461
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
Double sampling for rice in Bangladesh using Landsat MSS data
Barry N.Haack
George Mason University, Fairfax, Va., USA
Abstract: Landsat Multispectral Scanner (MSS) data were utilized to assess the location and area of the
winter or boro rice crop for a study site in Bangladesh. Both digital classifications and visual interpre
tations of the Landsat data were conducted using a double sampling between the Landsat analyses and govern
mental reported acreages for ten administrative units. The digital data were classified using a table look
up of MSS Bands 5 and 7 and the visual analysis conducted with an enlarged false color composite subscene
and a rearview projector. Both techniques located about one-half of the reported acreage but had correlations
between the reported and Landsat data for the ten units of over 0.90. This study indicates that double
sampling with Landsat data can be an effective technique for the inventory of renewable resources in rural
areas.
1 INTRODUCTION
A common problem to many decision-makers,
particularly in less developed countries, is the
inadequacy of information on the available
resource base. Without accurate information
decision makers often fail to make decisions or
make incorrect decisions. Sound decisions depend
on accurate information, yet every low income
country faces severe competing demands for the
financial and human commitments necessary to staff
an information system equal to its decision-making
requirements (Cummings, 1977). The frequent
inadequacy of resource base information may be due
to difficulties in accessing some regions; lack of
trained personnel, equipment or funds to collect
information properly; or rapid changes in the
resource base not detectable by traditional data
collection methods such as the high rates of
deforestation in many areas of the world caused by
increasing population pressures. One technique
which may be utilized to resolve some of these
informational needs is remote sensing and
especially operational spaceborne remote sensing
systems such as Landsat (NAS, 1977). The Landsat
series of satellites have provided essentially
global data, relatively inexpensively and on a
repetitive basis since 1972.
Bangladesh has an intense need to increase the
efficiency of national land use for food and fiber
production. One action taken by the Government to
address this problem was the formation of the
Landsat Task Force composed of scientists and
resource managers from several ministries. The
purpose of the Task Force was to assess the
applicability of and utilize satellite and other
remote sensing technologies in Bangladesh.
Preliminary investigations conducted by the
Landsat Task Force produced promising results for
winter rice crop identification, water course
mapping, and forest inventory among others.
Because of these initial successes the Government
of Bangladesh has further developed its scientific
and technical capabilities in satellite remote
sensing.
This study examined the ability of Landsat data
to conduct an inventory of the winter rice crop in
Bangladesh. Once the location and extent of
current winter rice is adequately known, methods
to increase the yield of this rice or locate new
rice production areas can be pursued.
2 STUDY SITE AND CONDITIONS
Bangladesh is composed largely of alluvial and
deltaic terrain. The low surface relief,
relatively fertile soils, and intense population
pressures result in approximately 60 percent of
the land being used for agricultural production,
with rice being grown on over 70 percent of the
cropped area. The climate is tropical, with
monsoonal rains in late spring, summer and early
fall. Winter, however, is dry and cool, and only
a few areas where there are natural depressions
which were flooded in late fall, or where
irrigation is practiced, are suitable for rice
growing. In an effort to increase the production
of the winter or boro rice crop, the Government
has encouraged additional irrigation by supplying
the necessary water pumps. For management and
planning purposes, the Government would like to
know the effect of this program. This information
has proved either difficult to obtain or of
questionable accuracy because of the poorly
developed infrastructure associated with
agricultural statistics reporting. This project
examined techniques using Landsat data to provide
information on both the location and extent of
boro rice.
The study site selected for this analysis was
approximately 8500 square kilometers located east
of the capital of Dhaka. This area is along the
Meghna River and included parts of Buidya Bazar
Thana, Rubgonj Thana and the Dhaka-Narayangonj-
Demra Agricultural Project. This study area was
selected because it; 1) was an area of known boro
rice production both in large contiguous blocks
and small individual fields, 2) was suitable for
expansion of boro rice production, 3) had
excellent proximity to Dhaka for field work and A)
had available agricultural statistical information
by administrative units, unions, within the Thanas
(Chaudhury, 1978).
A crop calender was constructed for the boro
rice by discussions with agricultural officers and
farmers within the study area. From those
discussions, it was determined that the boro rice
seedlings are raised in seedbeds in November,
transplanted to the main fields in December and
January and harvested in April or May, prior to
the monsoon. The latter part of March is the time
of maximum green vegetative cover by the boro
rice. Other winter crops such as Rabi vegetables
