| ESA | Earth Home | Missions | Data Products | Resources | Applications |
| 24-May-2012 |
 EO Data Access
|
|||
 |
How to Apply | ||
 |
|
|
|
| |
How to Access | ||
|
|||
| |||
| 3rd ERS SYMPOSIUM Florence 97 - Abstracts and Papers | |||
| |||
| Identification of Rice Cultivating Systems in the Mekong Delta using Multitemporal ERS SAR Images | |||
| |||
| Services |
| Site Map |
|
| Frequently asked questions |
|
| Glossary |
|
| Credits |
|
| Terms of use |
|
| Contact us |
|
| Advanced Search |  |
Delineation of Rice Cropping Systems in the Mekong River Delta using Multitemporal ERS Synthetic Aperture Radar
|
Soo Chin LIEW(a) Suan Pheng KAM(b) To-Phuc TUONG(b) Ping CHEN(a) Vo-Quang MINH(c) Laarni BALABABA(d) Hock LIM(a) |
(a)Centre for Remote Imaging, Sensing and Processing, National
University of Singapore, Lower Kent Ridge Road, Singapore 119260
Tel: (+65) 7723220, Fax: (+65) 7757717 Email: phyliew nus.edu.sg , http://www.crisp.nus.edu.sg
(b)International Rice Research Institute, P.O. Box 933, 1099 Manila, Philippines
(c)Department of Soil Science, Faculty of Agriculture, University of Can Tho, Can Tho, Vietnam
(d)Natural Resources Program, Asian Institute of Technology, Bangkok, Thailand | |
Abstract
-
The aim of this study is to use multitemporal ERS-2 satellite
synthetic aperture radar (SAR) images in delineating and mapping
areas under different rice cropping systems in the Mekong river
delta, Vietnam. Five change index maps were generated from the
six images acquired between June and December, 1996. Using a 3-dB
threshold, the pixels in each change index map were classified
into one of three classes: increasing, decreasing or constant
backscattering. The five change index maps were used to generate
a composite map with 243 possible change classes. After discarding
the minority classes, and merging the dominant classes with similar
backscatter time series, the thematic classes of rice cropping
systems practiced in the study area were delineated.
Keywords: multitemporal, synthetic aperture radar, rice, Mekong delta, agriculture
Introduction
Rice cultivation in the Mekong river delta is largely governed by hydrology, rainfall pattern and the availability of irrigation. Given the diversity of the cropping systems, it would be very difficult, if not impossible, to discriminate the areas under different planting systems using one single-date image. Multi-temporal images are required to monitor the spatial and temporal growth patterns of the rice crops and subsequently to identify the cropping systems practiced at a particular region. Multispectral visible/near infrared images such as those from the SPOT or LANDSAT satellites could be used for this purpose. Unfortunately, a large part of the rice growing season coincides with the rainy period resulting in limited availability of cloud-free images throughout the growing season. The use of cloud penetrating SAR would overcome this problem.The objective of this study is to use multi-temporal ERS-2 SAR data to delineate and map the spatial distribution of the various rice cropping systems in the Mekong River Delta. The study was confined within a 100 km by 100 km area covered by one ERS-2 SAR scene, (track 75, frame 3411 shifted 30%) which includes parts of Soc Trang and Bac Lieu provinces in the Mekong river delta, Vietnam (see Fig. 1 and inset for the location map).
Rice Cropping Systems in the Mekong Delta
There are three main rice seasons in the Mekong river delta (see Table 1). Two of the seasons (HT and M) coincide with the rainy season which typically starts in May and lasts until November. The annual rainfall in the study area varies from 1600 to 2000 mm. The three rice seasons, in various combinations governed by hydrology, rainfall pattern and availability of irrigation, constitute the variety of rice-based cropping systems practiced in the Mekong river delta [1].The major rice cropping systems in this area are the single rice crop (SC), double rice crop (DC) and triple rice crop (TC) systems (see Table 2). The single rice crop is invariably the Mua (M), which has traditionally been the rainfed rice cropping system using local, traditional varieties with a longer growth period. It is practiced mainly in the tidally-inundated coastal area subjected to salinity intrusion prior to the rainy season. The double cropping system may be the Dong Xuan-He Thu (DX-HT) or the He Thu-Mua (HT-M) system. The DX crop, planted at the end of the rainy season, needs to be irrigated. In areas affected by deep flooding during the rainy season, the second crop (HT) is planted earlier in April, depends on pump-irrigation for its water, and is harvested before the onset of the flood. In areas where flooding is not severe or commences later, the HT crop is planted in May/June and makes use of rain water. The HT-M system is typically practiced in the salinity affected areas and in areas where irrigation is not available. Both crops are rainfed. Hence the crop calendar varies each year, depending on the onset of the rainy season. The rice crops may be planted using the transplanting or the direct seeding methods. In the direct seeding method, the seeds are either sown onto dry fields prior to the start of the rainy season (dry direct seeding) or pre-germinated seeds may be sown onto wet fields (wet direct seeding). The HT crop is generally wet or dry direct seeded while the M crop may be transplanted or wet direct seeded. The triple crop system combines all three rice seasons (DX-HT-M) and is practiced in limited riverine areas with favourable hydrological conditions and availability of irrigation for the DX crop.
Radar Backscattering Characteristics of Rice Plants
Various studies on the backscattering of radar from rice plants [2, 3, 4] indicate that the radar backscatter generally increases with time after planting during the vegetative phase when there is a rapid increase in plant biomass. The backscatter saturates at the reproductive phase of the growth cycle when the plants start flowering. Theoretical simulation [4] shows that the scattering mechanism is dominated by double scattering between the water surface and the rice plants. The backscattering coefficient is found to increase from -16 dB or less at the beginning of the growth cycle when the field is inundated and there is little biomass in the field, to about -8 dB at the saturation level [4].Hence, it should be possible to monitor the rice growth stage by measuring the backscattering coefficient from the plants as a function of time if radar images are acquired at appropriate time intervals. ERS-2 SAR images can only be acquired in the same mode (either descending or ascending) at 35-day intervals. Although the backscatter time series obtained at 35-day intervals would not have sufficient temporal resolution to capture the rapid increase in biomass during the vegetative phase for the short growth duration rice, which is about 30 days, it would still be able to capture the changes in backscatter between the beginning of the planting cycle and the end of the reproductive stage. 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 characterised by a high backscatter.
Images and Data Acquisition
Seven descending mode ERS-2 synthetic aperture radar images at 35-day repeat intervals were acquired (track 75, frame 3411 shifted down by 30% to cover areas of interest in Soc Trang and Bac Lieu provinces) during the following dates in 1996: May 5, Jun 9, Jul 14, Aug 18, Sep 22, Oct 27 and Dec 1. A reasonably cloud-free SPOT multispectral image of the part of the study area was also acquired on Jun 19, 1996. This image served as a useful aid in the interpretation of the earlier radar images. Ground truthing field trips were also carried out on dates coincidental with or close to the dates of SAR image acquisition. Over 90 plots of rice fields, each about 100 x 100 m2 in size were monitored for general field conditions, date of planting and crop growth stage. A final field check was carried out after completion of the radar data processing to ascertain the delineated spatial pattern of the rice cropping systems.
Image Preprocessing
The ERS-2 scenes were acquired and processed into the calibrated SAR Precision Image (PRI) format at the ground station of the Centre for Remote Imaging, Sensing and Processing (CRISP), Singapore. Each PRI product was first converted to 8-bit images by dividing the 16-bit pixel Digital Numbers (DN) by 4, low-pass filtered using a 5x5 averaging window and then downsampled to 50-m pixel size. An edge-preserving speckle removal filter based on the adaptive Wiener filter for multiplicative noise was applied followed by a 3x3 median filter. For ease of analysis, the ocean and Cu Lao (island) Dung on the left bank of the Hau Giang (one of the branches of the Mekong River) appearing in the scene were masked out. Multitemporal colour composite images were then generated to show the changes in the backscattering coefficients during the rice growing seasons. An example of such images is shown in Fig. 1a which is composed of the ERS images of May 5, Jun 9 and Jul 14 in the red, blue and green bands respectively. The multitude of colours illustrates the variety of rice cropping systems in the study area. The areas which appear grey are where the backscattering coefficient remained relatively constant throughout the season. These areas correspond to relatively invariant surface features such as the inland melaleuca forests, the coastal mangrove forests, the linear settlements, homestead gardens and orchards along canals and roads; and the built-up areas. Two other multitemporal colour composite images are shown in Fig. 1b and Fig. 1c.
Classification Method and Results
Classification of the rice cropping systems in this region was made by thresholding the change indices derived from the series of multitemporal SAR images. For each consecutive pair of images 1 and 2, a change index map was generated. The change index (CI) was defined as the change in the backscattering coefficient expressed in dB,
(dB) = 20 log10(DN2/DN1)
where DN1 and DN2 are the pixel digital numbers of the images 1 and 2 respectively. Six change index maps were generated from the series of seven SAR images. An arbitrary threshold was then applied to each change index map to produce a threshold-change-index (TCI) map. A threshold value of 3 dB was found to be suitable for delineating the different classes of rice cropping systems. For each TCI map, the pixels were classified as having a constant (-3dB < CI < +3dB), decreasing (CI < -3dB) or increasing (CI > +3dB) backscattering over the corresponding time period. The results of thresholding the change index maps using a 3-dB threshold are shown in Fig. 2. In each of these six TCI maps, the areas with decreasing backscatter are shown in black while the white areas are areas with increasing backscatter. The grey areas are the areas with constant backscatter.
Five TCI maps (covering the period from Jun 9 to Dec 1) were used in the classification. By combining these five TCI maps, a total of 243 possible classes could be formed. Each pixel was assigned a class number according to
3i-1 ti
where ti (i = 1, 2, ..., 5) is the pixel value of the ith TCI map, which assumes one of the values 0 (constant backscatter), 1 (decreasing backscatter) or 2 (increasing backscatter). Since these classes were obtained from the threshold change indices, they will be referred to as the "change classes" in the following paragraphs.
The change class of value zero corresponds to the regions in the scene where the radar backscattering coefficient remained relatively constant throughout the period of observation. These areas are the non-rice areas, covered with relatively permanent surface features described in the previous section. Of the 243 possible classes, 30 classes were found to account for 86.7% of the pixels with non-zero class value, i.e. where the backscatter changed by 3 dB or more in at least one of the series of TCI maps. These change classes are listed in Table 3 in descending order of dominance in terms of pixel numbers.
A number of change classes were associated with and merged into
a thematic class of rice cropping system on the basis of similarity
in their time series of radar backscattering (
in dB) as well
as their geographical distribution pattern. Field knowledge about
the type of rice cropping system that might be practiced in a
particular geographical region was also considered when assigning
the thematic classes. Of the 30 dominant change classes, 24 could
be assigned to the major thematic classes of rice cropping systems.
Six change classes remained unclassified. They were very likely
the non-rice classes. The backscatter time series of the
dominant change classes were grouped and graphed for the major
thematic classes of rice cropping systems (see Fig. 3a to 3h).
The resulting thematic map of the rice cropping
systems is shown in Fig. 4.
Discussion
The major rice cropping systems listed in Table 2 (except the triple crop system) have been delineated by thresholding the change index maps derived from the multitemporal SAR images. Some variations of these major systems have also been identified. Their backscatter time series characteristics are described in the following paragraphs.
The single crop system (rainfed Mua), planted with long
growth duration rice, is predominantly confined to the coastal
areas where there is still saline intrusion during the dry season
(coloured red in Fig. 4). The backscatter time series is characterised
by a sharp drop in the backscattering coefficient between Jun
9 and Jul 14 (see Fig. 3a), corresponding with the start of the
Mua rice crop. The backscatter then increases and remains
high till the end of the monitoring period in Dec 1. The shrimp-rice
system (coloured brown in Fig. 2) is practiced in the low depression
areas, where there is late planting of Mua crop (a sharp
increase in between Aug 18 and Sep 22;
see Fig. 3b) after the
harvest of shrimps and flushing out of saline water by the rains.
There are two main variations of the irrigated, double rice cropping
(DX-HT) system. The irrigated DX-HT areas coloured yellow in
Fig. 4 are low-lying and are subjected to severe flooding in Aug-Oct.
To avoid the floods, the HT crop is planted early, in April, and
is at the earlier part of the crop season. Correspondingly the
backscatter time series shows a generally increasing from May 5, peaking at Jul 14 (see Fig. 3c). The drop in
between Jul 14 and Aug 18 is probably due to flooding after the harvest of the HT crop. The fields remain fallow until the planting of the
DX crop in December.. The increase in between Aug 18 and Sep 22 is probably due to weed emergence.
The drop in at the end
of October coincides with the peak of flooding, while the subsequent
increase in in December marks the start of the DX crop.
The backscatter time series for the area coloured magenta in
Fig. 4 is characterised by a distinct drop in
on Oct 27, followed
by a steep increase in backscatter (see Fig. 3d). The earlier
part of the profile shows a slight drop in on Jun 9, followed by a gradual increase of radar backscatter, reaching a peak on
Sep 22. This category is found predominantly in areas where the
second variation of the irrigated DX-HT cropping system occurs.
In these areas which are not so severely affected by flooding
during the rainy season, the HT crop is planted later, in May/June,
to take advantage of the rains. The increasing from July to September indicates the HT crop in the fields. The drop in
on Oct 27 may
be due to the peak floods. It was also noted that in the irrigated
area, the class number 27 (the backscatter series of which is
included in Fig. 3d) occurs mainly in the periphery of the rice
areas, just behind the linear settlements and associated homestead
orchards and miscellaneous planting. Field investigation and interviews
with the local residents reveal that these locations are associated
with the sugarcane-rice (HT) cropping system.
The magenta-coloured pixels (Fig. 4) also occur scattered in the rainfed areas, with some concentrations along the main road leading from Soc Trang to Bac Lieu. Ground observations indicate that the rainfed HT-M system is practiced in some parts of this area, where the dry direct-seeded HT crop is planted between May/June and Aug/Sep, followed by the Mua crop in Oct. Because of the similar crop calendar of the HT crop for both the irrigated DX-HT and the rainfed HT-M cropping systems, it is coincidental that these two systems exhibit similar characteristics in their backscatter time series. For the HT-M system, the increase in backscatter between Oct 27 and Dec 1 is probably due to the Mua crop, while in the case of the DX-HT system, the increase could be due to weed emergence. This backscatter time series probably also reflects the sugarcane-HT rice cropping system mentioned above.
Four other variations of the double cropping rainfed HT-M systems
which exhibit distinctive backscatter time series can be identified.
The system practiced mainly in the Bac Lieu province (coloured
light green in Fig. 4) is characterised by two distinct drops
in the backscatter time series, i.e. on Jun 9 and Oct 27 (see
Fig. 3e). In these areas which are lower-lying, the fields are
flooded in June, hence the corresponding drop in
in the Jun 9 scene. The wet field conditions allow for wet direct seeding
of the HT crop. The increase in after October corresponds with the Mua crop. The bright green pixels which appear within
the irrigated DX-HT magenta area west of Phung Hiep correspond
with swampy areas which are not cultivated. It is also coincidental
that the backscatter time profile of this swampy area resembles
that of the wet direct seeded HT-M cropping system.
On the other hand in most of Soc Trang province which is on relatively
higher ground (coloured cyan in Fig. 4), there is no distinct
dip in in the backscatter time series
(see Fig.3f); there was
no early season inundation of the rice fields. The HT crop is
mainly dry direct seeded, hence the corresponding increase in
backscatter between Jul 14 and Aug 18.. As is the case of the
light green areas, the increase in after October corresponds with the Mua crop.
In the more inland region of Bac Lieu province (coloured dark green in Fig. 4), the HT-M system is characterised by two dips in the backscatter time series on Jul 14 (late HT) and Oct 27 (Fig. 3g). Roughly east of Soc Trang town, close to the Hua Giang river, is a small but rather homogeneous area coloured dark blue (Fig. 4) which has a distinctive backscatter time series (see Fig. 3h). The cropping system in this area is similar to that of the cyan-coloured area of Soc Trang province. However as this area can be subjected to tidal inundation, the backscatter profile could be altered.
The triple crop system has not been identified. Since this system is not commonly practiced, it is most probably included under the minority classes not considered in this study.
Conclusions
In conclusion, the dominant rice cultivating systems practiced in the study area have been delineated in this study by thresholding the change index maps derived from multitemporal SAR images followed by human interpretation of the radar backscatter time series. The thresholding process produces 243 possible change classes characterised by their unique signature of radar backscatter time series. The dominant change classes are merged into one of the several thematic classes of rice cropping systems based on their similarity in the backscatter time series and their geographical distribution. However, ambiguity still exists in assigning some change classes to their respective thematic classes. For example, one of the rainfed HT-M systems has similar radar backscatter time series as the irrigated DX-HT system, and possibly with a rice-nonrice cropping system. An acquisition in April or sometime during the peak of the DX crop is probably needed to resolve this ambiguity. Use of ascending mode images may help to increase the temporal resolution of the backscatter time series. However, mixing the two modes of acquisition may complicate the interpretation of the backscatter time series due to the different orientations of the ground with respect to the SAR line of sight.
References
- T.P. Tuong, C.T. Hoanh and N.T. Khiem. 1991.
Agro-hydrological factors as land qualities in land evaluation for rice cropping
patterns in the Mekong Delta of Vietnam, in P. Deturck and
F.N. Ponnamperuma (eds.) Rice Production on Acid Soils of the
Tropics. Institute of Fundamental Studies, Kandy, Sri Lanka.
- T. Le Toan, H. Laur, E. Mougin, and A. Lopes, 1989. Multitemporal and dual-polarization observations of agricultural vegetation covers by X-band SAR images. IEEE Trans. Geosci. Remote Sensing, 27, pp. 709-717.
- T. Kurosu, M. Fujita, and K. Chiba, 1995. Monitoring of rice crop growth from space using ERS-1 C-band SAR. IEEE Trans. Geosci. Remote Sensing, 33, pp. 1092-1096.
- T. Le Toan, F. Ribbes, L.-F. Wang, N. Floury, K.-H. Ding, J. A. Kong, M. Fujita, and T. Kurosu, 1997. Rice crop mapping and monitoring using ERS-1 data based on experiment and modeling results. IEEE Trans. Geosci. Remote Sensing, 35, pp. 41-56.
Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry