Implementation and Classification Results

448 class unclassified . These unfiltered pixels mean the areas related to these pixels represent land cover other than forest, barren land, residential areas, or paddy fieldsplantations. We implemented ten simulations using different ROs for classification process and accuracy assessment and the result is shown in Fig. . Unclassified Residential Area Forest Barren Land Paddy fieldsPlantations Figure 6. Classification result of joint processing of Landsat 8 December 11, 2014 and ALOS ‐2PALSAR‐2 December 12, 2014 combined image first simulation Based on classification results, the area for each class is presented in more detail in Table . As shown in the table, in most of the area classified as forest, followed by residential area, paddy fieldsplantations, and barren land. Table . Classification results of combined Landsat 8 on Dec. , and ALOS‐ PALSAR‐ on Dec. , Simulation Area a Residential Area Forest Barren Land Paddy FieldsPlantations , . , . , .8 , . 8, . , . , . , . , 8. , . , . , . , . ,8 . , . 8, . , . , . , . , . , . , . , . 8, . , . , . , . , . 8 , . , . , 8 . 8 ,8 8. , . , 8. ,8 . , . , .8 , . 8 , 8. , . Average , . , . , . , . We implemented the classification method using the data in . The Landsat 8 data was dated on January , and ALOS‐ PALSAR‐ was dated on February , . The date of both data was different since the limited number of good quality of Landsat data. The first simulation using RO as training samples and ROs as testing data is shown in Fig. . 449 Unclassified Residential Area Forest Barren Land Paddy fieldsPlantations Figure 7. Classification result of joint processing of Landsat 8 January 15, 2016 and ALOS ‐2PALSAR‐2 February 25, 2016 first simulation The overall simulation results, represented as the area for each class is shown in more detail in Table . As shown in the table, the composition of each class in almost the same with the case. Most of the area classified as forest, followed by residential area, paddy fieldsplantations, and barren land. Table . Classified area from combined Landsat Jan. , and PALSAR‐ Feb. , Simulation Area a Residential Area Forest Barren Land Paddy FieldsPlantations , . , . , . ,8 . 8 , . , 88. , . , . ,8 . , . , .8 , . , . 8 , . , . , .8 , 8 . , 8 . ,8 . 8, . , . , .8 , . 8 , 8. , . , . ,8 . , 8. 8 , . , . 8, 8 . 8, . , . , 8. 8, . , . , . , .8 , . , . 8 Average , . , . 8 8, . 8, . 3.3. Analysis of Classification Result We assessed the accuracy of classification results using confusion matrix method. Confusion matrix is performed by comparing the classification result using ROs that are randomly selected to other ROs from the training sample for each class. The results of accuracy assessment of combined Landsat 8 and PALSAR‐ in and are shown in Table . 4 0 Table . Accuracy rate of classified image in and Trial . . . 8 . . . . . . . . . .8 . 8 . . . . . . Average . . The accuracy rate for classified Landsat 8 on December , and ALOS‐ PALSAR‐ on December , combined image is . and is higher than the accuracy of ALOS‐ PALSAR‐ image classification without Landsat 8 . . The classified image showed that forest represented by green pixels dominates the study area. Barren Land and residential areas are located in the center of Bandung regency. Barren land as well as paddy fieldsplantations dominate the residential areas. The accuracy rate for classified Landsat 8 on January , and ALOS‐ PALSAR‐ on February , combined image is . and is higher than the accuracy of ALOS‐ PALSAR‐ image classified without Landsat 8 . . The classified image shows that paddy fields and plantations represented by light blue pixels dominates the study area. As in , the residential area in is increasing as shown by red pixels as well as barren land and paddy fieldsplantations. The results are also correlated to the fact, that the number of population in Bandung is also increasing. Paddy fields and plantations areas are increasing because barren land that is actually paddy fields and plantation areas that were not in planting or harvesting period in December entered harvesting period in February , so paddy or other crops are dominant. As reported in Gala Daily December , , in the end of December , Bandung regency is just entering planting period. Another news agency also reported as Kompas Daily on December , and West Java Province Agriculture authority stated that harvesting period will be in that February Disdag, . The comparison of each class in and based on the classification results is shown in Table . Table . The comparison of area from each class based on classified results Class Area a Difference Residential area . , . , . , Forest . , . , 8 ‐ . , Barren land . , 8. , . , Paddy fieldsplantations . , 8. , 8. , As shown in the table, all classes are increased during ‐year period, except the forest, that are decreasing by ‐ , . a ‐ 8. . The increasing area of paddy fieldsplantations is also reported by West Java statistic data released by BPS Badan Pusat Statistik . t stated that the area of paddy fields in Bandung regency is increasing 4 1 form ,8 a in to , a in . Referred to this study, paddy fieldsplantations area is increased by 8, . a. The probability error that occurred during the classification process is caused by the pixels from a specific class is classified to different classes. n this case, PALSAR‐ as a radar sensor, recorded backscattered signals from the different classes into the same values. For example, paddy fieldsplantations and forest pixels are recorded as similar class because both objects consist of vegetation. Another reasons that may occurred in the process are: . ALOS‐ PALSAR‐ and Landsat 8 images has a time difference so that the types of land cover can change during these dates period. . The noise in Landsat 8 image caused by cloud or other atmospheric disruptions. . The selection of training samples using Google Earth is not quite precise to determine the classes in classification process.

4. Conclusions

The accuracy rate of land use and land cover classification using joint data from Landsat 8 and ALOS‐ PALSAR‐ is . in and . in . The classification accuracy rate of joint data is increased by . 8 and . , in and , respectively, compared to classification using ALOS‐ PALSAR‐ only. Residential areas are increasing by . from , . a in to , . a in . While forest areas are decreasing by . from , . a in to , . 8 a in . Barren land is increasing by . from , . a in to 8, . a in . Paddy fieldsplantations area is increasing by 88. 8 from , . a in to 8, . a in . References Bapedas . Badan Pengelolaan DAS Citarum Ciliwung. https:bpdasctw.wordpress.com. Last accessed October . BPS . Badan Pusat Statistik Provinsi Jawa Barat. http:jabar.bps.go.id. Last accessed October . CCRS . Canada Centre for Remote Sensing, Fundamentals of Remote Sensing. http:www.nrcan.gc.caearth‐ sciencesgeomaticssatellite‐imagery‐air‐photossatellite‐imagery‐productseducational‐resources . Last accessed June . Disdag . Dinas Pertanian Tanaman Pangan Provinsi Jawa Barat. http:dagang.disperindag .jabarprov.go.id. Last accessed December . Google . Google Earth. https:www.google.comearth. Last accessed October . ogg, R.V. and Craig A.T. . ntroduction to Mathematical Statistics, th Ed., , ongkong Macau: Pearson Education Asia Limited igher Education Press. https:www.scribd.comdocument 88 Teori‐Peluang‐ Bab‐ . Last accessed November . JAXA . Advanced Land Observing Satellite‐ DAC‐ ALOS‐ . http:alos‐ ‐ restec.jpenstaticpagesindex.phpabt‐ . Last Accessed October . JAXA . ALOS‐ Advanced Land Observing Satellite‐ ; SAR Mission Daichi‐ . http:directory.eoportal.orgwebeoportalsatellite‐missionsaalos‐ . Last accessed October . Lehmann, E., et al. , Joint processing of landsat and ALOS‐PALSAR data for forest mapping and monitoring. EEE Transactions on Geoscience and Remote Sensing : ‐ · January NASA . Landsat Science. http:landsat.gsfc.nasa.gov?page_id= . Last accessed October . Suzuki, Y. . New Era of Global Monitoring by ALOS‐ : Advanced Land Observing Satellite‐ DAC‐ , nt. Relations and Res. Dep., JAXA. The 7th Indonesia Japan Joint Scientific Symposium IJJSS 2016 Chiba, 20‐24 November 2016 4 2 Geological Mapping for the Land Deformation Using Small UAV, DinSAR Analysis and Field Observation at The Siak Bridge I and

II, Pekanbaru City, Indonesia

usnul Kausarian a,b, , Josaphat Tetuko Sri Sumantyo a , Detri Karya b , Dewandra Bagus Eka Putra b, Evizal Abdul Kadir b a Josaphat Microwave Remote Sensing Laboratory, Center for Environmental Remote Sensing, Graduate School of Advance Integration Science, Chiba University, 1 ‐33, Yayoi, Inage,Chiba, 263‐8522, Japan b Faculty of Engineering, Universitas Islam Riau, Jl. Kaharuddin Nasution No. 113 Pekanbaru Riau 28284, Indonesia Abstract Pekanbaru is the capital city of Riau province, located in the center of the Sumatra sland ndonesia is experiencing rapid economic progress. The city is split by Siak river. Big bridges that connect this city are Siak Bridge and Siak Bridge. The Quality of Siak bridges paid attention seriously in this previous time, especially for Siak bridge deflection, despite completely built from past time. Also for Siak Bridge it needs more attention incase of this bridge is the road for the heavy truck way. Geological mapping using a small UAV conducted to determine the deformation points on the site of Siak bridge and Siak Bridge areas, the study of Siak Bridge and Siak Bridge are also supported by the DnSAR analysis using ALOS PALSAR data of Pekanbaru city and Deflection observation that occurs in Siak bridge and Siak Bridge was measured during field observation. Results of small UAV D models analysis shows no negative land deformation on the Siak Bridge and Siak Bridge areas. DnSAR analysis shows the number of positive deformation of Siak Bridge is 8 cm and Siak Bridge is 8 cm. Deflection on Siak Bridge was detected around ‐ cm. Keywords Keywords: Pekanbaru city; Siak Bridges; Small UAV; DnSAR; Deflection

1. Introduction

Pekanbaru city Figure is the capital city of Riau province, ndonesia located ° ‐ ° E and ° ‐ ° N with sea level average from ‐ meters. The northern part of this city is a plain area with average height ‐ meter. The total area of Pekanbaru city is . km². Pekanbaru city is the one of the main city in ndonesia with the population of . million people based on Center for ndonesia Statistic Board, Pekanbaru branch released on . The economical in Pekanbaru growth above in a year, this number is above of ndonesia’s economy growth which has . Center for 4 3 ndonesia Statistic Board, . This rapid growth needs infrastructure facilities to support people activities, meanwhile, Pekanbaru city is the strategic city located at the center of Sumatera sland. With this position, Pekanbaru city is the connector of the north‐ south littoral at the Sumatera sland, also as the connector between east and west of this island. This importance of Pekanbaru city requires supporting infrastructures inevitably. One them is bridge facility. Pekanbaru city was split into two because it passed the great river named Siak River, the river is recognized as well as the deepest river in ndonesia. Siak River divides Pekanbaru city located at the Senapelan sub‐district, Rumbai sub‐ district, and Rumbai Pesisir sub‐district. The establishment of the bridge as the community liaison activities in Pekanbaru, provide enormous positive impact by making the groove Pekanbaru ground transportation becomes easier, but the construction of this bridge also pose a risk to their land‐use change and the condition of the bridge due to the pressure exerted by the bridge and traffic activities that are in it. This study aims to determine the effect given or experienced by land due to the construction of the bridge. The bridge is located on the Siak River area of Pekanbaru city. Siak bridge located at ° . 8 N and ° . E, was built in and available to use in with the length meter. Siak bridge located at ° . N and ° .8 E, was built in with length meters. Figure 1. Map of Pekanbaru city, Riau Province, Indonesia.

2. Data and Method

Three methodologies were used for this research. Field mapping on Siak , Siak bridges were used small UAV with specifications: built‐in camera with FOV Field Of View ° mm mm format equivalent f .8 lens, . ” sensor and effective pixels: M connected to the built‐in GPS. Pictures were taking among weeks for the whole bridges, starting from May until June . The pictures were rendered using some software to produce a D model of mapping area.