GIS for Flash Flood Hazard Mapping in Oman
Ghazi Al-Rawas, Dr. Magaly Koch, and Dr. Farouk El-Baz
Methodologies in geographic information
systems (GIS) provide a way to address common environmental
problems, including flash floods in arid areas. Flash floods
are generated when precipitation saturates the drainage capacity
of the basin slope and causes impoundment of the drainage network,
resulting in exceptionally high discharge amounts at the basin's
Flash flood hazard map of the Saham basin in northern Oman
Flash-flood-prone wadis (dry river channels cut into the terrain)
can be assessed and delineated through the use of GIS. As a
part of a research project, we used this methodology to determine
the hazard of flash floods in the wadis of the Sultanate of
Oman. Data used in this study consisted of two raster layers:
a digital elevation model (DEM) at a 90m resolution, and an
annual mean rainfall layer at a 1km resolution. These layers
were used as input data for ArcView GIS software in order to
delineate the watersheds and calculate the flow accumulation
in each of the 28 major drainage basins of Oman (Figure 1).
Several morphometric parameters
were identified for each basin in order to estimate its flash
flood potential. The morphometric analysis consisted of classifying
and measuring the geometry of drainage basins and their components
in a quantitative manner. There are many morphometric parameters
describing the geometry of drainage basins that have an effect
on the creation of flash floods. The GIS hydrologic modeling
approach was used along with a DEM, plus Landsat TM images,
in order to identify the following parameters in an automated
way: flow accumulation, mean land slope, mean rainfall, and
To calculate these identified parameters, a customized ArcView
GIS hydrology program was employed. The parameters were calculated
for each basin based upon the DEM at a 90m resolution. Applying
this program to the DEM, watersheds and their topographic properties
were delineated in a faster and highly accurate way, giving
more reliable measurements than do traditional techniques -
based on manual calculation - that use hard copies of maps.
1. Flow Accumulation
To calculate flow accumulation at the outlet of a stream, two
steps are required. First, the program needs to fill sinks (if
there are any) in the DEM data. Second, it must acquire the
flow direction grid that shows the direction of flow for each
cell in the DEM.
Sinks: There are grid
cells with no neighboring cells of lower elevation, so that
there is no downslope flow path to a neighboring cell. This
means the inclusion of flat areas and closed depressions that
can be largely eliminated from sub-segment calculations. Some
sinks may be landscape features, but most result from errors
during DEM generation, plus the limited resolution of the DEM
itself (Figure 2).
Flow direction: The flow
direction for every cell of the filled DEM is one of the keys
to deriving the hydrologic characteristics of a surface. This
function directs the flow out of each cell. There are eight
output directions, each related to the eight adjacent cells
where flow can travel. The flow direction is determined by finding
the direction of steepest decent or maximum drop from each cell.
Flow accumulation is calculated
based upon the flow direction. This function calculates the
accumulated flow as the weight of all cells flowing through
each downslope cell in the output grid. The output of flow accumulation
represents the amount of rain that would flow through each cell,
assuming that all rain would become runoff and there is no interception,
evapotranspiration, or loss to groundwater. This can also be
viewed as the amount of rain that falls on the surface, upslope
from each cell (ArcInfo 6.1 user's guide: cell-based modeling
Watershed: The watershed
is created based upon the DEM, and by specifying the minimum
number of the cells that comprise it. It is defined as an area
that drains surface water into a common outlet as concentrated
drainage. Watersheds are delineated from the flow direction
layer. Other terms for watersheds include basin, catchment,
or contributing area. This area is normally defined as the total
area flowing into a given outlet, or pour point. Sub-watersheds
are divisions based upon the drainage divide of a watershed.
Pour points are the points at which water flows out of an area.
This is the lowest point along the boundary of the watershed.
Once the watersheds have been identified, they are used as measurement
areas to calculate the parameters for estimating hazard floods
in each respective basin (Figure 3).
2. Mean Land Slope
Mean-land-slope values are derived for each watershed. The slope
plays a large role in the estimation of hazardous floods. High
slopes can lead to flash flood generation. Using the DEM, the
slope of each basin is derived. A raster layer of slope is used
with an ArcInfo grid program to generate a theme that shows
the mean slope of each sub-watershed within the basin.
3. Mean Rainfall
The rainfall coverage with 1km resolution for the whole country
was used in this study. The mean rainfall was calculated for
each sub-watershed within each basin. 4. Shape Factor Shape
factor is a measure of the shape of a basin computed as the
ratio of length of the basin to its computed area. The formula
for such computation is as follows: Shape factor = (LB)2 / A.
A comparative study conducted
in three selected sites in the Egyptian desert suggests that
circular basins create flash floods with sharp discharge peaks
(quick flow), whereas elongated basins are generally characterized
by slower flood-water flow (Table 1).
Results and Conclusions
Wadis that are expected to carry flash floods are ranked according
to risk factors to establish the most dangerous places in the
event of a flash flood. One example of these results is given
here for the Saham basin of northern Oman (Number 5 in Figure
1). Figure 4 shows all parameters that were used to determine
the hazardous wadis in this basin. These parameters are calculated
and entered in the attribute table of the basin, and by setting
a threshold for flash flood hazards. The most dangerous areas
are identified based upon the calculated parameters of the basin
and buffer (Figure 5). For example, the buffered wadis have
the highest risk for generating flash floods as compared to
others in the sub-watersheds of the Soham basin. The graduated
red color represents the level of danger in terms of flow accumulation
for every wadi. A TM image of this basin shows that a hazardous
wadi leads to a dam. The shape factor shows that this wadi is
more circular than the others, which is more dangerous in this
case than the elongated one as mentioned above.
This study further shows
that the parameters related to flash flood hazards are more
accurate and reliable when directly calculated from a DEM. Automation
of morphometric parameters calculation is faster than with traditional
techniques. The drainage network digitized from Landsat TM data,
to the drainage network delineated directly from the DEM, shows
similar patterns, especially the main flow-channel segments
of the network. However, the flat areas are less well represented
(straight lines) by the flow accumulation and the stream network.
The flash-flood-hazard wadis were determined and mapped for
most of the basins in the Sultanate of Oman. These maps may
prove to be very important for infrastructure planning, tourism,
and water management. For example, based upon these maps, the
best location for building dams can more easily be identified
in order to recharge wells and prevent flash flood occurrence,
especially in residential areas.
About the Authors:
Mr. Ghazi Al-Rawas is a graduate student at the Center
for Remote Sensing, Boston University. He is also an assistant
lecturer at Sultan Qaboos University in Oman. He may be reached
via e-mail at firstname.lastname@example.org,
or at email@example.com.
Magaly Koch, Ph.D., is a research assistant professor
at the Center For Remote Sensing at Boston University. His e-mail
address is firstname.lastname@example.org.
Farouk El-Baz, Ph.D., is the director of the Center for
Remote Sensing, Boston University. His e-mail address is email@example.com.