- BSc (Technical Geography, 1985); Drs (Remote Sensing, 1989); PGradDiploma (Rural&Land Ecology ITC, 1991); M.Sc (Rural... moreBSc (Technical Geography, 1985); Drs (Remote Sensing, 1989); PGradDiploma (Rural&Land Ecology ITC, 1991); M.Sc (Rural&Land Ecology ITC, 1993); Ph.D (Mapping Science/Remote Sensing, UQ, 2007)edit
- Prof. Stuart Phinn, Prof. Geoff McDonald, Dr. David Pullaredit
Research Interests:
Research Interests:
Research Interests:
... ШЦяЛа JAVA S £ A I Study area INDIAN Ttvt»txrs 114ГЭОТ»00*Е Figure 1. Landsat Thematic Mapper imagery recorded on 25 June ... Sw 18 Forest vegetation on shaded slopes Fsh 3 Typical indundated ricefield Inu 19 Forest vegetation on... more
... ШЦяЛа JAVA S £ A I Study area INDIAN Ttvt»txrs 114ГЭОТ»00*Е Figure 1. Landsat Thematic Mapper imagery recorded on 25 June ... Sw 18 Forest vegetation on shaded slopes Fsh 3 Typical indundated ricefield Inu 19 Forest vegetation on neutral slopes Fne 4 Bare ground/dry ...
ABSTRACT This research developed a versatile land-use information system (VLUIS) based on moderate- and high-spatial resolution imagery for supporting local planning in Indonesia. It was motivated by the fact that the existing land-use... more
ABSTRACT This research developed a versatile land-use information system (VLUIS) based on moderate- and high-spatial resolution imagery for supporting local planning in Indonesia. It was motivated by the fact that the existing land-use information contained by the Key Dataset for Local Development (KDLD) was not adequate to support environmental planning at local levels in Indonesia. This was due to its inconsistent mapping methods, contents/classification scheme, and inflexibility to be used as an input to local physical planning processes. Although the KDLD was developed by most local coordinating agencies for development planning (Bappedas), the land-use map was not used as a common reference by various local and provincial institutions in assessing the state of environment. Therefore, each institution had a tendency to develop its own land-cover/land-use information, resulting redundant works of land-cover/land-use mapping, which were incompatible to each others. With regard to that problem, the objectives of this study were: (a) to specify land-use related planning tasks at local level in Semarang-Salatiga area, Java, Indonesia; (b) to design a versatile landuse classification scheme for urban and rural environment at local level in Java in order to support various applications in the local planning context; and (c) to develop and verify the versatile land-use mapping methods based on moderate- and high-spatial satellite imagery. Semarang-Salatiga area was chosen due to its relatively complex land-use phenomena and data availability. In this study, two types of satellite image dataset were used, Landsat-7 ETM+ and Quickbird, representing moderate- and high-spatial resolution imagery respectively. To achieve the research objectives, a methodology comprising three stages of activity was developed. The first stage specified local physical planning tasks and their required land-cover/land-use information, based on literature study and interview with 36 stakeholders in the study area. In the second stage, versatile land-use information contents were specified in a classification scheme containing five land-use dimensions, i.e. spectral, spatial, temporal, ecological, and socio-economic. In the third stage, a set of image classification methods was developed for generating all land-use dimension maps with the specified classes. For each type of imagery, the study area was divided into northern and southern parts. The northern part represents more developed/urbanised area, while the southern part represents less developed or rural areas. Multi-spectral classification in terms of both standard and non-standard approaches were explored to derive the spectral-related land-cover classes, while visual interpretation and object-oriented image segmentation were compared to find most accurate method in generating the spatial dimension classes. The standard multi-spectral classification approach made use of original bands as input to the classification process, while the non-standard approach involved texturally filtered and texturally aggregated bands in addition to the original ones. The spectral-related land-cover and spatial dimension maps, supported by a terrain unit map, were integrated in a raster GIS environment to derive the temporal, ecological, and socio-economic maps in separate processing methods. After that, all derived maps were integrated into a single dataset of VLUIS, ready for query-based activation at will and translation to other classification systems. Based on the interview with the respondents, a list of variables related to land-cover/land-use information required by various local planning tasks was regrouped with respect to the developed five land-use dimensions. After that, a classification scheme containing five columns representing spectral-related land-cover, spatial, temporal, ecological, and socioeconomic dimensions were created. The specified classes under each dimension referred to the variables used in various local planning and to the existing, widely used, classification systems. The spectral-related land-cover mapping results showed that standard multi-spectral classification methods using the original spectral bands gave higher accuracy results (84.63% or Kappa=0.8276 for Landsat-7 ETM+ and 68.75% or Kappa=0.6813 for Quickbird) than non-standard classification methods involving textural filtering (80.55% or Kappa=0.7988 for Landsat-7 ETM+ and 66.45 or Kappa=0.6503 for Quickbird) and textural aggregation (66.68% or Kappa=0.6512 for Landsat-7 ETM+ and 63.91% or Kappa=0.6222 for Quickbird) approaches. This was due to the fact that the texture is closer to spatial rather than spectral concept, while the specified categories in the spectral-related land-cover dimension is purposively developed for spectral classification. For the same image coverage and number of classes, Landsat-7 ETM+ gave higher accuracies (84.63% or Kappa=0.8276 for 40 classes, and 87.05% or Kappa=0.8535 for 25…
Research Interests:
Kelud eruption in February 13rd 2014 has a huge impact for the environment. The ashfall spread out around Java island and make several airports need to close their flight schedule for a week. It's become obstruction for the movement... more
Kelud eruption in February 13rd 2014 has a huge impact for the environment. The ashfall spread out around Java island and make several airports need to close their flight schedule for a week. It's become obstruction for the movement of the economics, people and also goods. In the other side, about 4 days after the eruption event, Lahar coming down in the northern flank of volcano and then destruct a paddy field, settlement and killing a victims. Until now, the debris of pyroclastic materials still covered the area surrounding Kelud and it needs to be mapping out to help the inventory of damage and losses assessment done well. We will test both of high spatial resolution data to do this work. The imageries are GeoEye-1 and FORMOSAT-2. GeoEye-1 provide the condition before the eruptions happen while FORMOSAT-2 has an after event imagery, so that by these data we will evaluate the impact of eruption to the environment of Kelud Volcano especially for land-use. The OBIA techniques wa...
Research Interests:
Research Interests: Image Processing, Remote Sensing, Spatial Analysis, Local Development, Land Cover, and 12 moreLocal governance, Land Use, Decision Support, Environmental Assessment, Landsat TM, High Resolution, Information Content, Spatial resolution, Extraction Method, High Spatial Resolution, Area of Interest, and Information System
ABSTRAK Penelitian ini bertujuan untuk mengkaji pengaruh perubahan penutup lahan terhadap debit puncak di DAS Gesing, Purworejo, dan dampaknya terhadap bencana banjir di daerah hilir dari DAS tersebut. Dalam penelitian ini, citra landsat... more
ABSTRAK Penelitian ini bertujuan untuk mengkaji pengaruh perubahan penutup lahan terhadap debit puncak di DAS Gesing, Purworejo, dan dampaknya terhadap bencana banjir di daerah hilir dari DAS tersebut. Dalam penelitian ini, citra landsat TM tahun perekaman 1992 dan citra Aster VNIR tahun perekaman 2003 digunakan sebagai dasar ekstraksi informasi penutup lahan. Sementara itu, data penutup lahan yang diperoleh dari citra juga digunakan sebagai salah satu masukan pemodelan koefisien aliran permukaan (C) untuk masing-masing tahun, di samping data masukan lain berupa kemiringan lereng, infiltrasi tanah dan simpanan permukaan. Hasil pemodelan koefisien aliran permukaan pada masing-masing tahun kemudian dijadikan masukan untuk pemodelan debit (Q) dengan menggunakan metode rasional, di samping data luas area (A) dan intensitas hujan (I) yang juga disajikan secara spasial. Metode rasional sederhana ini diimplementasikan dengan pendekatan sel (piksel) menggunakan perangkat lunak PCRaster, di mana komputasi terebut menggunakan ukuran piksel sebagai A parsial sehingga dihasilkan peta distribusi spasial debit pada setiap piksel, dan menjaid debit puncak di bagian outlet DAS. Hasil pemodelan debit pada dua tahun yang berbeda ini kemudian dibandingkan dengan perubahan penutup lahan, baik dalam perubahan jenis (hasil klasifikasi multispektral) maupun kualitas (perubahan indeks vegetasi), serta dibandingkan pula dengan tinggi banjir di lapangan yang merupakan model interpolasi spasial jejak banjir dan wawancara dengan masyarakat lokal. Hasil penelitian ini menunjukkan bahwa perubahan penutup lahan di DAS Gesing bagian hulu telah terjadi secara signifikan, dan hal ini berpengaruh besar terhadap peningkatan koefisien aliran permukaan serta debit puncak. Banjir tahun 1992 dan 2003 merupakan dampak langsung dari perubahan tersebut. Penelitian ini juga mencoba membahas beberapa kelemahan dari model yang digunakan. Kata kunci: data penginderaan jauh, penutup lahan, DAS, banjir, debit puncak 1. PENGANTAR Banyak wilayah di Jawa saat ini telah mengalami peningkatan jumlah dan kepadatan penduduk. Di wilayah perdesaan, jumlah penduduk yang meningkat pesat juga kebanyakan menggantungkan hidupnya pada aktivitas pertanian. Peningkatan ini secara langsung menyebabkan beberapa masalah lingkungan, di mana kebutuhan akan lahan pertanian dan juga permukiman pada umumnya dikompensasi dengan penyusutan lahan hutan dan vegetasi alami/semi-alami lainnya. Dengan demikian, terjadilah proses perubahan penutup dan penggunaan lahan sekaligus.
Research Interests:
The study utilizes Landsat-7 ETM+ 2001and Landsat TM5 2009 based on Normalized Differences Vegetation Index (NDVI) and 457 colour composite at the study area located in Leihitu Peninsula, Ambon City District, Ambon Island, Moluccas... more
The study utilizes Landsat-7 ETM+ 2001and Landsat TM5 2009 based on Normalized Differences Vegetation Index (NDVI) and 457 colour composite at the study area located in Leihitu Peninsula, Ambon City District, Ambon Island, Moluccas Province. The classified satellite data under NDVI and 457 colour composite of 2001 and 2009 of 2001 and 2009 were used to determine land cover change that have occurred in the study areas. This study attempts to use a comparative change detection analysis in land cover that has occurred in the study area with NDVI and 457 colour composite over 9 year period (2001 to 2009). The results of the present study disclose that total area increased their land cover were bare land and impermeable surface, herbaceous and shrubs, low density vegetation, and medium density vegetation, while high density vegetation is decreasing in both NDVI and 457 colour composite analysis. Overall accuracy was estimated to be around 94.3 % for NDVI and for 457 Colour composites was 84.7%. The study area has experienced a change in its land cover between 2001 and 2009 in both NDVI and 457 false colour composite analyses. The whole land cover types have experienced increased in both methods, except high density vegetation. The transformations of spectral vegetation (NDVI) product more closely with actual land cover compared with 457 colour composite product.
Research Interests:
Object based image analysis (OBIA) approaches on landform study are considered relatively fewer as compared to landuse/landcover studes. By using OBIA approach, subjectivity on conventional manual interpretation can be further... more
Object based image analysis (OBIA) approaches on landform study are considered relatively fewer as compared to landuse/landcover studes. By using OBIA approach, subjectivity on conventional manual interpretation can be further minimalized. The OBIA, until recently, have not been carried out to extract karst morphology. This study aimed to build a new prosedure to extract mayor karst morphology using OBIA on DEM data on land facet level, and evaluate the results. This study used digital topographic map at 1:25.000 scale to generate digital elevation model (DEM). Slope, shaded relief, topographic position index, and elevation percentile were then derived from the DEM to be used as input to segmentation and classification processes. Parameterization on segmentation and classification was applied by trial and error method. The resulted rule sets for each data were then applied to two pairs of 3 x 3 km wide area of interest. Each of these two areas has different karst type. All the results were then evaluated based on efficiency and consistency of segmentation and classification. Karst features obtained by the proposed method were also visually compared to those which were identified using manual digitization on 1:30.000 panchromatic aerial photographs. It was found that the overall segmentation and classification results on proposed method worked better on karst with conical hill type, while for other types, the rule sets need to be further adjusted. On land facet level, conical hills boundary found to be more comparable and easier to evaluate than any other. However, the results also proved that OBIA provides a better, quicker, more objective, and repeatable alternative to fieldwork and manual digitization.
Research Interests:
Research Interests:
GISdevelopment.net-> Application-> Land Information System ABSTRACT Recently, high-spatial resolution imagery has widely been used for environmental assessment and mapping in Indonesia. However, most of these studies made use of visual... more
GISdevelopment.net-> Application-> Land Information System ABSTRACT Recently, high-spatial resolution imagery has widely been used for environmental assessment and mapping in Indonesia. However, most of these studies made use of visual interpretation instead of digital classification for generating land-cover/land-use information from such imagery. On the other hand, digital classification was usually applied to deliver land-cover information or mixed information between land-cover and land-use with relatively general categorisation, so that the results were not adequate to support planning. This study tried to extract land-use information related to socioeconomic function by combining spectral classification, image segmentation and visual interpretation of Quickbird imagery covering Semarang area, Indonesia. To do so, a multi-spectral classification was run to derive detailed spectral-related land-cover classes. Image segmentation and visual interpretation were also carried out to generate spatial pattern of the land-cover features. A classification scheme under the versatile land-use classification system (VLUIS) was used as a reference. Integration of the spectral-related land-cover and spatial pattern maps was controlled using a knowledge-based approach, by formalising knowledge about spatial relationship between land-cover, socioeconomic function, spatial pattern and their ecological context into a set of GIS rules. The result showed that Quickbird imagery could be used for generating socioeconomic function of land-use at 83.63% accuracy (Kappa=0.821). In addition, several limitations related to the methods used and inaccurately mapped categories were identified.
Research Interests:
Lossy image compression techniques can affect information content of remotely sensed images. This study tried to compare classification results' accuracy of compressed ALOS-AVNIR2 multispectral images covering Salatiga-Ambarawa area... more
Lossy image compression techniques can affect information content of remotely sensed images. This study tried to compare classification results' accuracy of compressed ALOS-AVNIR2 multispectral images covering Salatiga-Ambarawa area (Central Java) using per-pixel and object-based image analysis (OBIA) approaches. The image was compressed using JPEG format at various compression levels from 10 to 90% with 10% increment, prior to the classification process. Meanwhile, the original uncompressed data was kept as a control. This study showed that the original uncompressed data gave the highest accuracies using both per-pixel classification approach. However, the per-pixel classification was also found more sensitive to the decrease of data quality caused by the compression levels, while the OBIA classification performed more consistently at lower accuracies for both original and compressed datasets.
Research Interests:
Spectral approach in satellite image analyses such as vegetation indices have been widely used in many applications, and it frequently gave good results. However, problems came up when spectral indices based on multispectral data were... more
Spectral approach in satellite image analyses such as vegetation indices have been widely used in many applications, and it frequently gave good results. However, problems came up when spectral indices based on multispectral data were used for analysing rocks or soils in wet tropical regions due to the presence of vegetation cover. Therefore, spectral index-based analyses for rocks and soils were only effective in large, non-vegetated bare lands. In order to overcome such problem, study tried to integrate spectral and geostatistical approaches for mapping soil surface clay content in Gunung Kidul area, Yogyakarta, Indonesia. As an initial stage, a Landsat-8 image was corrected radiometrically and geometrically, so that its pixel values were transformed to at-surface reflectance, while the geometric position of each pixel refers to Indonesian topographic (RBI) map. After that, two spectral indices for accentuating surface clay content, i.e. SRCI=Band7/Band6 and NDCI=(Band7 – Band6)/(Band7 + Band6) were applied for generating two tentative clay content-related images. Since those images also represent areas other than open soils, two other processes were undertaken in order to isolate the open soils. The first process was applying normalised difference vegetation index (NDVI), by which the areas representing vegetation cover and water were masked out. The second one was multispectral classification for generating built up features such as asphalt, concrete and housing rooftops, since those objects could not be separated from open soils using previously mentioned indices. The clay index images were then correlated with field samples containing information on laboratory-analysed clay contents, resulting nearly the same correlation coefficients for Band7/Band6 image (r=0.65) and (Band7– Band6)/(Band7 + Band6) (r=0.63). Regression equations were used to transform the images to surface clay content maps of the barren land. Since these maps only represent clay content in the open soil area, a geostatistical approach in terms of kriging method was utilised to interpolate the pixel values. By this method, every pixel value of surface clay content is considered as point, and thus the semivariogram analysis could be run. As a result, this method could predict surface clay content in the vegetation-covered areas. Accuracy assessment using independent clay content samples showed that SRCI gave a slightly better results than NDCI. Moreover, when the accuracy tester samples were also taken from vegetated areas, the SRCI-and NDCI-based spatial interpolation results gave lower accuracy than those of spectral index-based model. The resultant models were also impeded by the presence of settlements with clay rooftops, which were hard to be automatically separated from the scene. This study also showed that the integration of spectral and geostatistical approach could improve soil characteristics mapping accuracy in the area with scattered, small open soils.
Research Interests:
Water is an indispensable requirement for every creature on the earth's surface. Water is closely related to the hydrological cycle, one of the processes that occurs during that cycle is evapotranspiration both actually and potentially.... more
Water is an indispensable requirement for every creature on the earth's surface. Water is closely related to the hydrological cycle, one of the processes that occurs during that cycle is evapotranspiration both actually and potentially. However, information in terms of spatial distribution of actual evapotranspiration is very rare, particularly in Indonesia. Landsat-8 satellite data offers opportunity in coping with this need, since the image dataset is available in both reflective and thermal spectral bands. Those bands can be generated to land-cover and temperature information, which are important to surface energy balance computation including evapotranspiration. The purposes of this study were (1) to understand the capability of Landsat 8 for deriving actual evapotranspiration (ETa) parameters estimation, (2) to know its accuracy according to data obtained from meteorological and climatology stations, and (3) to determine the spatial distribution of ETa based on land cover information. The ETa can be extracted from remotely sensed images using Surface Energy Balance System (SEBS) algorithms. ETa estimation made use of SEBS algorithm comprising several parameters, i.e. net radiation (Rn), which is proportional to the amount of soil surface heat flux (G0), sensible heat flux (H) and latent heat flux (λE). ETa is part of the λE which is calculated based on SEBS algorithms. The parameters required for SEBS include albedo, emissivity, land surface temperature, NDVI, vegetation fraction, LAI, surface roughness momentum transfer (Z0m), canopy height, and elevation represented by digital elevation model (DEM). Each of these parameters serves to establish the elements of energy balance of Rn, G0, H, or λE. The land surface temperature was computed on a pixel basis using split-window algorithm. The results showed that all parameters have good accuracies in comparison with the reference data to built SEBS algorithm. ETa accuracy results referring to the data from meteorological dan climatology stations showed standard error of estimates of 0.99 mm/day, 2.18 mm/day, and 2.66 mm /day at 3 different station locations. The highest ETa value was located in the objects of body of water, i.e. at 9.6 mm/day; while the lowest one was located in the objects of zinc roof, i.e. at 5.6 mm/day. This study demonstrated the advantages of spatial data like Landsat-8 satellite images and DEM over meteorological station data, particularly in modelling the spatial distribution of ETa in a relatively small area, which could not be done using data obtained from meteorological stations.
Research Interests:
Land-use studies based on satellite imagery usually focused on information extraction methods, change monitoring, agricultural yield estimates, and the influences of particular land-use types on the environmental processes such as erosion... more
Land-use studies based on satellite imagery usually focused on information extraction methods, change monitoring, agricultural yield estimates, and the influences of particular land-use types on the environmental processes such as erosion and flood. Studies emphasizing on ecological aspects of land fragmentation were fewer as compared to those areas of interest., although the information on such fragmentation can provide indicators of environmental fragility. This study tried to analyse the agricultural land-use fragility in Semarang area, Central Java, Indonesia based on the fragmentation levels in three observation scales. Landsat TM and ETM+ images recorded in 2002 was used as a basis for information extraction. In this study, the image was processed using multispectral classification in order to generate land cover classes. A knowledge-based technique was used to derive land-use map by integrating the land-cover classes and the terrain characteristics information in a raster-based GIS environment. The land-use classes were then analysed with respect to their fragmentation levels in 3x3, 5x5 and 7x7 moving window sizes respectively; by which the window sizes represent different scales of observation. Based on the generated fragmentation map, it was found that the agricultural land in the study area has been severely fragmented in 2002. Another important result of this study is that the use of different window sizes have generated different results of agricultural land-use fragmentation index. This result comply with the theory of modifiable areal unit problem (MAUP), particularly the effect of scales on the drawn conclusion of spatio-ecological phenomena.
Research Interests:
This research developed methods for generating land-use information which is relevant to a broader study, i.e. land capability assessment and classification for sustainable development of brackish water aquaculture systems in Indonesia.... more
This research developed methods for generating land-use information which is relevant to a broader study, i.e. land capability assessment and classification for sustainable development of brackish water aquaculture systems in Indonesia. The broader study requires land-use information covering coastal areas with various utilisations, e.g. coastal fishponds, rice fields, rural settlement, mangrove-based conservation, and urban uses. In order to meet that requirement, the land-use information needs to be delivered in terms of spatial, ecological, and socioeconomic dimensions. To do so, a versatile land-use information system (VLUIS) which has been developed for local planning in Indonesia was used as a reference. In the VLUIS, the land-use information is broken down into five layers representing spectral, spatial, temporal, ecological, and socioeconomic dimensions. As the study area, a small portions of Landsat ETM+ image covering Maros, South Sulawesi, Indonesia was chosen. In this study, a combination of multispectral classification and object-based image segmentation was applied. The multispectral classification was carried out to generate spectral-related land-cover types, while the object-based image segmentation was run to derive spatial dimension of land-use. A terrain unit map obtained from visual interpretation was used to support the integration of the spectral-related land-cover and the spatial dimension maps. A knowledge-based classification incorporating spectral, spatial and terrain characteristics of the study area was carried out. By this method, new spatial information in terms of maps representing ecological and socioeconomic dimension of land-use were generated using different rules. This study showed that a single source of imagery could be processed in various ways to derive different types of spatial information, and all information could then be integrated to generate versatile land-use information relevant to particular planning tasks. The multispectral classification was found to be accurate enough to provide spectral-related land-cover types. It was also found, however, that the object-based image segmentation was still less accurate to classify objects with respect to their shape, size, and pattern simultaneously, particularly in comparison with the visual interpretation. Nevertheless, in the near future, improved methods of this approach may be expected to provide more useful and accurate information.