ershed 
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
The quantification of floodplain inundation by the use of LANDSAT 
and Metric Camera information, Belize, Central America 
50 
30 
80 
10 
75 
50 
20 
25 
60 
15 
5 
25 
45) 
45 
10 
5 
30 
10 
35 
10 
_5 
20) 
55 
l) 
elands has 
3 ofoLandsat TM 
,e of the large 
the vegetation 
:y also to dis- 
aquatic vege- 
iter areas, 
inges during 
¡en Landsat TM 
iresent too ma- 
S.T. Miller 
University of Aston, Birmingham, UK 
ABSTRACT: Investigations are made into the quantification of a flood event using a single post flood LANDSAT 3 
image aided by basic topographic and hydrological information. Comparisons are made between LANDSAT, Metric Camera 
and ground survey data to assess the contribution that space borne remote sensing can make where limited ground- 
truth is availible. The results indicate that such images can provide detailed accounts of flood behavior and 
estimates of flood volumes in circumstances where ground surveys are impracticable. 
I. BACKGROUND 
The Belize river catchment covers an area of more than 
6,000 km . in Central America and its floodplain is 
inundated every few years. Economic dislocation due to 
the flooding of the road link (the Western highway) 
between the capital Belmopan and the main seaport of 
Belize City, is a serious concequenoe. The floodplain 
area is largely uninhabited and has few other roads. 
Access within the area is difficult at all times and 
impossible at times of flooding. In such circumstances 
remotely sensed images provide an excellent opportunity 
for the quantification of flood extents. In this paper 
the flood of December 1979 is investigated. 
While the flood distribution is considered through 
out the floodplain area, estimates of flood volumes 
are made only for the part above Davis Bank gauging 
station, the upper floodplain. In the case of the 
lower floodplain, comparisons are made with other 
information sources as to floodwater distribution and 
the locations of flooding and destruction of the 
Western highway. The LANDSAT scene 020/46» 29th. 
December 1979 was aquired at least two weeks after the 
inception of flooding. 
2. THE UPPER FLOODPLAIN 
2 .1. The hydrological data base 
In common with many developing countries, hydrological 
and meteorological records in Belize are short, incre 
asing the problems of flood volume assessment. The 
first stage of this investigation was to establish a 
base for return period peak flows that could be used 
in flood volume calculation. 
Several options were considered before the 'Peak- 
Over-Threshold (P.O.T.) method was selected, for the 
following reasons(Flood Studies Report 1972): 
1. It is based on a simple exponential series, the 
validity of which is easily tested. 
2. It is suitable for short periods of records. 
3. It is adaptable for seasonal and non-seasonal 
variations. 
4. It is adaptable to conditions where flow peaks 
do not adhere to the same distribution at all stages. 
Five gauging stations, two outside the floodplain 
and three within it, were checked for conformity to 
an exponential distribution of high flows. The three 
floodplain stations were Big Falls Ranch, Bermudan 
handing and Davis Bank having 3» 4 and 2 years record 
respectively, obtained between I968 and 1972. It was 
noted that while the two stations outside the flood- 
plain conformed to a consistant P.O.T. distribution 
at all stages, the three inside did not. Their 
distributions were found to have varying exponential 
gradients. 
Since the P.O.T. distributions depend upon the thresh- 
hold value (qo) and the exponential gradient (b), the 
former was set to assure an exponential gradient and 
distribution of observed peak flows for return periods 
of five years or more. The number of peaks per year 
that exceed the threshold (qo) is termed A . The 
equations that define return period flows are of the 
form Q(T)= qo + b(ln A . In T) and are given below in 
table I for the floodplain stations. (Flood Studies 
Report 1972:189). 
Table I. P.O.T. formulae for floodplain stations 
Station 
P.O.T. formulae 
Big Falls Ranch 
Bermudan Landing 
Davis Bank 
Q(T)= 400 
Q(t)= 400 
Q(T)= 280 
+ 26,35(ln2 + 
+ 20,20(lnl.5 
+ 28.00(lnl + 
InT) 
+ InT) 
InT) 
where Q(T) , the T year flood and qo are in cumecs. 
In this way a flood peak return base was identified 
for later use, incorporating a slow growth in keeping 
the limited ability of the river channel within the 
floodplain to conduct flood discharge. 
2.2 Synthesis of the 1979 flood event 
Having determined a suitable hydrological base for the 
determination of flood peaks, an assessment of the 
flood volume of the 1979 flood event was necessary. 
No hydrological records of the event were available. 
However past hydrological and meteorological records 
enabled the construction of unit hydrographs. Two 
previous events occuring in 1969 and -1971 were used 
and from their averages unit hydrographs for each 
station were obtained (Flood Studies Report 1972:375- 
402). 
The time period for the description of these unit 
hydrographs was four days, determined by the nature 
of the rainfall periods and daily rainfall records. 
They were converted to one day unit hydrographs for 
convenient application to the 1979 flood (Flood Studies 
Report 1972:397) 
The reconstruction of the 1979 flood flow hydrographs 
was completed using rainfall rcords from all available 
meteorological stations. Areal weighting was applied, 
as were estimates of probable base flow at the time of 
the 1979 event, consideration of the I969 and 1971 flow 
hydrographs indicated that a 15%increase in the derived 
hydrogxaph was appropriate in the case of each station. 
Although the practical limitations of such methods 
are recognised (Flood Studies Report 1972:379-581),