Determination of Swamps Area Suitable for Paddy Field Using Remote Sensing Approach in Banyuasin Regency, South Sumatera Province

STATEMENT

I, Estri Rahajeng, hereby stated that this thesis entitled:

DETERMINATION OF SWAMPS AREA SUITABLE FOR PADDY FIELD USING REMOTE SENSING APPROACH IN BANYUASIN REGENCY, SOUTH SUMATERA PROVINCE

Are results of my own work during the period of February until July 2011 and that it has not been published before. The content of this thesis has been examined by the advising committee and external examiner.

Bogor, August 2011

Estri Rahajeng G051090131

ABSTRACT

ESTRI RAHAJENG. Determination of Swamps Area Suitable for Paddy Field Using Remote Sensing Approach in Banyuasin Regency, South Sumatera Province. Under the Supervision of I WAYAN ASTIKA and HARTANTO SANJAYA.

Swamps area become more and more important for Indonesia and will be the future for agricultural development potential outside of Java because mainly in Java, there is a continuous loss of agricultural lands due to urbanization, industry and roads infrastructure. The spatial and temporal distribution of swamps area is an important parameter to be correctly characterized in order to get the information about the area suitable for agricultural purpose. Remote sensing techniques can be used to obtain the spatial distribution of swamps area over a large area, reducing expensive and time consuming of field measurement.

The objective of this research is to develop an identification method for determining swamps area suitable for paddy field. The method was then applied in area of interest in Banyuasin Regency. This research was intended to integrate between Normalized Difference Water Index (NDWI) analysis, Land Surface Temperature (LST) analysis and land covers analysis by applying the supervised classification.

The swamps identification method was tested by using Image acquisition on 15 April 2000 imagery with acquisition date on 15 April 2000 and Image acquisition on 16 May 2006 with acquisition date on 16 May 2006. The suitability analysis of swamps area for paddy field based on land system data only for the constricted factors data that are peat depth, pH, slope and salinity data.

For image acquisition on 15 April 2000, an accuracy of 73,5% was obtained with LST value range of 23oC - 33oC and NDWI value range of -0.35 – 0.1. For image acquisition on 16 May 2006, an accuracy of 63.5% was obtained with LST value range of 23oC - 33oC and NDWI value range of -0.43 – 0. The suitability classes of swamps area for paddy field in whole area of interest were found to be 54% area for S1 (highly suitable) or 3,426 km2, 40.2% for S2 (moderately suitable) class or 2,550 km2 and for S3 (marginally suitable class) is around 5.8% area or 369 km2.

Keywords : NDWI, LST, suitability analysis, swamps.

ABSTRAK

ESTRI RAHAJENG. Penentuan Lahan Rawa Yang Sesuai Untuk Lahan Sawah Menggunakan Pendekatan Penginderaan Jauh di Kabupaten Banyuasin, Provinsi Sumatera Selatan. Bimbingan I WAYAN ASTIKA dan HARTANTO SANJAYA.

Lahan rawa menjadi semakin penting bagi Indonesia dan merupakan potensi masa depan untuk pembangunan pertanian di luar Pulau Jawa karena di Pulau Jawa telah banyak pengurangan lahan pertanian akibat urbanisasi, industri dan pembangunan infrastruktur jalan. Distribusi spasial dan temporal dari lahan rawa merupakan parameter penting yang harus ditentukan dengan tepat dalam rangka memperoleh informasi wilayah rawa yang cocok untuk pertanian. Penginderaan jauh dapat digunakan untuk memperoleh distribusi spasial lahan rawa di daerah yang luas sehingga dapat menggantikan pengukuran lapangan yang memerlukan biaya mahal dan memakan waktu yang lama.

Tujuan penelitian ini adalah untuk membangun metode identifikasi penentuan wilayah rawa yang cocok untuk sawah. Metode tersebut kemudian diterapkan di Kabupaten Banyuasin. Metode ini mengintegrasikan antara Indeks Normalisasi Perbedaan Air (NDWI) dan Suhu Permukaan Tanah (LST) serta tutupan lahan menggunakan klasifikasi terbimbing.

Metode ini dicoba dengan menggunakan Image acquisition on 15 April 2000 dengan waktu akuisisi pada tanggal 15 April 2000 dan Image acquisition on 16 May 2006 dengan waktu akuisisi pada tanggal 16 Mei 2006. Analisa kesesuaian lahan dilakukan dengan menggunakan data faktor pembatas dalam budidaya padi di lahan rawa yaitu kedalaman gambut, pH, kemiringan lahan (slope) dan salinitas.

Untuk citra akusisi 15 April 2000, akurasi tertinggi sebesar 73,5% diperoleh pada rentang nilai LST sebesar 23oC - 33oC dan nilai NDWI -0,35 – 0,1. Sedangkan citra akuisisi 16 May 2006, akurasi tertinggi sebesar 63,5% diperoleh pada rentang nilai LST dari 23oC - 33oC dan nilai NDWI -0,43 - 0. Hasil akhir berupa kesesuian lahan rawa untuk sawah menunjukkan bahwa untuk keseluruhan wilayah uji coba mempunyai kelas S1 (sangat sesuai) sebesar 54 % atau 3.426 km2, kelas S2 (cukup sesuai) sebesar 40,2% atau 2.550 km2 dan untuk kelas S3 (kurang sesuai) sebesar 5,8 % atau 369 km2.

Kata kunci : NDWI, LST, analisa kesesuaian, rawa.

SUMMARY

ESTRI RAHAJENG. Determination of Swamps Area Suitable for Paddy Field Using Remote Sensing Approach in Banyuasin Regency, South Sumatera Province. Under the Supervision of I WAYAN ASTIKA and HARTANTO SANJAYA.

Swamps area become more and more important for Indonesia and will be the future for agricultural development potential outside of Java. Mainly in Java, there is a continuous loss of agricultural lands for urbanization, industry and roads infrastructure. The spatial and temporal distribution of swamps area is an important parameter to be correctly characterized in order to get the information about the area suitable for agricultural purpose.

South Sumatera government has set Banyuasin Regency as one of the development centre of rice crops. Nearly 80 percent of the Banyuasin Regency is swamps where the areas are marginal lands unsuitable for industrial purposes, but potential for food crops, rice, and coconut, also interesting for the plantation area, such as oil palm. Determination swamps area suitable for agriculture by using terrestrial method needs considerable amount of time, resources and cost. The information obtained from terrestrial method is also limited only at the point observation and certain time period.

The objective of this research is to develop an identification method for determining swamps area suitable for paddy field. The method was then applied in Banyuasin Regency. This research was intended to integrate between Normalized Different Water Index (NDWI) and Land Surface Temperature (LST) analysis as well as land cover analysis by applying supervised classification. The suitability analysis of swamps area for paddy field based on land system data only for the constricted factors data that are peat depth, pH, slope and salinity data

The method was tested by using Image acquisition on 15 April 2000 with acquisition date on 15 April 2000 and Image acquisition on 16 May 2006 with acquisition date on 16 May 2006. The study area is a subset from full scene image acquisition on 15 April 2000 and acquisition on 16 May 2006. In order to get LST value for 2 sets satellite imagery, the LST processing only used thermal band (Band 61 and Band 62 for Image acquisition on 15 April 2000 and Band 6 for Image acquisition on 16 May 2006). The steps to do LST analysis are converting digital number (DN) to radiance and then converting radiance to brightness temperature. Meanwhile, NDWI analysis use of reflected near-infrared radiation (band 4) and visible green light (band 2) to enhance the presence of such features while eliminating the presence of soil and terrestrial vegetation features. The maximum likelihood method was used for the classification process. This

method is based on the priority of type coverage. Suitability analysis uses peat depth, pH, slope and salinity data by applying overlay processing.

For image acquisition on 15 April 2000 the LST range values are 2oC to 35oC and mean value is 22.8oC. For image acquisition on 16 May 2006 the range values are 3oC to 37oC and mean value is 25.4oC. For NDWI result, image acquisition on 16 May 2006 has the same characteristics with the image acquisition on 15 April 2000, whereas water class having the positive value range and bigger value than others. It indicates that the water presence in the nature such as water body, ocean and inundation area always having positive value. NDWI value indicates high correlation with moisture content of land cover. Bigger NDWI value means bigger the moisture content of land cover than others.

Image acquisition on 15 April 2000 was classified into the 9 classes namely cloud shadow, forest, paddy field, shrub, settlement, bare land, water, inundation area and cloud. Shrub area is the biggest area with the percentage 29.20% and followed by paddy field with percentage 21.00%. Image acquisition on 16 May 2006 was also classified into the 9 classes namely water, mangrove, forest, settlement, paddy field, shrub, bare land, inundation area, and fish pond. Paddy field area is the biggest area with the percentage 23.00% and followed by shrub area with percentage 16.00%. Result of classification is noted that have more classes than those of interest (swamps and not swamps). Next, the pixels of each input image were reclassified to get the swamps and not swamp classes. For image acquisition on 15 April 2000, swamps areas were formed from bare land, inundation area, shrub and paddy field. The others were formed as not swamps areas. Meanwhile for image acquisition on 16 May 2006, swamps areas were formed from bare land, inundation area, shrub, mangrove and paddy field. The others were formed as not swamps areas.

Reclassification process was also done to the LST and NDWI result. In this process, some threshold values was applied to get the good accuracy in reclassifying area become 2 types : swamps and not swamps type. Threshold I was applied LST value 23oC – 33oC and NDWI value -0.43 to 0. Threshold II was applied LST value 23oC – 33oC and NDWI value -0.35 to 0.1. Threshold III was applied LST value 12oC – 20oC and NDWI value -0.4 to 0.2.

The swamps area distribution resulted from reclassification process, image acquisition on 16 May 2006 had smaller area proportion than image acquisition on 15 April 2000. It’s caused by the acquisition date of the image influencing the temperature and water distribution in whole area of interest. It implies to the large area of swamp distribution. Swamps distribution obtained from image acquisition on 15 April 2000 with threshold II analyzing had the area around 6,389 km2 and swamps area distribution obtained from image acquisition on 16 May 2006 with threshold I had the area is around 5,493 km2.

Overall accuracy of determination swamp area based on the LST, NDWI and supervised classification is still in moderate accuracies. For image acquisition

on 15 April 2000, highest accuracy is 73,5% for the threshold II with LST value range of 23oC - 33oC and NDWI value range of -0.35 – 0.1. Meanwhile for Image acquisition on 16 May 2006, highest accuracy is 63.5% for the threshold I with LST value range of 23oC - 33oC and NDWI value range of -0.43 – 0.

The most problem occurred in the swamps determination is in the image classification processing. In this step, the initial cluster gathering was undertaken using unsupervised classification. The resulting clusters were later edited, assigned names and used in final supervised classification of the image as a base reference for each spectral. Training area was formed by using visual interpretation of the image referring to the each spectral from unsupervised classification. Some land cover types have the similar spectral, but based on field knowledge and visual interpretation is different land cover types. In image acquisition on 15 April 2000, one constraint in image classification is haze. It caused the spectral value of two different land cover type almost similar for example paddy field and shrub have the similar spectral value therefore in visual interpretation it makes confusion among others. Likewise in image acquisition on 16 May 2006, settlement and bare land have the similar spectral value. For the next processing, the classification results would be reclassified into 2 type of land cover that are the swamps and not swamps area. Misclassification in the image classification process can caused the error in determining the swamps area in whole area of interest.

The percentage of suitability classes for paddy field based on land system for whole area of interest were S1 (highly suitable) around 52.2%, S2 (moderately suitable) around 41.8% and S3 (marginally suitable) around 6.07%. This result was overlaid with the swamps distribution from previous method to get the distribution swamps area suitable for paddy field. Suitability classes for paddy field were S1 (highly suitable) class covered 54% area or 3,426 km2 , S2 (moderately suitable) class covered 40.2% area or 2,550 km2 and for S3 (marginally suitable) class around 5.8% area or 369 km2. There is no area that included in N (not suitable) class.

This method can be applied as an alternative method in determining alternative area for wetland to support the agricultural development. Referring to the limitation of the method for improving of the accuracy, it is suggested to use some other thresholds value range more detail in order to get the more variety of result possibility. The accuracy assessment is suggested to do in each step by using more detail of ground truth data.

Keywords : NDWI, LST, analisa kesesuaian, rawa.

Copyright ® 2011, Bogor Agricultural University Copyright are protected by law,

1. It is prohibited to cite all or part of this thesis without referring to and mentioning the source

a. Citation only permitted for the sake of education, research, scientific writing, report writing, critical writing or reviewing scientific problem.

b. Citation does not inflict the name and honor of Bogor Agricultural University.

2. It is prohibited to republish and reproduce all part of this thesis without the written permission from Bogor Agricultural University

DETERMINATION OF SWAMPS AREA SUITABLE FOR

PADDY FIELD USING REMOTE SENSING APPROACH

IN BANYUASIN REGENCY, SOUTH SUMATERA PROVINCE

ESTRI RAHAJENG

A thesis submitted for the degree Master of Science

In Information Technology for Natural Resources Management Study Program

GRADUATE SCHOOL

BOGOR AGRICULTURAL UNIVERSITY BOGOR

2011

Research Title : Determination of Swamps Area Suitable for Paddy Field Using Remote Sensing Approach in Banyuasin Regency, South Sumatera Province.

Student Name : Estri Rahajeng

Student ID : G051090131

Study Program : Master of Science in Information Technology for Natural Resources Management

Approved by, Advisory Board

Dr. Ir. I Wayan Astika, M.Si Hartanto Sanjaya, S.Si, M.Sc Supervisor Co-Supervisor

Endorsed by,

Program Coordinator Dean of the Graduate School

Dr. Ir. Hartrisari Hardjomidjojo, DEA Dr. Ir. Dahrul Syah, M.Sc.Agr

Date of Examination: August 8th, 2011 Date of Graduation:

ACKNOWLEDGEMENTS

First of all, I would like to express my earnest gratitude to Alloh SWT for the favor, mercy and blazing me to carry out this thesis.

I would like to express my thanks and sincere appreciation to Ministry of Agriculture (Kementerian Pertanian), Indonesia for giving me the opportunity to continue my studies and providing the financial support essential for overall expenses during the whole study at Master of Science in Information Technology for Natural Resources Management, Bogor Agricultural University.

I would like to express my deepest gratitude to Dr. I Wayan Astika as Supervisor and Hartanto Sanjaya, M.Sc as Co-Supervisor for encouraging, guidance, supporting and kindly provided me the valuable advice. Moreover, I would like to express my gratitude to the external examiner Dr. M. Buce Saleh for giving advice and valuable correction to make my thesis acceptable.

I am also thanks to all responsible persons from the Public Work Office and Spatial and Regional Planning Board of Banyuasin Regency also Lowland-Wetland and Coastal Area Data Information Center, Sriwijaya University, South Sumatera for supporting necessary data related to my research.

I am obliged to many persons who have given me encouragement and beneficial inspiration in realizing the completion of this research (MIT 2009 student, MIT secretariat and all my colleagues in MIT program).

Most of all, I would like to dedicate this thesis to all my lovely family (my family in Lampung and my big family in Palembang), especially my lovely husband M. Daud Rusdi, SKM. MKM and my lovely children (Nadiyah Tzurayya, M. Hilmiy Rasyid and M. Hashif Rifqiy) that providing patient, understanding, and encouragement during my study. Without their sincere support, I never could have finished this thesis.

Hopefully, this thesis could give positive and valuable contribution for anyone who read it.

Bogor, August 2011 Estri Rahajeng

CURRICULUM VITAE

The author, Estri Rahajeng was born on January 13th, 1975 in Kotabumi, Lampung, Indonesia. She finished her Elementary, Junior and High School in Lampung, Indonesia and passed in 1992. Her bachelor degree was received in Agricultural Engineering Department, Faculty of Agricultural Technology, Bogor Agricultural University in 1997.

In 2009, she was selected as state scholarship to study in Master of Science in Information Technology for Natural Resources Management, Bogor Agricultural University and received the M.Sc degree in 2011.

As for working experience, since 2005 to 2008 she worked as trainer in Jambi Agricultural Training Center, Ministry of Agriculture. Then, since 2008 to present, she has been working in Development and Agricultural School, Ministry of Agriculture in Sembawa, Banyuasin Regency, South Sumatera Province.

TABLE OF CONTENTS

Page

LIST OF TABLES ... xiv 

LIST OF FIGURES ... xvi 

LIST OF APPENDICES ... xviii 

I.  INTRODUCTION ... 1 

1.1 Background ... 1 

1.2 Problem Statement ... 4 

1.3 Research Objectives ... 4 

1.4 Scopes of Research ... 4 

II.  LITERATURE REVIEW ... 5 

2.1 Swamps Potential in Indonesia ... 5 

2.2 Remote Sensing Application ... 7 

2.2.1 Image Rectification ... 9 

2.2.2 Image Enhancement ... 11 

2.2.3 Image Classification ... 11 

2.3 Landsat TM Characteristics ... 12 

2.4 Land Suitability ... 13 

2.5 Previous Related-Researches ... 15 

III. METHODOLOGY... 19 

3.1 Time and Location ... 19 

3.2 Data Preparation ... 21 

3.2.1 Data Acquisition ... 21 

3.2.2.  Image Data Pre-Processing ... 22 

3.2.3 Vector Data Pre-Processing ... 23 

3.3 Framework of Research ... 24 

3.3.1 Swamps Identification ... 24 

3.3.2 Suitability Analysis ... 30 

3.3.3 Accuracy Assessment ... 31 

IV. RESULT AND DISCUSSION ... 33 

4.1 Swamps Identification ... 33 

4.1.1 LST Analysis ... 34 

4.1.2 NDWI Analysis ... 36 

4.1.3 Image Classification ... 38 

4.1.4 Spatial Analysis ... 41 

4.2 Accuracy Assessment ... 50 

4.3 Suitability Analysis ... 52 

V.  CONCLUSIONS AND RECOMMENDATIONS ... 59 

5.1 Conclusions ... 59 

5.2 Recommendations ... 59 

REFERENCES ... 61

APPENDICES ... 65 

LIST OF TABLES

Page

Table 1 Swampy land resource in four major islands (in thousand hectares) ... 6 

Table 2 Suitable tidal swamps in four major islands (in thousand hectares) ... 6 

Table 3 Thematic mapper spectral bands ... 13 

Table 4 Land suitability order ... 14 

Table 5 Land suitability classes ... 15 

Table 6 Swamps reclamation area in Banyuasin ... 19 

Table 7 ETM and TM thermal band calibration constants ... 26 

Table 8 Suitability criteria for paddy field ... 31 

Table 9 Land use area summary report for base map ... 34 

Table 10 Summary of LST value for each imagery ... 34 

Table 11 Minimum, maximum and mean NDWI value for Image acquisition on 15 April 2000 ... 36 

Table 12 Minimum, Maximum and Mean NDWI value for image acquisition on 16 May 2006 ... 37 

Table 13 Land covers area summary report for image acquisition on 15 April 2000 ... 40 

Table 14 Land covers area summary report for image acquisition on 16 May 2006 ... 40 

Table 15 Percentage of swamps and not swamps area for each image ... 42 

Table 16 Threshold in swamps area analysis* ... 44 

Table 17 Result probability for each threshold values ... 45 

Table 18 Percentage area of swamps and not swamps for image acquisition on 15 April 2000 ... 47 

Table 19 Percentage area of swamps and not swamps for image acquisition on 16 May 2006 ... 49 

Table 20 Accuracies of determination results under each threshold. ... 52 

Table 21 Suitability classes under constricted factors. ... 52  Table 22 Swamps area suitable for paddy field ... 57

LIST OF FIGURES

Page

Figure 1 Electromagnetic spectrum ... 7 

Figure 2 Electromagnetic interactions ... 9 

Figure 3 Research area ... 20 

Figure 4 Detail of area of interest ... 20 

Figure 5 Landsat 7 imagery acquisition on 15 April 2000... 23 

Figure 6 Landsat 5 imagery acquisition on 16 May 2006 ... 23 

Figure 7 General flowchart for the research ... 24 

Figure 8 Image classification (Lillesand and Kiefer, 2000) ... 28 

Figure 9 Swamps identification framework ... 29 

Figure 10 Suitability analysis framework ... 30 

Figure 11 Base map of swamps analysis ... 33 

Figure 12 LST distribution of image acquisition on 15 April 2000 ... 35 

Figure 13 LST distribution of image acquisition on 16 May 2006 ... 35 

Figure 14 NDWI distribution for image acquisition on 15 April 2000... 37 

Figure 15 NDWI distribution for image acquisition on 16 May 2006 ... 38 

Figure 16 Classification result for image acquisition on 15 April 2000 ... 39 

Figure 17 Classification result for image acquisition on 16 May 2006 ... 39 

Figure 18 Swamps classification for image acquisition on 15 April 2000 ... 42 

Figure 19 Swamps classification for image acquisition on 16 May 2006 ... 42 

Figure 20 Swamps distribution of image acquisition on 15 April 2000 under threshold I ... 46 

Figure 21 Swamps distribution of image acquisition on 15 April 2000 under threshold II ... 46 

Figure 22 Swamps distribution of image acquisition on 15 April 2000 under threshold III ... 47 

Figure 23 Swamps distribution of image acquisition on 16 May 2006 under

threshold I ... 48 

Figure 24 Swamps distribution of image acquisition on 16 May 2006 under threshold II ... 48 

Figure 25 Swamps distribution of image acquisition on 16 May 2006 under threshold III ... 49 

Figure 26 Reference map in swamps determination ... 50 

Figure 27 Percentage of suitability classes under constricted factors ... 53 

Figure 28 Suitability map based on peat depth factor ... 53 

Figure 29 Suitability map based on pH factor ... 54 

Figure 30 Suitability map based on slope factor ... 54 

Figure 31 Suitability map based on salinity factor... 55 

Figure 32 Suitability classes for paddy field ... 56 

Figure 33 Swamps distribution from image processing ... 56 

Figure 34 Swamps area suitable for paddy field ... 57 

Figure 35 Comparison between suitability map and reclamation area ... 58 

xviii

LIST OF APPENDICES

Page

Appendix 1 Ground truth data ... 65 

Appendix 2 Error matrix for 2 classes in the each threshold value for each satellite imagery ... 66

Appendix 3 LST distribution ... 69 

Appendix 4 NDWI distribution ... 71 

Appendix 5 Image classification result ... 73 

Appendix 6 Swamps area suitable for paddy field... 75 

Appendix 7 Suitability classes in reclamation area... 76 

I.

INTRODUCTION

1.1 Background

Indonesia is one of the dense population countries in the world. In the last three decades, the annual population growth rate in the country has declined from 2,1 % in 1971 to 1,5 % in 2000. However due to the young structure of the population, the number of people still rose from 119 million to 205 million during the same period. The population is estimated to reach about 300 million in 2025 (BPS, 2010). The increasing number of population has been followed by the increasing in urbanization. According to United Nation Population Fund Agency (UNFPA) (2003), the percentage of people living in urban areas has risen from 17,2% in 1971 to 42,2% in 2000. This was at the expense of the annual decrease of the rural population from 1,8% to 0,3%.

The government has put food security to be one of the highest priorities in agricultural development. Securing food and livelihood is apparently linked to the exploitation of the natural resources base (land, water and forest) for instance by reclaiming swamps. The pressure of intense human activity has encouraged the development to move from arable land to marginal spaces such as slopping area, swampy area and coastal region. To develop these marginal areas for agricultural production requires careful measures that minimize constraints of erosion, land slide, land subsidence, acidity hazards, tidal stream and sea level rise.

In view of swamps for agriculture, so far swamps reclamation has been carried out as a gradually long term process. Swamps areas by their nature are generally unsuitable for development, which is mainly caused by the soil condition, water logging and regular/permanently covered with water of their environmental values. Due to these conditions in order to optimize land productivity as well as to protect the areas from flooding, the development of swamps areas needs careful alternative approaches and techniques in land and water management.

The development of swamps for agriculture has been carried out in various part of the world for many centuries. More recent large scale swamps

development projects have in tidal swamps in South Asia such as Indonesia, Malaysia, Thailand and Vietnam. So far, the projects were carried out as gradually long term process or step-wise development strategy. This strategy was based on some considerations, which are: 1) limited availability of construction budget as well as the need to reclaim large areas 2) lack of knowledge, experience and design criteria in this field 3) social cultural background of transmigration people who most of them are coming from dry land and are no familiar with the wet tidal land conditions

Since 1960, the Indonesia government has been reclaiming swamps and most of them are tidal swamps. The objectives have been 1) to increase the national food production, mainly rice; in order to obtain self-sufficiency 2) to provide agricultural land for transmigrates, in order to support the government transmigration program 3) to support regional development 4) to increase income per capita and 5) to increase the security on coasts along the border line (Suryadi, 1996).

The development of swamps has to create a suitable condition for agriculture and settlement. From the 34 million ha of existing and potential swamps in four major islands in Indonesia (Sumatera, Kalimantan, Papua and Sulawesi), about 5,5 million ha or 14 % has been identified as very suitable for agriculture development by government 2,1 million ha, 39% of the 5.5 million ha has been reclaimed (1,1 million ha by the central government and the rest by local government, local community and migrants).

The major objective of the 1,1 million ha swamps reclamation conducted in the period 1960s-1990s by the government was resettlement of people from densely region in Java, Bali, Madura islands. Since that time almost no new reclamation were carried out. Regarding the population growth and the need of increasing the food production, the new development of agricultural land and most importantly optimizing the productivity of the existing land is essential if Indonesia is be able to meet the food demand of its population.

Swamps become more and more important for Indonesia and will be the future for agricultural development potential outside of Java because mainly in Java, there is a continuous loss of agricultural lands for urbanization, industry and

roads infrastructure. Rice production on Java will continue to decrease in future as the urban requirements or land and water are in strong competition with the requirements for rice cultivation and it is estimated that the rate of the loss of agricultural land is about 30,000 to 40,000 ha/year.

The development of swamps in Sumatera has started from 1960s by Directorate of Swamps under The Directorate Generale of Water Resources Development, in the project called “Proyek pembukaan persawahan pasang surut (P4S). Swamps cover about 30% out of 8.7 million ha of South Sumatera province. About 300.000 ha had been reclaimed from 1969-1988 with more than 50.000 households resettled on the sites. The crops yield of rice was satisfactory in the period of 1981-1990. However, in the next period the settlers found some difficulties with agriculture on the drained swamps included water delivery as lack of water in the dry season and pests. Furthermore, many development projects suffer from excessive soil problems i.e low soil fertility and high acidity, hence limited by hydro-topography of adjoining river and basin.

The spatial and temporal distribution of swamps area is an important parameter to be correctly characterized. Numerous applications rely on information about swamps area especially for swamps land from soil characteristic until micro water management aimed at optimizing best management practices in agricultural management. Remote sensing techniques can be used to obtain the spatial distribution of swamps land over large area, reducing expensive and time consuming field measurement.

Remote sensing approach is the only way for consistent mapping of overwhelming proportion, if not all of the swamps of the world. This will need development of methods and datasets for rapid delineation of swamps, to map spatial distribution, and to identify their specific characteristics such as biophysical, ecological and socio-economic. Referring to the wetland delineation, the US Army Corps of Engineers (USACE) in the Wetland Delineation Manual (1987) also supports such delineation without field visit: “in a routine wetland determination when the quality and quantity of information obtained are sufficient for wetland determination, onsite inspections of the study area may not be necessary”. However, at larger spatial scales, applicability of remote sensing

techniques could vary significantly at different localized areas due to the higher degree of variability in the spectral signatures of the associated ground features. Thereby, the need to investigate methods that can consistently map wetlands over large area becomes important (Kulawardhana et al., 2007).

1.2 Problem Statement

The government has set South Sumatera Province as one of the rice centre in Indonesia. As the implication of the program, the local government of South Sumatera has set Banyuasin Regency as one of the development centre of rice crops. Nearly 80 percent of the Banyuasin Regency is swamps where the areas are marginal lands unsuitable for industrial purposes, but potential for food crops, rice, and coconut, also interesting for the plantation area, such as oil palm.

Determination swamps area suitable for agriculture by using terrestrial method needs a considerable amount of time, resource and cost. The information obtained from terrestrial method is also limited only at the point observation and certain time period. In order to get updating information about “where is the swamps area suitable for paddy field”, it is important to have the alternative method for solving the problem.

This research will be used to complete the study about swamps area in Banyuasin Regency besides the study was already done in Lowland-Wetland and Coastal Area Data Information Center, Sriwijaya University, South Sumatera.

1.3 Research Objectives

The objective of this research is to develop an identification method for determining swamps area suitable for paddy field. The method was then applied in Banyuasin Regency.

1.4 Scopes of Research

Scopes of the research are:

1. This research was intended to integrate between Normalized Different Water Index (NDWI), Land Surface Temperature (LST) and land cover analysis by applying supervised classification.

2. The suitability classes of swamps area for paddy field based on the land system data.

II.

LITERATURE REVIEW

2.1 Swamps Potential in Indonesia

Swamps are a form of wetland with some flooding of large areas of land by shallow bodies of water. A swamps generally covered by vegetation, or vegetation that tolerates periodical inundation. The two main types of swamps are "true" or swamps forest and "transitional" or shrub forest. The water of swamps may be fresh water, brackish water and sea water.

Swamps area is considered as low or level water or any broad expanse of land with a general low level. This type of land is relatively flatter or lower than adjacent land. The typical characteristic of swamps is inundation for some time due to poor drainage, where seasonal flooding may occur. Swamps in Indonesia become more important remaining land resources for the development of food crops. The development of swamps has to create a suitable condition for agriculture and settlement. Almost 9 million ha of tidal swamps is indicated potential for agriculture and more than 3,6 million ha has been reclaimed. Most of reclaimed swamps are in Sumatera and Kalimantan (Bappenas 2006; Wignyosukarto 2006).

In general, the swamps development in Indonesia is driven by transmigration program, whereby major reclamation projects were implemented in order to settle farmers from densely isles of Java, Madura and Bali. In the time period, the issues of food security and food sufficiency become the main objective of swamps development and management, which is to fulfill the food requirement by intensifying the existing reclaimed swamps area (Suryadi 1996; Euroconsult 1997).

Swamps in Indonesia can be grouped into three categories (Suryadi 1996) : 1. Tidal swamps. These swamps are located along the coasts and along lower

reaches of rivers where the river regimes are dominated by tidal fluctuation. They include a generally narrow zone of mangrove, followed by extensive fresh water swamps. Land elevations are generally around the tidal high water level. From water management point of view these area are characterized by shallow inundation in the wet season, caused mainly by stagnant rainwater.

The daily low tide in the rivers offers in the principle good opportunities for drainage of excess water. In the certain areas the high tide offers opportunities for tidal irrigation

2. Non tidal swamps, beyond the zone of tidal swamps seasonal fluctuations in the river water levels are more pronounced and may cause deep inundation of lands in the wet season. The absence of daily low water in the rivers requires adopted approaches for drainage. In the many cases flood protection will be required.

3. Inland swamps, separated from the above swamps by surrounding uplands. These swamps cover relatively small areas.

Commonly, the swamps are found in the coastal zone (tidal swamps) and river floodplain (non tidal, inland swamps). Table 1 has identified approximately 39 million swampy land in the whole country, 65% of these area is located in Sumatera and Kalimantan and 35% of swamps area can be found in Papua and Sulawesi. Moreover, up to 65% of the swampy land is tidal swamps and 37% is inland swamps. Table 2 shows the suitable tidal swamps for agricultural cropland in four major islands.

Table 1 Swampy land resource in four major islands (in thousand hectares) Type of Sumatera Kalimantan Papua Sulawesi Total

Tidal swamps 9.771 7.054 7.798 84 24.707

Inland swamps 3.440 5.710 5.181 385 14.716

Total 13.211 12.764 12.979 469 39.423

Source : Settlement and Regional Infrastructure Department and Rijks Waterstaat Netherlands. 2002

Table 2 Suitable tidal swamps in four major islands (in thousand hectares)

Type of Sumatera Kalimantan Papua Sulawesi Total

Not Cultivated 1.380 1.392 2.808 12 5.599

Cultivated 2.062 1.460 6 72 3.600

Source : Settlement and Regional Infrastructure Department and Rijks Waterstaat Netherlands. 2002

2.2 Remote Sensing Application

Remote sensing is the science and art of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device that is not in contact area or phenomenon under investigation (Lillesand and Kiefer 2000).

Figure 1 Electromagnetic spectrum

The earth’s surface and atmosphere emit individual characteristic signatures within the visible light and electromagnetic radiation spectrum. The spectrum is divided into spectral bands ranging from short gamma rays to long radio waves. As indicated, most remote sensing devices make use of electromagnetic energy. However, the electromagnetic spectrum is very broad and not all wavelengths are equally effective for remote sensing purposes. Furthermore, not all have significant interactions with earth surface materials of interest to us. Figure 1 illustrates the electromagnetic spectrum. The atmosphere itself causes significant absorption of shorter wavelengths such as the ultraviolet (UV). As the result, the first significant window (i.e., a region in which energy can significantly pass through the atmosphere) opens up in the visible wavelengths. Even here, the blue wavelengths undergo substantial attenuation by atmosphere scattering, and are thus often left out in remote sensed images. However, the green, red and near-infrared (IR) wavelengths all provide good opportunities for gauging earth surface interactions without significant interference by the atmospheric. In addition, these regions provide important clues to the nature of many earth materials. Chlorophyll, for example is very strong absorber of red visible wavelengths, while

the near-infrared wavelengths provide important clues to the structures of plant leaves. As a result, the bulk remotely sensed images used in GIS-related applications are of taken in these regions.

Extending into the middle and thermal infrared regions, a variety of goods windows can be found. The longer of the middle infrared wavelengths have proven to be useful in a number of geological applications. The thermal regions have proven to be very useful for monitoring not only the obvious cases of the spatial distribution of heat from industrial activity, but a broad set of applications ranging from fire monitoring to animal distribution studies to soil moisture conditions.

After the thermal IR, the next area of major significance in environmental remote sensing is in the microwave region. A number of important windows exist in this region and area of particular importance for the use of active radar imaging. This can thus be used as a supplement to information gained in other wavelengths, and also offers the significant advantage of being usable at night (because as an active system it is independent of solar radiation) and in regions of persistent cloud cover (since radar wavelength are not significantly affected by clouds).

When electromagnetic energy strikes a material, three typed of interaction can follow: reflection, absorption and/or transmission. It can be seen in Figure 2. Our main concern is with the reflected portion since it is usually this which is returned to the sensor system. Exactly how much is reflected will vary and will depend upon the nature of the material and where in the electromagnetic spectrum our measurement is being taken. The result can be characterized as a spectral response pattern or sometimes called as a signature. Most human are familiar with spectral response patterns since they are equivalent to human concept of color.

Finding distinctive spectral response patterns is the key to most procedures for computer-assisted interpretation of remotely sensed imagery. This task is rarely trivial. Rather, the analyst must find the combination of spectral bands and the time of year at which distinctive patterns can be found for each of the information classes of interest.

Figure 2 Electromagnetic interactions

Remote sensing is used extensively to gather measurements. Satellite-based systems can measure phenomena that change continuously over time and cover large, often inaccessible areas (Aronoff 1991). The ideal of perfect remote sensing system has yet to be developed. Consequently, error creeps into the data acquisition process can degrade the quality of the remote sensor data collected. These errors may have an impact on the accuracy of the subsequent human or machine-assisted image analysis. Therefore, it is usually necessary to pre-process that remotely sensed data prior to analyze it to remove some of these errors (Lillesand and Kiefer 2000).

Digital image processing involves the manipulation and interpretation of digital images with the aid of a computer. The central idea behind digital image processing is quite simple. The digital image is fed into a computer one pixel at a time. The computer is programmed to insert these data into an equation, or series of equations and store the result of computation for each pixel (Lillesand and Kiefer 2000).

The procedures of digital image processing are following some broad types of computer assisted operations: image rectification, image enhancement, image classification.

2.2.1 Image Rectification

Image rectification are operations aiming at correcting distorted or degraded image data, which stem from image acquisition; to create a more faithful

representation of original scene. The procedures of image rectification consist of geometric correction, radiometric correction and noise removal.

a. Geometric Correction

Raw digital images usually contain geometric distortions so significant that they cannot be used as maps. The geometric correction process is normally implemented as two-step procedure. First, those distortions those are systematic or predictable. Second, those distortions that are essentially random or unpredictable are considered (Lillesand and Kiefer 2000).

As systematic distortions are constant and predictable they do not constitute a problem to the user of satellite imagery. The agencies that supply the imagery do the corrections. The main systematic distortions are panoramic (or scanner), distortion, scan skew and change in scanning velocity (Meijerink et al. 1994).

Systematic distortions are well understood and easily corrected by applying formulas derived by modeling the sources of the distortions mathematically. Random distortions and residual unknown systematic distortions are corrected by analyzing well-distributed ground control points (GCPs) occurring in an image. As with they counterparts on aerial photographs, GCPs are features of known ground location that can be accurately located on digital imagery. Some features that make good control points are highway intersections and distinct shoreline features (Lillesand and Kiefer 2000).

b. Radiometric Correction

The radiometric correction is necessary to remove variations in the radiometry that influence the compatibility of multi-temporal/multi-sensor data. The sources of radiometric distortions in VIR (Visible InfraRed) data are

• Atmosphere (atmosphere scattering, absorption) • Sensor (stripping)

• Sun illumination

Image enhancement is procedures that applied to image data in order to effectively display or record the data for subsequent visual interpretation.

Steps that most commonly applied digital enhancement technique can be categorized as contrast manipulation, spatial features manipulation or multi image manipulation.

2.2.2 Image Enhancement

Image enhancement procedures are applied to image data in order to more effectively display or record the data for subsequent visual interpretation. Normally, image enhancement involves techniques for increasing the visual distinction between features in a scene. The objective is to create “new” images from the original image data in order to increase the amount of information that can be visually interpreted from data (Lillesand & Kiefer 2000).

2.2.3 Image Classification

The overall objective of image classification procedures is to automatically categorize all pixels in an image into land cover classes or themes. Normally, multispectral data are used to perform the classification and indeed, the spectral pattern present within the data for each pixel is used as the numerical basis for categorization. That is, different feature types manifest different combination of DNs based on their inherent spectral reflectance and emission properties. In this light, a spectral pattern is not all geometric in character. Rather, the term pattern refers to the set of radiance measurements obtained in the several of classification procedures that utilizes this pixel by pixel spectral information as the basis for automated land cover classification.

A classification describes the systematic framework with the names of the classes and the criteria used to distinguish them, and the relation between classes. Classification of remotely sensed data is used to assign corresponding levels with respect to groups with homogeneous characteristics, with the aim of discriminating multiple objects from each other within image. Classification thus necessarily involves definition of class boundaries that should be clear, precise, possibly quantitative, and based on objective/criteria.

Supervised classification is the procedure most often used for quantitative analysis of remote sensing image data. It rests upon using suitable algorithm to label the pixels in an image as representing particular ground cover types, or

classes. In supervised classification, this is realized by an operator who defines the spectral characteristics of the classes by identifying sample areas (training areas). Supervised classification requires that the operator must be familiar with the areas of interest. The operator needs to know where to find the classes of interest in the area covered by the image. This information can be derived from general area knowledge or from dedicated field observations (Janssen and Goerte 2000).

2.3 Landsat TM Characteristics

The kind of Landsat that are useful for image interpretation for much wider range applications is Landsat Thematic Mapper ™. The characteristic of Landsat Thematic Mapper ™ which first loaded on Landsat 4 in 1982 was designed to provide improved spectral and spatial resolution over the Multi Spectral Scanner (MSS) instrument. Landsat Thematic Mapper ™ is designed to capture electromagnetic in 7 spectral bands. It has three bands in visible region (band 1, 2 and 3), one band in near infra red (band 4), two bands in mid infra red (band 5 and 7) and one band in thermal infra red (band 6). All bands have 30 m spatial resolution except for for Image acquisition on 15 April 2000, support for TM imagery with the addition of a co-registered 15 m panchromatic band (Lillesand and Kiefer 2000).

Landsat TM image is useful for image interpretation of a much wider range of applications than other satellite images. This is due to the Landsat has both an increase in the number of spectral bands and an improvement in spatial resolution. A list of the seven spectral bands of the Landsat TM, along with a brief summary of the intended principal application of each, that shown in Table 3.

Table 3 Thematic mapper spectral bands Band Wavelength (µm) Nominal Spectral Location Principal Applications

1 0,45-0,52 Blue Designed for water body

penetration, making it useful for coastal water mapping. Also useful for soil/vegetation discrimination, forest type mapping.

2 0,52-0,60 Green Designed measure green

reflectance peak of vegetation discrimination and vigor assessment.

3 0,63-0,69 Red Designed to sense in chlorophyll absorption region aiding in plant species differentiation.

4 0,76-0,90 Near-Infrared Useful for determining vegetation types, vigor, and biomass content, for delineating water bodies and for soil moisture discrimination. 5 1,55-1,75 Mid-Infrared Indicative of vegetation moisture.

Also useful for differentiation of snow from clouds.

6 10,4-12,5 Thermal Infrared Useful in vegetation stress analysis, soil moisture discrimination and thermal mapping applications.

7 2,08-2,35 Mid-Infrared Useful for discrimination of mineral and rock types. Also sensitive to vegetation moisture content.

Source : (Lillesand and Kiefer 2000)

2.4 Land Suitability

Land evaluation is the process of assessing of land performance when the land is used for specified purposes (FAO, 1976). The land is the ultimate source of wealth and the foundation on which civilization is constructed. Due to the benefit of the land, then are merged efforts to utilize it. Land evaluation leads to

rational land use planning and appropriate and sustainable use of natural and human resources. Land suitability represents a method of land evaluation.

Land suitability analysis estimates which areas suitable or not suitable for certain development. The land suitability can be determined by using matching methods between land suitability criteria and land characteristics. The process of land suitability classification is the appraisal and grouping of specific areas of lands in terms of their suitability for a defined use. The suitability is the aptitude of a given type of land to support a defined use.

To produce the land suitability, two concept of land evaluation are known, i.e. physiographic approach and parametric approach. Physiographic approach utilizes landform framework to identify the natural unit, while parametric approach divides the land following the distinguish land value and its combination. Parametric approach is more suitable for this research, due to all of parameters that are quantized. In this study two categories are recognized: orders and classes. The orders indicate whether or not given types of land are suitable for the concerned land utilizations type and are expressed by symbols S and N. It can be seen in Table 4.

Table 4 Land suitability order

Order Remarks S (Suitable) Land on which sustained use is expected to yield

benefits which justify the inputs, without un acceptable risk of damage to land resources

N (Not Suitable) Land whose qualifies appear to preclude sustainability for the considered land use

Source : FAO, 1976

Classes reflect degrees of sustainability within the order “suitable”. Normally three classes are recognized:

Table 5 Land suitability classes

Classes Remarks S1

(highly suitable)

Land which has no significant or only minor limitations to the sustained application of the given land utilization.

S2 (moderately suitable)

Land which has limitations that are moderately severe for sustained application of the given land utilization. The limitations will reduce productivity or benefits and will increase the required inputs

S3 (marginally suitable)

N (not suitable)

Land which has severe limitations for sustained application of the land utilization

The limitations are so severe that they preclude the successful application of the given land utilization

Source : FAO, 1976

Because it is based only on physical aspect of suitability orders, there will be no differentiation between N1 and N2. In this case, the “not suitability” of land for paddy field will be assumed as N (not suitable).

2.5 Previous Related-Researches

Land surface temperature is sensitive to vegetation and soil moisture, hence it can be used to detect land use and land cover changes, e.g. tendencies towards urbanization, desertification etc.

Hanqiu Xu. (2007) proposed a technique to extract urban built up land features from Landsat Thematic Mapper ™ and Enhanced Thematic Mapper (ETM+) imagery using three indices, Normalized Difference Built up (NDBI), Modified Normalized Difference Water Index (MNDWI) and Soil Adjusted Vegetation Index (SAVI) respectively. These indices is used to represent three major urban land use classes, built up land, open water body and vegetation, respectively. As a result, the spectral signatures of the three urban land use classes are more distinguishable in the new composite image than in the original seven band image as the spectral clusters of the classes are well separated.

Nugroho et al. (2007) present detecting tidal flood pattern with Landsat TM remote sensing data in South Sumatera coastal area. The method in the research is using transformation image that included Maximum Likelihood, Principle Component Analysis (PCA) and Tasseled Cap (all in IDRISI version Window 2.0). The result stated that the humidity of earth surface is one of the results from the transformation moist.

Guha and Lakshmi (2002) proposed one way of determining soil moisture contents from remotely sensed data is by using the thermal emissions from soils in the microwave range, generally sensitive to moisture variations in the top five cm of the soil. Saturated surfaces emit low levels of microwave radiation, whereas dry soils emit much higher levels of microwave radiation. However, in many applications it is difficult to separate the microwave signal from saturated and unsaturated soil due to competing effects of moisture content, surface roughness, vegetation, liquid precipitation, and complex topography unless the variables are known a priori (Schmugge 1985; Bindlish et al. 2003).

Hence, an extensive amount of calibration is necessary to fit the parameters and prior knowledge of the surface cover and state must be known (Kerr 2007). Indeed, Wagner et al. (2007) state that microwave remote sensing systems can capture the general trends in surface soil moisture conditions, but cannot be used to estimate absolute soil moisture values. A more promising approach to obtain soil moisture variability is to remotely sense the greenness variations of biomass within an otherwise homogeneous canopy (Yang et al. 2006), because variations in soil water directly affect the growth patterns of the overlying vegetation.

Frazier et al. (2000) applied the simple digital image processing techniques to map riverine water bodies with Image acquisition on 16 May 2006 TM imagery. The image classification method used a single density slicing then compared to a 6-band maximum likelihood classification over the same area. The water boundaries delineated by each of these digital classification procedures were compared to water boundaries delineated from color aerial photography acquired on the same day as the TM data.

Frazier et al. (2003) applied satellite imagery in many studies that seek to relate river flow to floodplain inundation. This method accounted for rapid

variation in daily discharge using before-flood and after-flood sequences of Landsat TM imagery in reference to predefined wetland vector coverage. This procedure established a relationship between wetland inundation and river discharge.

Owor et al. (2007) applied remote sensing approach to detect seasonal change and inundation in the wetland using NDWI, NDVI and unsupervised classification supported with the ground check. Change detection showed a slight decrease in vegetated and exposed wetlands areas from five set Landsat satellite imagery (August 1987, January 1995, September 1999, March 2001, December 2001).

Yu et al. (1998) did the automated identification of swamps land incorporating Landsat TM Image and GIS data. This research describes an ongoing project on the potential study of Landsat TM for the monitoring of wetland resources with a concern of peat deposit. Preliminary results show the following improvements that are: 1) The outlines of the swamp are clearly drawn out by incorporating geomorphical data and the accuracy is reasonable 2) The estimation of peat deposit could be improved with the DEM 3) The whole procedure can be easily and automatically repeated when new data are available.

III.

METHODOLOGY

3.1 Time and Location

This research was conducted from February until July 2011 in part of the swamps reclamation area in Banyuasin Regency which geographic location between 2o16’48’’ - 3o13’48’’ South and 104o46’48’’ - 105o32’24’’ East.

Swamps reclamation areas in Banyuasin Regency consist of several locations that can be seen in Table 6.

Table 6 Swamps reclamation area in Banyuasin

No Location Area (ha)

1 Pulau Rimau 40,263

2 Delta Telang I 26,680

3 Delta Telang II 13,800

4 Delta Upang 8,423

5 Delta Saleh 19,780

6 Sugihan Kiri 49,557

7 Sugihan Kanan 29,835

8 Padang Sugihan 51,000

9 Kumbang Padang 14,227

10 Gasing Puntian 6,900

11 Air Rengit 2,411

12 Air Limau 2,576

13 Air Senda 6,730

14 Tenggulang 8,794

15 Bertak I 4,600

16 Bertak II 5,100

17 Karang Agung Hulu 9,000

18 Karang Agung Hilir 20,317

19 Karang Agung Tengah 30,000

Source : Public Work Office of Banyuasin Regency, 2010

Area of interest for this research are consisted of 7 schemes of reclamation schemes in Banyuasin Regency that are Telang I, Delta Upang, Sugihan Kiri, Sugihan Kanan, Delta Saleh, Padang Sugihan and Kumbang Padang. This area is crossed by Anak Telang, Anak Kumbang and Anak Musi River. The detail of area of research can be seen in Figure 4.

Research area

Figure 3 Research area

Figure 4 Detail of area of interest

The climate type of the area is tropical monsoon, hot and humid throughout the year maximum temperature ranging from 29oC - 32oC, minimum temperature from 21oC - 22oC and humidity around 84% - 89%. The wet months (over 200 mm rainfall per month) occur during November – April and average driest month is August (less than 100 mm rainfall per month). Average annual rainfall is about 2.400 mm. According to the Oldeman classification, the agro climate is C-1, with the 5 to 6 wet months (rainfall over 200 mm) and 0-1 dry months (rainfall less than 100 mm). The climate and rainfall type supports a range of crops.

The macro topography is generally flat or gently sloping towards natural drainage outlets. At micro level, the topography is very irregular and disturbed as a result of reclamation activities, e.g. the subsidence of mineral soils and the oxidation of peat. Human activities such as leveling and construction of sorjans further contributed to the disturbance of the original topography.

3.2 Data Preparation

3.2.1 Data Acquisition

 

The remotely sensed data used in this research are a Image acquisition on 15 April 2000 +ETM image (path 124, row 62) acquired on 15 April 2000 and a Image acquisition on 16 May 2006 TM (path 124, row 62) acquired on 16 May 2006. The data is obtained from the United States Geological Survey (USGS) website. The area of interest of image acquisition on 15 April 2000 can be seen in Figure 5 and for image acquisition on 16 may 2006 can be seen in Figure 6.

Other data used in the research were Swamps Distribution Map acquired in 2009 from Public Work Office of Banyuasin Regency, Rupa Bumi Indonesia acquired on 2003 from National Coordination Agency for Surveys and Mapping, Land System Map acquired on 1989 from Indonesian Center for Agricultural Land Resources Research and Development, Ministry of Agriculture.

Ground data obtained from ground truth using GPS Garmin was done in the Delta Upang, Delta Telang I and Muara Sugihan.

3.2.2. Image Data Pre-Processing

 

The study area was a subset from full scene image acquisition on 15 April 2000 and image acquisition on 16 May 2006. Some pre-processing was done for each image before to do the swamps analysis. The pre-processing image included:

A. Geometric Correction

Raw digital images usually contain geometric distortions; it cannot be used as maps. Geometric correction is needed because usually the raw digital image contains geometric distortions caused by variation in the surface ground curve, altitude, attitude of sensor platform, etc. The intent of geometric correction is to compensate the distortions introduced by these factors so that the corrected image will have the geometric integrity of a map.

Geometric correction was done using digital map as reference and done by :

1. Collection of GCPs. These are defined as points that are clearly identifiable on both the satellite imagery and reference image (vector data). The GCPs should be widely distributed and the RMS error not more than 0,5. In this research, Landsat imagery was registered using topographic map (scale 1:50,000). In selecting the GCPs, one has to be careful, not only one should check that the object selected on the two images is one and the same, but one also has to be sure that the two have the same location on each image.

2. Rectification : Rectify data set using Polynomial (Control Point). 3. Resampling : Using Nearest Neighbor Resampling.

B. Radiometric Correction

The effect of atmospheric scattering caused by water molecules is a problem in imagery that should be eliminated or minimized to avoid bias for each spectral band. Histogram or atmospheric adjustment is one of the ways to minimize this bias.

C. Image Cropping

Image cropping was done in order to extract the study area, because the original image covers a large area, while the study is only part of that

image. Cropping the image is needed because the study area is limited only in centre of swamps reclamation area.

Figure 5 Landsat 7 imagery acquisition on 15 April 2000

Figure 6 Landsat 5 imagery acquisition on 16 May 2006

3.2.3 Vector Data Pre-Processing

 

Most the data for suitability analysis were secondary data, which included in Land System Data. The data were peat depth, pH, salinity and slope map. That map represents features being necessary to GIS process.

3.3 Framework of Research

The frame work for this research included swamps identification procedures and land suitability analysis. General framework for this research can be seen in Figure 7. It is combination of Remote Sensing and Geographic Information System. The research has been done under two phases as follows:

• Swamps Identification. This section was done by using satellite imagery analysis based on the different characteristic between 2 satellite imagery with different season acquisition. Referring to the archive image availability, image acquisition on 15 April 2000 is used as the image in wet season and image acquisition on 16 May 2006 is used as the image in dry season. The framework of this section can be shown in Figure 9.

• Land Suitability Analysis. This section was done by using spatial data to produce the suitability classes for the swamps area that resulted from the difference area between 2 satellite imagery results after some processing from previous phase.

General flowchart for this research can be seen in Figure 7.

Figure 7 General flowchart for the research

3.3.1 Swamps Identification

Swamps identification was done by calculating and analyzing 3 parameters that are:

A. Calculating LST

Land surface temperature is how hot the “surface” of the Earth would feel to the touch in a particular location. From a satellite’s point of view, the “surface” is whatever it sees when it looks through the atmosphere to the ground. It could be snow and ice, the grass on a lawn, the roof of a building, or the leaves in the canopy of a forest. Thus, land surface temperature is not same as with air temperature that is included in the daily weather report. In order to get LST value for 2 sets satellite imagery, the LST processing only used thermal band (Band 61 and Band 62 for Image acquisition on 15 April 2000 and Band 6 for Image acquisition on 16 May 2006).

According to Ghulam (2010), the steps to do the LST analysis are:

a. Converting DN to Radiance

The Landsat satellite imagery has Digital Number (DN) values range between 0 and 255.

The value of Lmin and Lmax is obtained from the header files. Open the .MET file using Word Pad or any text editor.

……….. (1)

‰ L is the radiance value for band i;

‰ Lmin is the minimum spectral radiance (the spectral radiance that is scaled to QCALMIN in watts/(meter squared * ster * m) can be seen from the header file;

‰ Lmax is the maximum spectral radiance ((the spectral radiance that is scaled to QCALMAX in watts/(meter sq)) can be seen from the header file;

‰ DN is the digital number; ‰ Qcalmax = 255; and ‰ Qcalmin = 1.

b. Converting Radiance to Brightness Temperature

The steps to do this process using Planck’s Radiance Function

Where, C1=1.19104356×10-16 W m2; C2=1.43876869×10-2 m K

In the absence of atmospheric effects, T of a ground object can be theoretically determined by inverting the Planck’s function as follows:

This equation can be reformed as :

Let K1 = C1/ 5 , and K2 = C2/ , and satellite measured radiant intensity B (T) = L , then above mentioned equation is collapsed into an equation similar to the one used to calculate brightness temperature from Landsat TM image :

Therefore, K1 and K2 become a coefficient determined by effective wavelength of a satellite sensor. The value of K1 and K2 can be seen in Table 7.

Table 7 ETM and TM thermal band calibration constants

……….. (2)

……….. (3)

………... (4)

………... (5)

K1 (Wm-2sr-1µm-1) K2 (Kelvin)

Image acquisition 666.09 1282.71

Image acquisition 607.76 1260.56

Source : Ghulam, 2010

B. Calculating NDWI

The Normalized Difference Water Index (NDWI) was developed to delineate open water in satellite imagery and enhance their presence in remotely-sensed digital imagery (McFeeters 1996). The NDWI makes use of reflected near-infrared radiation and visible green light to enhance the presence of such features

while eliminating the presence of soil and terrestrial vegetation features. The NDWI formula is :

……….……... (6)

where GREEN is a band that encompasses reflected green light such as band 2 and NIR represents reflected near infrared band such as band 4. The selection of these wavelengths was done to : (1) maximize the typical reflectance of water features by using green light wavelengths (2) minimize the low reflectance of NIR by water features (3) take advantage of the high reflectance of NIR by vegetation and soil features (Mc Feeters 1996). As a result, water features are enhanced owing to having positive values and vegetation and soil are suppressed due to have zero or negative values. Image processing software can easily be configured to delete negative values. This effectively eliminates the terrestrial vegetation and soil information and retains the open water information for analysis. The range of NDWI is then from zero to one. Multiplying that value by a scale factor (e.g., 255) enhances the resultant image for visual interpretation.

C. Supervised Classification

Supervised classification methods are based upon prior knowledge of the image, specifically the statistical nature of the spectral classes used to classify the image (Mather, 1986). As the preliminary step in supervised classification method, collecting training sample in image that will be classified is very important task.

In this research, the maximum likelihood method was used for the classification process. This method is based on the priority of type coverage. Maximum likelihood method is one of effective methods and will give a good image if the size and representative of the sample data used in the training area. Other terms of similar meaning are in situ data or collateral data, but both refer to sample data gathered in order to establish a relationship between the sensor response and particular surface condition. It is commonly used to determine the accuracy of categorized data obtained through classification.

28 Supervised spectral classification was used to automatically group pixels of multispectral images into groups of predefined classes, based on the variation of reflectance among bands. The general image processing can be seen in Figure 8.

Figure 9 Swamps identification framework

3.3.2 Suitability Analysis

The suitability assessment was conducted by considering the land system analysis. According to Hardjowigeno et al. (2001), the suitability criteria for paddy crops are divided into highly suitable (S1), moderately suitable (S2), marginally suitable (S3), and not suitable (N). The factors that analyzed to classify the suitability class are limited only the constricted factors for paddy field in swamps area namely peat depth, pH, slope and salinity. Table 8 describes the suitability criteria of the constricted factors for paddy field in swamps area. Figure 10 shows the suitability analysis framework.

Figure 10 Suitability analysis framework

Table 8 Suitability criteria for paddy field

Factors S1 S2 S3 N

Peat depth (cm) <50 < 100 100-150 >150-200

pH 5.5-7.0 >7.0-8.0 or

4.5-5.5 >8.0-8.5 or 4.0-4.5 >8.5 or <4.0 Slope (%)

< 3 3-8 > 8-15 > 15-25

Salinity (mmhos/cm)

< 3.5 3.5-5.0 > 5.0-6.6 > 6.6-8.0

Source : Hardjowigeno et al., 2001

Overlaying for vector datasets is conducted to obtain an aggregate of layers that determines the suitability. In vector data, the map overlay operation is done in pairs. For a more than two layers to be overlaid, it will be taken several steps. In this research includes four map layers. So, to make its process integrated into single process, the each vector data were converted into raster data in ArcGIS 9.3. After that by using weighted overlay processing with the same weighting for each suitability maps, the combination layer map was produced and can be treated as single map to the next processing.

3.3.3 Accuracy Assessment

Methods for automated land features mapping are concluded with an accuracy assessment of their result (Congalton, 1991). The validation sampling points is used on this account. Accuracy assessment is done to verify the analysis and classification result using “point sampling accuracy” approach according to “confusion matrix”. The confusion matrix summarizes the relationship between two datasets that are classification map and reference information or alternative model. The accuracy assessment for this research use formulation below:

a. Omission Error (as known as Producer’s Accuracy)

Takes into account the accuracy of individual classes; indicates the percentage of the time a particular land cover type on the ground was identified as that land cover

type on the map. It expresses how well the map producer identified a land cover type on the map from the satellite imagery data.

……….……... (7) O = (Xii / X+i) x 100%

Xii = total number correct cells in a class

X+i = sum of cell values in the row

O = Omission error

b. Commission Error ((as known as User’s Accuracy)

Takes into account the accuracy of individual classes; indicates the percentage of the time a particular land cover type on the map is really that land cover type on the ground. It expresses how well a person using the map will find that land type on the ground.

C = (Xii / X+i) x 100%

Xii = total number correct cells in a class

X+i = sum of cell values in the column

C = Commission error

c. Overall Accuracy

Summarizes the total agreement/disagreement between the maps; only incorporates the major diagonal and excludes the omission and commission errors. It indicates how well the map identifies all land cover types on the ground.

……….……... (8)

……….……... (9) A = (D / N) x 100%

D = total number correct cells as summed along the major diagonal N = total number of cells in the error matrix.

A = Overall accuracy

IV.

RESULT AND DISCUSSION

4.1 Swamps Identification

The first step in the swamps identification is to define the areas considered as swamps by using a GIS layer of the “base swamps extent” was created by extracting of the swamps feature from Rupa Bumi Indonesia. The base swamps extent layer is used as a template to compare each image set in the LST, NDWI and image classification processing.

Figure 11 Base map of swamps analysis

Figure 11 shows the land use distribution in area of interest based on the Rupa Bumi Indonesia with scale 1:50,000 acquired on 2003. Land use area summary report for the base map can be seen in Table 9. Swamps area is the largest area with the area 4,869 km2 or 67,19% from whole area of interest.

Ghulam A. 2010. Calculating Surface Temperature Using Landsat Thermal Imagery. Department of Earth and Athmospheric Sciences. Saint Louis University.

Guha A, Lakshmi L. 2002. Sensitivity, Spatial Heterogeneity and scaling of C-Band microwave brightness temperatures for land hydrology studies, IEEE Trans. Geosci. Remote Sensing.

Hanqiu Xu. 2007. Extraction of Urban Built-up Land Features from Landsat Imagery Using a Thematic-oriented Index Combination Technique. Photogrammetric Engineering & Remote Sensing Vol 73, No. 12. 1381-1391.

Hardjowigeno S, Widiatmaka. 2001. Kesesuaian Lahan dan Perencanaan Tataguna Tanah. Soil Department. Agriculture Faculty, Bogor Agricultural University.

Imanudin, M.S. 2006a. Strategi Pengelolaan Lahan Untuk Peningkatan Index Pertanaman di Lahan Pasang Surut. Proceeding Seminar “Peran & Prospek Pengembangan Rawa dan Pembangunan Nasional ”Jakarta, 27-28 November 2006. Dep. PU. Dirjen SDA. Jakarta.

Janssen LLF, Goerte BGH. 2000. Digital Image Classification in Principle of Remote Sensing. ITC, Enschede, The Netherlands.

Karnieli A et al. 2009. Use of NDVI and Land Surface Temperature for Drought Assessment: Merits and Limitations. Journal of Climate. American Meteorological Society.

Kerr YH. 2007. Soil Moisture from Space. Where are we?. Hydrogeology Journal. 15, 117-120

Kulawardhana RW et al. 2007. Evaluation of the Wetland Mapping Methods using Landsat ETM+ and SRTM Data. Journal of Spatial Hydrology Vol 7, No 2.

Lillesand TM, Kiefer RW. 2000. Remote Sensing and Image Interpretation. Fourth Edition. John Willey and Son.Inc, New York.

[LWMTL] Land and Water Management Tidal Lowland. 2005a. Technical Guideline on Tidal Swamps Development. Volume I: General Aspects, Report of the Joint Indonesian - Netherlands Working Group, Jakarta, Indonesia.

[LWMTL] Land and Water Management Tidal Lowland. 2005b. Technical Guideline on Tidal Swamps Development. Volume II: Water Management, Report of the Joint Indonesian - Netherlands Working Group, Jakarta, Indonesia.

[LWMTL] Land and Water Management Tidal Lowland. 2005c. Technical Guideline on Tidal Swamps Development. Volume III: Operation and Maintenance, Report of the Joint Indonesian - Netherlands Working Group, Jakarta, Indonesia.

Mather PM. 2004. Computer processing of remotely-sensed images: an introduction. John Wiley and Sons Ltd. 3rd ed. 324 pp.

McFeeters SK. 1996. The use of the normalized difference water index (NDWI) in the delineation of open water features. International Journal of Remote Sensing 17:7, 1425-1432

Meijerink AMS, de Brouwer HAM, Mannzerts CM, Venezuela CR. 1994. Introduction to the use of Geographic Information System for Practical Hydrology. ITC and UNESCO International Hydrological Programme, Netherlands.

Munyati C. 2000. Wetland Change Detection on The Kafue Flats, Zambia, by Classification of a Multitemporal Remote Sensing Image Dataset. International Journal of Remote Sensing Vol 21, No. 9, 1787-1806

[NCSS] Natural Cooperative Soil Survey. 1980. Soil Survey of Pecos County. Texas. United States Department of Agriculture, Soil Conservation Service (USDA-SCS) and with Texas Agricultural Experiment Station. Texas. USA

Nugroho K, Wiradisastra US, Arsyad S, Pawitan H, Sudarsono. 2007. Detecting Tidal Flood Pattern with Landsat TM Remote Sensing Data in South Sumatera Coastal Area. Jurnal Tanah. Balai Besar Sumber Daya Lahan Pertanian. Kementerian Pertanian. Indonesia.

Owor M, Muwanga A, Pohl W. 2007. Wetland Change Detection and Inundation North of Lake George, Western Uganda Using Landsat Data. African Journal of Science and Technology, Vol. 8 No.1 pp 94-106

Richard JA. 1993. An introduction to Remote Sensing Digital Image Analysis. Second Edition. Spinger-Verlag. Berlin.

Schmugge T. 1985. Remote Sensing of Soil Moisture in Hydrological Forecasting, edited by Anderson, M.G. and Burt, T.P., Wiley, New York. Settlement and Regional Infrastructure and Rijks WaterstaatNetherlands. 2002.

Informasi Umum tentang Rawa Pasang Surut di Indonesia. http://www.tidal-lowlands.org/ind/General.htm (accessed July 2011).

Sitorus SRP.1995. Evaluasi Sumber Daya Lahan. Penerbit Tarsito. Bandung. Indonesia

Suryadi FX. 1996. Soil and Water Management Strategies for Tidal Swamps in Indonesia. PhD Disertation. Delft University of Technology-IHE Delft. Balkema, Rotterdam, Netherlands.

[UNFPA] United Nation Population Fund. 2003. UNFPA-Indonesia. http://unfpa-indonesia (accessed on February 2011)

[USGS] United States Geological Survey. 2011. Landsat Search and Download. Department of The Interior. United States. http://landsat.usgs.gov (accessed February 2011).

[USACE] US Army Corps of Engineers. 1987. Wetland Delineation Manual-Technical Report. Department of the Army. Washington, DC. 127-241 Wagner N, Naeimi V, Scipal K, de Jeu R, Martinez-Fernandez J. 2007. Soil

Moisture from Operational Meteorological Satellite. Hydrogeology Journal., 15, 121-131

Wang JR, Schmugge TJ. 1980. An Empirical Model for The complex dielectric Permitivitty of Soil as a function of Water Content, IEEE Trans . Geosci. Remote Sensing., 18 (4).

Wignyosukarto BS. 2006. The Hydraulic Performance of Tidal Irrigation System in The Reclamation of Acid Sulphate Soil. http://budiws.wordpress.com/2008/01/bs-wignyosukarto (accessed on February 2011)

Yang JP, Ding YJ. Chen RS. 2006. Spatial and Temporal of Variations of Alpine vegetation cover in the source regions of the Yangtze and Yellow River of the Tibetan Plateau from 1982 to 2001. Environmental Geology., 50(3) Yu G et al. 1998. Automated Identification of Swamp land Incorporating

Landsat Image and GIS Data. Chengdu Sub Centre of Agriculture Remote Sensing, Ministry of Agriculture. JingJusi 20, Chengdu China.

SUMMARY

ESTRI RAHAJENG. Determination of Swamps Area Suitable for Paddy Field Using Remote Sensing Approach in Banyuasin Regency, South Sumatera Province. Under the Supervision of I WAYAN ASTIKA and HARTANTO SANJAYA.

Swamps area become more and more important for Indonesia and will be the future for agricultural development potential outside of Java. Mainly in Java, there is a continuous loss of agricultural lands for urbanization, industry and roads infrastructure. The spatial and temporal distribution of swamps area is an important parameter to be correctly characterized in order to get the information about the area suitable for agricultural purpose.

South Sumatera government has set Banyuasin Regency as one of the development centre of rice crops. Nearly 80 percent of the Banyuasin Regency is swamps where the areas are marginal lands unsuitable for industrial purposes, but potential for food crops, rice, and coconut, also interesting for the plantation area, such as oil palm. Determination swamps area suitable for agriculture by using terrestrial method needs considerable amount of time, resources and cost. The information obtained from terrestrial method is also limited only at the point observation and certain time period.

The objective of this research is to develop an identification method for determining swamps area suitable for paddy field. The method was then applied in Banyuasin Regency. This research was intended to integrate between Normalized Different Water Index (NDWI) and Land Surface Temperature (LST) analysis as well as land cover analysis by applying supervised classification. The suitability analysis of swamps area for paddy field based on land system data only for the constricted factors data that are peat depth, pH, slope and salinity data

The method was tested by using Image acquisition on 15 April 2000 with acquisition date on 15 April 2000 and Image acquisition on 16 May 2006 with acquisition date on 16 May 2006. The study area is a subset from full scene image acquisition on 15 April 2000 and acquisition on 16 May 2006. In order to get LST value for 2 sets satellite imagery, the LST processing only used thermal band (Band 61 and Band 62 for Image acquisition on 15 April 2000 and Band 6 for Image acquisition on 16 May 2006). The steps to do LST analysis are converting digital number (DN) to radiance and then converting radiance to brightness temperature. Meanwhile, NDWI analysis use of reflected near-infrared radiation (band 4) and visible green light (band 2) to enhance the presence of such features while eliminating the presence of soil and terrestrial vegetation features. The maximum likelihood method was used for the classification process. This

method is based on the priority of type coverage. Suitability analysis uses peat depth, pH, slope and salinity data by applying overlay processing.

For image acquisition on 15 April 2000 the LST range values are 2oC to 35oC and mean value is 22.8oC. For image acquisition on 16 May 2006 the range values are 3oC to 37oC and mean value is 25.4oC. For NDWI result, image acquisition on 16 May 2006 has the same characteristics with the image acquisition on 15 April 2000, whereas water class having the positive value range and bigger value than others. It indicates that the water presence in the nature such as water body, ocean and inundation area always having positive value. NDWI value indicates high correlation with moisture content of land cover. Bigger NDWI value means bigger the moisture content of land cover than others.

Image acquisition on 15 April 2000 was classified into the 9 classes namely cloud shadow, forest, paddy field, shrub, settlement, bare land, water, inundation area and cloud. Shrub area is the biggest area with the percentage 29.20% and followed by paddy field with percentage 21.00%. Image acquisition on 16 May 2006 was also classified into the 9 classes namely water, mangrove, forest, settlement, paddy field, shrub, bare land, inundation area, and fish pond. Paddy field area is the biggest area with the percentage 23.00% and followed by shrub area with percentage 16.00%. Result of classification is noted that have more classes than those of interest (swamps and not swamps). Next, the pixels of each input image were reclassified to get the swamps and not swamp classes. For image acquisition on 15 April 2000, swamps areas were formed from bare land, inundation area, shrub and paddy field. The others were formed as not swamps areas. Meanwhile for image acquisition on 16 May 2006, swamps areas were formed from bare land, inundation area, shrub, mangrove and paddy field. The others were formed as not swamps areas.

Reclassification process was also done to the LST and NDWI result. In this process, some threshold values was applied to get the good accuracy in reclassifying area become 2 types : swamps and not swamps type. Threshold I was applied LST value 23oC – 33oC and NDWI value -0.43 to 0. Threshold II was applied LST value 23oC – 33oC and NDWI value -0.35 to 0.1. Threshold III was applied LST value 12oC – 20oC and NDWI value -0.4 to 0.2.

The swamps area distribution resulted from reclassification process, image acquisition on 16 May 2006 had smaller area proportion than image acquisition on 15 April 2000. It’s caused by the acquisition date of the image influencing the temperature and water distribution in whole area of interest. It implies to the large area of swamp distribution. Swamps distribution obtained from image acquisition on 15 April 2000 with threshold II analyzing had the area around 6,389 km2 and swamps area distribution obtained from image acquisition on 16 May 2006 with threshold I had the area is around 5,493 km2.

Overall accuracy of determination swamp area based on the LST, NDWI and supervised classification is still in moderate accuracies. For image acquisition

on 15 April 2000, highest accuracy is 73,5% for the threshold II with LST value range of 23oC - 33oC and NDWI value range of -0.35 – 0.1. Meanwhile for Image acquisition on 16 May 2006, highest accuracy is 63.5% for the threshold I with LST value range of 23oC - 33oC and NDWI value range of -0.43 – 0.

The most problem occurred in the swamps determination is in the image classification processing. In this step, the initial cluster gathering was undertaken using unsupervised classification. The resulting clusters were later edited, assigned names and used in final supervised classification of the image as a base reference for each spectral. Training area was formed by using visual interpretation of the image referring to the each spectral from unsupervised classification. Some land cover types have the similar spectral, but based on field knowledge and visual interpretation is different land cover types. In image acquisition on 15 April 2000, one constraint in image classification is haze. It caused the spectral value of two different land cover type almost similar for example paddy field and shrub have the similar spectral value therefore in visual interpretation it makes confusion among others. Likewise in image acquisition on 16 May 2006, settlement and bare land have the similar spectral value. For the next processing, the classification results would be reclassified into 2 type of land cover that are the swamps and not swamps area. Misclassification in the image classification process can caused the error in determining the swamps area in whole area of interest.

The percentage of suitability classes for paddy field based on land system for whole area of interest were S1 (highly suitable) around 52.2%, S2 (moderately suitable) around 41.8% and S3 (marginally suitable) around 6.07%. This result was overlaid with the swamps distribution from previous method to get the distribution swamps area suitable for paddy field. Suitability classes for paddy field were S1 (highly suitable) class covered 54% area or 3,426 km2 , S2 (moderately suitable) class covered 40.2% area or 2,550 km2 and for S3 (marginally suitable) class around 5.8% area or 369 km2. There is no area that included in N (not suitable) class.

This method can be applied as an alternative method in determining alternative area for wetland to support the agricultural development. Referring to the limitation of the method for improving of the accuracy, it is suggested to use some other thresholds value range more detail in order to get the more variety of result possibility. The accuracy assessment is suggested to do in each step by using more detail of ground truth data.

Keywords : NDWI, LST, analisa kesesuaian, rawa.