. Regression analysis was applied for model selection. The optimal rainfall estimation model
selected was the best multiple linear regression equation that consists of topography parameters as independent variables, applied over
the best area of interpolation. Cross-validation was used for the optimal model evaluation and for comparing with six alternate
estimation methods. The optimal model provided the best cross-validation result. The output was a GIS based Rainfall Estimation
System for estimating point and are-average monthly rainfall at an unmonitored site, and for distributing point rainfall estimates to
regular spaced grid cells in producing isohyets of monthly rainfall over the study area.
In Sawarak, many human and economic activities involving
agriculture, industry, commerce and transportation are
extremely reliant on the weather conditions. Observation of
rainfall includes raingauge measurement that is the direct and
quantitative measurement of rainfall, provided by all the rainfall
stations. Space observation of rainfall refers to infrared (IR)
satellite and radar data that give an indirect measurement of
rainfall in terms of temperature and reflectivity, provided by
satellite and radar at Kuching Meteorological Station only.
Quantitative measurement of rainfall are preferred and
required by many activities and models. The main problem
with raingauge observations lies in the lack of dense network.
As compared to vast landmass and complex terrain, there is a
lack of conventional gauge measurements in Sarawak, which
prevents adequate sampling of rainfall observations for many
applications such as agriculture and hydrological models. In
view of this scarcity of observation, estimates of the amount
and spatial distribution of rainfall are critical inputs to a variety
of these models. However, obtaining reliable estimates is
particularly difficult when the area coverage provided by the
surrounding stations is sparse and when rainfall varies greatly
with locations due to topographic effects which involve normal
land or sea breeze effects and upslope or downslope motion
of the monsoon flow over the mountains. In order to take into
account these topographic effects, which involve spatial data,
into rainfall estimation, there is a need of analytical tools
dealing with spatial data and analysis, which can not be done
manually, to improve rainfall estimation. This acts as a main
hindrance in an effort to develop rainfall estimation model to
estimate rainfall. Until recently, the advancement in computer
technology especially the recent development of GIS provides
tools tackling the spatial data. GIS, systems which deal with
spatial information, map processing, spatial database and
spatial analysis, can play an important role in dealing with
organizing and integrating apparently disparate data sets
together by geography and performing more complex spatial
The main objective of this research is to estimate rainfall at
unmonitored site using rainfall data from surrounding
raingauge stations using the interpolation procedure with the
integration of Geographic Information System (GIS). With the
aid of GIS, topography factor itself, consists of other variables
such as elevation, barriers and land-sea distribution, affecting
the rainfall and its distribution greatly can be analyzed to study
its extent of effects to rainfall estimation which, otherwise, is a
very difficult task. Thus, GIS enables rainfall estimation
technique to include as many parameters of the topography
factor which involves spatial data. This study will focus on
how to incorporate spatial analysis into the rainfall estimation
and monitoring using raingauge data.
The rest of this paper is organized as follows: Section 2
consists of literature review on the rainfall estimation
techniques. Section 3 discusses the components of the
methodology in deriving the optimal rainfall estimation model
and developing GIS-based system. Section 4 describes the
implementation of the whole system. Analysis and conclusion
are given in sections 5 and 6 respectively.
2.1. Rainguage - Rainfall Estimation
Raingauge networks measure rainfall accumulations on the
ground at fixed locations. There is a tendency to
underestimate rain amount due to physical and human errors
such as wind direction, surrounding object, overflow and
observation errors. Inspite of this, raingauges still give very
good point accuracy. However, raingauges also suffer serve
limitation in sampling especially over ocean and remote land
areas as most of raingauges tend to be distributed with a
pronounced spatial bias toward populated areas along the
coast and against areas with high elevation and/or slope . In
spite of this, several studies have demonstrated the utility of
simple models based on the point data to diagnose factors
such as topography, geology, hydrographic and vegetation
which are observed to cause rainfall variations even over short
distances such as a diagnostic model for estimating rainfall