×

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

Title
Technical Commission VII


International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia

SYNTHETIC APERTURE RADAR (SAR) AND OPTICAL IMAGERY DATA FUSION:
CROP YIELD ANALYSIS IN SOUTHEAST ASIA
S. M. Parks“
* Exelis Visual Information Solutions, 4990 Pearl East Circle, Boulder, Colorado, 80304
Working Group VII/6
KEY WORDS: Remote Sensing, SAR, Radar, Data Fusion, LandSAT ETM+, ENVISAT ASAR, ENVI, SARscape
ABSTRACT:
With the expanding energy crisis and rising food prices, crop yield analysis in Southeast Asia is an increasingly important topic in
this region. Rice is the most important food crop in Southeast Asia and the ability to accurately predict crop yields during a growing
season is useful for decision-makers, aid providers, and commercial trade organizations. The use of optical satellite image data by
itself is difficult due to the almost constant cloud in many parts of Southeast Asia. However, Synthetic Aperture Radar (SAR), or
SAR data, which can image the Earth's surface through cloud cover, is suitable for many agricultural purposes, such as the detection
of rice fields, and the identification of different crop species. Crop yield analysis is difficult in this region due to many factors. Rice
cropping systems are often characterized by the type of rice planted, the size of rice field, the sowing dates for different fields,
different types of rice cropping systems from one area to another, as well as cultural practices such as sowing and transplanting. This
paper will discuss the use of SAR data fused with optical imagery to improve the ability to perform crop yield analysis on rice crops
in Southeast Asia.
1. INTRODUCTION
Food source security is a major concern, particularly in Asia,
due to the rapid population expansion happening in that
region. Accurate evaluations of food crops can be difficult in
many countries due to the lack of information available
regarding yields. The regular usage of optical satellite image
data for crop yield analysis is difficult due to the almost
constant cloud in many parts of Southeast Asia. However,
SAR data, which can image the Earth's surface through
cloud cover, is suitable for many agricultural purposes, such
as the detection of rice fields, and the identification of
different crop species. Crop yield analysis is difficult in this
region due to many factors. Rice cropping systems are often
characterized by the type of rice planted, the size of rice
field, the sowing dates for different fields, different types of
rice cropping systems from one area to another, as well as
cultural practices such as sowing and transplanting.
Space-borne radar imagery has great potential for the
delineation and monitoring of rice crop paddies. SAR
images have proven to be suitable for many agricultural
remote sensing purposes, for example, detection of
agricultural land such as rice fields, and even identification
of different crop species is possible using well-timed SAR
images (Karjalainen, Kuittinen, Junnikkala, Karvonen,
Nguyen, & Tran, 2010). Due to the abundance of cloud
coverage in Southeast Asia, SAR data is often a better choice
over optical data (Abu Bakar, Shaari, Chuah, & Ewe, 1997).
Multi-temporal and multi-sensor data fusion has also been
successfully used to identify irrigated rice fields. Rice is
often planted in paddy fields and grows in distinct stages,
including germination, emergence, tillering, heading, and
maturing (Wang, 2009). Rice backscatter coefficients in
SAR imagery display higher temporal variation than other
types of land cover (Wang, 2009). Rice in the planting stage
exhibits a lower backscatter coefficient from flooded water
because the rice plant is short and sparse, whereas, in the
tillering stage, the backscatter coefficient increases rapidly
when more tillers emerge and develop into a denser canopy
(Wang, 2009). The backscatter coefficient is slightly less in
the late heading stage when the leaves start to dry up and
mature (Wang, 2009).
It is possible to monitor the rice growth stage during crop
yield analysis, by measuring the backscattering coefficient
from the plants as a function of time if radar images are
acquired at appropriate time intervals during the growing
season. Due to the nature of rice growth, there is a rapid
increase in biomass during the 30-day vegetative phase for
the short growth duration rice, so it is necessary to acquire
time series data on a monthly basis in order to capture the
changes in backscatter between the beginning of the planting
cycle and the end of the reproductive stage (Wang, 2009).
Generally, the beginning of a rice season would be identified
by a low backscatter in the time series when the field was
inundated while the end of the reproductive stage is
characterized by a high backscatter (Wang, 2009).
In this study, multi-temporal ENVISAT ASAR APS dual
polarization (HH, HV) data from five different dates in 2011
were used to estimate rice crop yields in the Ganges Delta of
Bangladesh. This data corresponds to the growing season of
Aman rice crops, which typically occurs from July through
November.
2. METHODS
Ten ENVISAT ASAR APS datasets (5 HH and 5 HV
datasets) in the original ESA format that were acquired over
the same track and frame (i.e. same viewing geometry) were
used in this example. The ENVISAT ASAR data were