“3D urban geological information development for resilient society”
Susumu NONOGAKI;
Keynote Talk
To achieve safe, resilient and sustainable society, it is necessary to develop reliable intellectual infrastructures, such as basic research, ideas, and general-purpose technologies, and share it in an easy-to-use form. Geological information is one of the key intellectual infrastructures in the field of disaster management and plays an important role in evaluating earthquake-induced geohazard risk, such as ground motion amplification and liquefaction/subsidence. Recently, as part of the intellectual infrastructure development by national government, cooperating with local governments, Geological Survey of Japan (GSJ) has been working on creating three-dimensional (3D) geological maps of urban areas that represent the shallow subsurface geological conditions using drilling surveys data, and on developing a web-based system to make the developed information to the public. To date, the 3D geological maps of northern area of Chiba Prefecture and central Tokyo, Japan have been completed and are available for free on the GSJ website. These 3D geological information will help to enhance the accuracy of geo-risk assessment and subsurface environmental assessment and improve the efficiency of urban planning, thereby contribute to the achievement of resilient cities and communities and other goals set forth in the SDGs. This presentation will provide an overview of the 3D urban geological information development by GSJ and the upcoming challenges.
“A Review of Thai Geographical Research (2000-2023): Trends, Themes, and Future Directions”
Assoc.Prof.Pathana Rachavong;
Keynote Talk
This presentation provides an in-depth review and categorization of geographical research conducted by Thai scholars between 2000 and 2023. By analyzing key themes, methodologies, and regional focus areas, this study offers insights into the evolving landscape of geography in Thailand. The research is grouped into several categories, highlighting shifts in academic focus, the integration of technology in spatial analysis, and emerging topics such as climate change and urbanization. Additionally, the presentation reflects on how Thai geography has responded to global challenges and suggests future research directions that align with national and international priorities.
“A Study on the Distribution of Village Names Derived from Vegetation in Northeastern Thailand, Cambodia, and Laos”
Yoshikatsu Nagata;
GIS-IDEAS Oral Presentation
The place names of villages often use words that are appropriate to describe the land in a straightforward manner. Words such as “river” and “pond” evoke a place with easy access to water resources that are essential for life. Words such as “valley” and “hill” also evoke landforms that are landmarks. Words derived from plants useful for life are also often used, either alone or in combination with words describing the topographical environment.
This report examines the frequency and distribution of words derived from vegetation used in village-level place names in northeastern Thailand, Cambodia, and Laos, a region and countries located in continental Southeast Asia.
The number of village-level place names analyzed is approximately 29,000 for Northeastern Thailand, 13,100 for Cambodia, and 11,600 for Laos. The place name data for Northeastern Thailand is based on the KCC2K surveys conducted by the Ministry of the Interior of Thailand around 1990 and does not include urban centers. For Cambodia, the place name data is based on topographic maps at 1:50,000 published in the 1960s, and therefore only representative place names are included for urban areas. In Laos, the names of villages were extracted from the 1995 census and include urban areas. Thus, although the age and the range of the collected data are not homogeneous from region to region, it is sufficient material to study the occurrence of words of vegetation origin.
The most common word derived from vegetation is "mango," which is an edible fruit tree. In Cambodia, it occurs in about 3% of village names. It appears in about 1.5% and 0.6% of village names in Northeastern Thailand and Laos, respectively. Although Laos and northeastern Thailand are closely related as the basin of the Mekong River, there is a large difference in the frequency of occurrence. One of the reasons for this is that the Lao language is the mother tongue of about half of the total Lao population. It is possible that there are many place names with words derived from languages other than Lao. In northeastern Thailand, "lotus" is the most common word, used in about 1.8% of village names. It appears in about 0.6% of village names in both Cambodia and Laos. Lotus is a word that reminds people of beautiful bodies of water, not only for food but also for scenery.
Useful trees such as Ceylon oak, “samrong” in Thai, and “pradu” in Thai, also appear as words often used in village names. Details will be given at the presentation.
The words used in place names depend on the language used in the village. Therefore, the local vernacular name of a tree is not always the same as its name in the national standard language. In this report, the author mainly used language dictionaries and botanical dictionaries that can be searched in each language to confirm the meaning of the words used in place names. Since many research results on the utilization of useful plants include not only scientific names but also local vernacular names in each region, these research results will be included in due course.
“Advancing Geoscience and Geospatial Awareness Through Educational Outreach: The CERG Story”
Natraj Vaddadi;
Keynote Talk
Geoscience and geospatial knowledge are foundational to understanding the Earth’s processes and addressing many environmental challenges humanity faces today. In an effort to promote geoscience literacy, a variety of outreach activities have been developed by The Centre for Education and Research in Geosciences (CERG), India. These programs, aimed at both laypeople and students, emphasize the importance of Earth Science in daily life and foster an appreciation of the interdisciplinary nature of geoscience. This abstract highlights several key outreach initiatives designed to inspire the next generation of geoscientists and raise awareness about the importance of geospatial techniques.
A major annual event organized by CERG is the Geoweek celebration, a week-long program filled with diverse activities such as mineral, fossil, and poster exhibits, essay, art, and photography contests, and a series of talks, workshops, and film screenings. Geoweek’s central theme is exploring how Earth Science influences the everyday lives of people. Through these activities, the program provides a comprehensive platform for participants to engage with the material in a hands-on, interactive manner. This event attracts students and engages the general public, helping bridge the gap between geoscience experts and everyday citizens.
Another integral part of CERG's outreach efforts is Geoquest, an interschool Earth Science quiz that allows school children to demonstrate their knowledge of geoscience. Conducted live, this event encourages friendly competition while fostering curiosity about Earth processes, natural resources, and environmental challenges. Similarly, the Geolab project contest serves as a dynamic platform for young minds to present creative and innovative solutions to contemporary geoscience-related issues. Although Geolab focuses on Earth Sciences, the interdisciplinary nature of many projects has been noteworthy, with entries incorporating fields like information technology, data science, mathematics, physics, and biology into geological studies. This blend of disciplines emphasizes the increasingly multidisciplinary aspect of geosciences and demonstrates how different scientific fields converge in addressing global environmental issues. Additionally, students participating in Geolab have the opportunity to engage with distinguished professionals from various geoscience-related fields, providing them with mentorship and inspiration for future academic and career pursuits.
In addition to these competitions, Geotrails are organized to expose school children to the practical aspects of geology. These geotrails consist of guided field trips to easily accessible rock outcrops in and around Pune district, Western Maharashtra. The region, predominantly covered by Deccan Volcanics, offers an ideal setting to observe volcanic and geomorphic features firsthand, such as lava tubes, caves, channels, and natural formations like arches and potholes. By providing participants with an immersive field experience, these trips offer a tangible connection between classroom learning and the natural world, making the Earth’s geological history come alive for young students.
A unique initiative, Jal Katha, an annual international short film festival, focuses on water conservation and the responsible use of water resources. By highlighting the role of the common man in protecting water, Jal Katha serves as a platform for raising awareness about one of the most critical environmental issues of our time. This event uses the power of visual storytelling to convey the importance of water conservation, particularly in the context of everyday behaviors and choices.
Another significant outreach effort is the Maps & Me workshop, which aims to introduce school and college students to the fundamentals of geospatial technologies and open-source mapping tools. In an era where digital maps and satellite imagery are becoming ubiquitous, the workshop provides students with basic skills in navigating and interpreting maps. By understanding the basics of map reading, satellite images, and digital mapping, participants gain insights into how these tools are used to solve real-world problems in fields such as urban planning, disaster management, and environmental conservation. These programs are conducted in collaboration with ‘Geo-info Lab, (now AGILE)’ and the ‘International Institute of Information Technology (I2IT)’, a prominent educational institute in Pune, India.
Overall, CERG’s outreach programs serve as powerful tools for building geoscience and geospatial awareness among the public, particularly schoolchildren. By offering a range of interactive, interdisciplinary, and engaging activities, these programs help cultivate a new generation of environmentally conscious individuals who understand the importance of geosciences in addressing the pressing challenges of today’s world. Through hands-on learning and real-world applications, these initiatives are fostering a deeper appreciation of Earth Science and inspiring young students to consider careers in this vital field. As the need for geospatial awareness and geoscience literacy continues to grow, these outreach activities provide a blueprint for how education and public engagement can be successfully intertwined to build a more informed and responsible society.
“Application of Data Mining Techniques and GIS in Land Suitability Assessment for Pomelo tree Development Planning in Vinh Cuu District, Dong Nai Province, Vietnam”
Huy Anh Nguyen;
GIS-IDEAS Poster Presentation
The combination of data mining methods (using decision tree models) and GIS is widely applied in land suitability assessment for the development planning of various crops. This approach aims to overcome the subjective limitations of traditional land suitability assessment methods, such as the limiting factor method and the Analytic Hierarchy Process (AHP). In this study, data mining involves applying analytical and machine learning techniques to extract information from data on soil characteristics and crop yields to assess land suitability for Tan Trieu pomelo in Vinh Cuu District, Dong Nai Province. Decision trees in data mining are considered an effective method for processing training data sequences, specifically by efficiently handling discrete variables and categorical variables of soil characteristics. This is a new method for quantifying the relationships between soil properties and crop yields. Geographic Information Systems (GIS) are powerful tools for land evaluation. GIS enables the collection, analysis, and visualization of spatial data on individual factors such as soil type, slope, mechanical composition, and soil layer thickness, thereby supporting decision-making related to the rational planning and proposal of crops, as well as the sustainable management and use of land. Integrating GIS with decision tree models will transform land suitability assessment results into quantitative data, which can be represented through suitability maps. These maps guide the planning and development of pomelo cultivation in Vinh Cuu District, Dong Nai Province.
“APPLICATION OF GIS TO BUILD A LAND DATABASE IN A MOUNTAINOUS DISTRICT OF SON LA PROVINCE, VIETNAM”
Hanh Tran;
GIS-IDEAS Poster Presentation
As a unique production resource, land is an invaluable national asset. The establishment of a database stands at the core of digital transformation. A land database consists of four key components: a cadastral database, a land statistical and inventory database, a land use planning database, and a land price database. This paper aims to create a land database to enhance the efficiency of land exploitation and management within the context of the Moc Chau mountainous district in Son La province. The study was carried out using various methodologies, including synthesis and analysis, field investigation and surveying, and notably, the integration of multiple software systems to develop the land database. Consequently, the necessary component databases have been successfully established, ensuring the requisite accuracy. These databases are compiled using the VBDLIS online software developed by Vietnam. This research holds practical significance, offering valuable insights for policymakers and aiding in effective land use planning. It establishes a robust data foundation, a crucial step towards operating within a digital government, a digital economy, and a digital society.
“APPLICATION OF GROUND PENETRATING RADAR AND MACHINE LEARNING ALGORITHMS FOR THE AUTOMATIC RECOGNITION OF UNDERGROUND OBJECTS”
Pham Minh Hai;
GIS-IDEAS Oral Presentation
In Vietnam, surveying and mapping underground utility networks, including water and sewer pipelines, electric cables, and telecommunication cables, are the contents of Law on Surveying and Mapping announced in 2019. Ground Penetrating Radar (GPR) is defined as a geophysical technique that transmits electromagnetic waves into subsurface and records echoes reflected from interfaces between media with contrasting dielectric properties. Data is a digital image with pixels having values of gray levels from 0-255, and underground objects show in the GPR as parabolic curves. Reading these data often based on the experience of interpreters, so accuracy is still subjective. Therefore, the objective of this study is to apply GPR and machine learning algorithms for the automatic recognition of underground utility networks, with the case study in Hanoi, Vietnam. The specific goals are to develop an approach for automatically bounding and labeling underground objects in GPR images. The study used RIS Hi-Mod #4 GPR system from IDS Company (Italy) to capture underground data. The research findings will provide a comprehensive method for mapping underground utilities, benefiting the local government’s urban management efforts
“Applying 3D gis technology to simulate high-rise buildings, serving urban land management - research at Vinhomes Central Park residential area in Ward 22, BT District, Ho Chi Minh City”
Huy Anh Nguyen;
GIS-IDEAS Poster Presentation
3D GIS technology creates vivid, intuitive digital products, accurately simulates objects, and shares information easily and quickly. Because of the above advantages, 3D GIS is widely used in all fields around the world, especially in urban management. 3D GIS technology is very rich and highly effective. This technology also opens up the ability to build 3D urban models in a modern, fast, vivid and accurate way. Vinhomes Central Park Ward 22, Binh Thanh District, Ho Chi Minh City has an area of 43.91 hectares. With large scale and skyscrapers (over 40 floors, 81 floors). Using traditional 2D maps is not as effective as 3D maps in showing perspective, specificity, and spatial visualization. Currently, there are many professional 3D construction software capable of expressing detailed model ideas with in-depth effects on model size, materials and 3D space lighting. The study successfully built a 3D model of the research area using CityEngine software with many advantages such as the ability to integrate 3D models simply and conveniently; 3D models have a smaller capacity but still ensure quality and level of detail compared to reality; The construction process is quick and simultaneous, and the block programming is similar to reality, saving time.
“ASSESSING LAND USE FUNCTION VARIATION IN URBAN TRANSITION: A CASE STUDY OF VUNG TAU CITY, VIETNAM”
Ho Dinh Duan;
GIS-IDEAS Oral Presentation
Vung Tau is a coastal class-1 city of Vietnam. The city is undergoing rapid urbanization with the expansion of the urban land, concurrent with the process of land function change. This study aimed to assess land function change in the city over the 2010-2015 and 2015-2020 periods. GIS and Markov chains techniques were applied to assess land function change of Vung Tau city. Through converting the land use code from purpose to function, the study has synthesized and identified 3 groups of land functions including civil land, non-civil land and other land. The results of assessing land function changes showed that civil and non-civil land continuously expands with an average expansion rate of +11.7% and +14.87% for the two periods. Meanwhile, other land continuously shrinks with an average rate of -5.39% for the two periods. The results of assessing land functions change in Vung Tau city can provide a basis for evaluating the local urban space development process.
Keywords. Landuse, Land function, Urban transition, Markov chain, Transition matrix, Vung Tau city.
“Climate change impacts agricultural production in the coastal area of Thi Nai lagoon, Binh Dinh province: An analysis of the livelihood vulnerability index”
Vu Khac Hung;
GIS-IDEAS Poster Presentation
This study identifies the vulnerability of agricultural household livelihoods to climate change in Thi Nai lagoon area, mainly concentrated in Quy Nhon City and Tuy Phuoc district. This area is assessed as vulnerable to severe climate change in Binh Dinh province. The study uses the Livelihood Vulnerability Index (LVI) according to the guidelines of the Intergovernmental Panel on Climate Change (IPCC), combined with the use of a dataset from the 2024 household survey. The results show that Tuy Phuoc has a higher level of vulnerability in all factors compared to Quy Nhon. This suggests the need for special support measures to improve adaptive capacity and reduce risks for people.
“Enhanced Site Location Selection: A Hybrid Framework Integrating GIS, SWOT, MCDM, and Game Theory”
Rahul, Bharath;
GIS-IDEAS Oral Presentation
The choice of a suitable site location for business establishments is a critical decision that
significantly impacts their success. This research proposes a novel, hybrid approach that
integrates Geographic Information Systems (GIS), SWOT Analysis, Multi-Criteria Decision
Modeling (MCDM), and Game Theory to provide an advanced, data-driven methodology for
site selection. This study commences by leveraging GIS to create geospatial models that
incorporate essential factors, such as accessibility, infrastructure, and demographic factors.
This approach facilitates the identification of potential business sites with optimal
characteristics. Next, the SWOT Analysis is introduced to evaluate the strengths, weaknesses,
opportunities, and threats in the proposed sites. This step provides crucial insights into the
competitive landscape, ensuring that the selection process accounts for the business's unique
requirements and external factors.
To further refine the site selection process, the study employs MCDM techniques, such as AHP
and TOPSIS. These methods systematically rate the potential locations based on multiple
criteria, allowing businesses to holistically assess alternatives and weigh the importance of
each criterion. Finally, Game Theory is applied to model the complex interdependencies among
key stakeholders, such as competitors, suppliers, and government entities. By anticipating
market shifts and understanding strategic interactions, businesses can make more informed
decisions about their site location.
The proposed hybrid approach provides an industry-standard, scientific framework for
businesses to select the most suitable site location. This comprehensive methodology accounts
for a plethora of factors, ensuring businesses make well-informed decisions that set the
foundation for their success.
Keywords: Geographic Information Systems (GIS), SWOT Analysis, Multi-Criteria Decision
Modeling, Game Theory, Site Selection, Business Strategy, Location Analysis
“ESTABLISHMENT OF GROUNDWATER DATABASE IN THE COASTAL AREA OF SOC TRANG PROVINCE BY USING QGIS”
Nguyen Vo Chau Ngan;
GIS-IDEAS Oral Presentation
The research was conducted to support the management of exploitation and use of underground water for Ward 2, Vinh Chau town, Soc Trang province, Vietnam. A total of 1,504 households with wells were interviewed directly about water use. In addition, water samples from 9 wells were also collected and measured for quality parameters. Survey information was compiled and then established into a database using QGIS software. Households exploit groundwater for daily life and crop irrigation with 73.54% and 72.54%, respectively. Additionally, households use groundwater for purpose of aquaculture. Most households do not treat water used for daily life purposes. The water used to irrigate crops increases in the dry season, causing difficulties for households using conventional water pumps. Households must upgrade water pumps through the most common forms: syringes, air pumps, and rocket pumps. Water source quality is still within allowable standards for pH and TDS parameters. Several specific maps have been built with the database built with QGIS software, and users can retrieve information according to personal requirements. Research results have been transferred to authorities to exploit and support the management of underground water resources.
“Estimate the amount of carbon emissions from deforestation and forest degradation over two decades in Kon Ha Nung Plateau, Gia Lai Province, Vietnam”
Hieu Huu Viet Nguyen;
GIS-IDEAS Poster Presentation
The Forests play an important role in reducing carbon emissions and stabilizing climate change. Research on forest status and biomass fluctuations helps us understand and evaluate factors affecting forest carbon sequestration capacity, biomass fluctuations and recovery rates. Research on assessing deforestation and natural forest degradation based on remote sensing technology with deep learning methods. In this study, we used SPOT-4 and Sentinel-2 remote sensing data to assess natural forest changes in the Kon Ha Nung Plateau, Vietnam during the period 2000 - 2022 (Overall Accuracy is 90,3%, Kappa = 0,87). Thereby, the causes of deforestation and forest degradation in Kon Ha Nung Plateau from 2000 to 2022 were assessed. The research results showed that the current forest status in the study area from 2000 to 2022 had major fluctuations, specifically: The total area of lost forest reached 56.159,9 ha, and the area of degraded forest was 17.206,4 ha. From data on the developments and causes of deforestation and forest degradation, the study calculated the amount of CO2 emissions of the Kon Ha Nung Plateau in the period 2000 - 2022 with a positive value of about 658.842 tons of CO2/year. This shows that natural forests in the study area are emitting more CO2 than the amount of forests can absorb.
“Extracting Travel Distance and Slope Failure Height Using Oriented Bounding Boxes”
Tatsuya Nemoto;
GIS-IDEAS Poster Presentation
In this study, a method has been developed to simply extract travel distance and slope failure height from polygon data of slope failure areas and DEM using oriented bounding boxes. The travel distance and height extracted by this method were compared with the results measured manually to verify their accuracy. Based on the result, we infer that this method is effective.
“Geological consideration for development in Banding Island, Malaysia using remote sensing and GIS”
Mohammad Firuz Ramli , Zulkiflee Abd Latif and Nasyairi Mat Nasir;
GIS-IDEAS Oral Presentation
Development in steep hilly area is consistently at risk of landslides. Banding Island in Malaysia was originally a summit of a hill with steep slopes on both sides, but became an island after the surrounding areas was inundated, turning it into parts of Temenggor Dam. The main objective of this study was to investigate the effect of development to landslide occurences in Banding Island, Malaysia utilising archived remote sensing data within the geographical information system (GIS) environment. Free archived Landsat imagery had been used to illustrate the development history of the area and its relationship to the landslide occurrences. The results indicated that slopes cleared for road construction were more prone to landslides compared to the areas that were not cleared. This paper demonstrates that integrating remote sensing and GIS is effective for incorporating geological consideration in planning development in hilly area.
“Geospatial Data Integration in Smart Soil Erosion Modeling”
Hoa Thi Tran;
GIS-IDEAS Oral Presentation
Soil erosion poses a significant threat to agricultural productivity, environmental sustainability, and infrastructure stability. Practically, traditional soil erosion models, such as USLE (Universal Soil Loss Equation) or RUSLE (Reversed Universal Soil Loss Equation) often lack the precision needed for effective prediction and management. This article explores the integration of geospatial data to enhance the accuracy and efficiency of soil erosion modeling. By leveraging remote sensing technologies, Geographic Information Systems (GIS), and advanced data integration techniques, we propose a comprehensive framework for smart soil erosion modeling. The integration of different geospatial data resources, such as freely remote sensed images, digital elevation models, land cover maps, and meteorological data facilitates a multi-dimensional approach to understanding soil erosion dynamics. Additionally, the application of deep learning neural networks algorithms provides robust predictive capabilities, addressing data quality and uncertainty. Through a series of case studies, we demonstrate the practical applications and benefits of this integrated approach. To what expect, the research findings underscore the potential of geospatial data integration in improving soil erosion predictions, informing policy decisions, and promoting sustainable land management practices.
“Identification of Morphological Features in Slope Failure Areas due to Heavy Rainfall Using Change Vector Analysis and Random Forest Classifier”
Mitsunori Ueda;
GIS-IDEAS Poster Presentation
Identification of slope failure areas is essential for solving geological hazard problems such as disaster prevention, urban planning and land development. Slope failure areas are identified by human interpretation using aerial photographs and satellite images. However, the manual interpretation has the problem of requiring a lot of time and labor. Furthermore, objectivity is insufficient due to differences in human interpretation. Therefore, there is a need for development of technology to automatically detect slope failure areas. Also, in order to understand the characteristics of the slope failure, it is necessary to investigate the relationship between the structures inside the slope failure areas, such as the scarp and main body. Slope failure is a phenomenon that creates spatially irregular terrain in a short time. In this study, we automatically identified slope failure areas with focus on temporal changes. Change Vector Analysis (CVA) and Random Forest Classifier (RFC) were used to identify scarp and main body in slope failure areas from changes in the Digital Elevation Models (DEMs) during two periods. CVA compares paired images from two different time periods and numerically analyzes the changes between the two periods by expressing them as the vector. In this study, changes in pairs of topographic features were analyzed using CVA. RFC was used to extract slope failure areas using change vectors as training data. The study area is located in a part of Tamba city in Hyogo prefecture, Japan. The target to be identified is the slope failure that occurred due to heavy rainfall in 2014. In this research, 1 m DEMs were used which was generated from airborne laser survey data acquired before and after slope failure event. Topographical characteristics were calculated from the DEMs at the two time periods and the pairs of each topographical features were combined. The results of the CVA showed different characteristics depending on the strength and direction of the change vector. The strength of the change vector suggested the possibility of determining the slope failure areas from the histograms. The direction of the change vector indicated that they could be useful for classifying scarp and main body in slope failure from the distribution of values. As the result of RFC, the learning features which contributed most to learning were the pair of terrain normal vectors and their variances, the second was the pair of elevation and Laplacian and the third was the pair of elevation and slope angle. The feature importance in RFC revealed that the geometric characteristics of the terrain significantly contributed to the classification, while the terrain conditions had a minimal impact on the learning process. The extraction accuracy was verified using the Cohen's kappa statistic. The result of accuracy was 0.75, which corresponds to “Substantial”. The results of extraction using the RFC were generally good. However, there were misinterpretations in the rivers and non-slope failure areas. This study focused on the changes in topographical characteristics from DEMs acquired before and after the event and identified scarp and main body in the slope failure areas. Identification of slope failure areas using topographical changes is useful for quickly generating the inventory maps of slope failure. Also, allows geological hazard mapping and updating, including location, shape, and morphology information of slope failure. Furthermore, revealing morphology of slope failure will be useful in predicting slope failures. The methodology described in this paper be applied to pre and post failure DEM derived from InSAR or high-resolution global DEM such as AW3D.
“INVESTIGATING LAND SUBSIDENCE BY PROCESSING MULTI-TEMPORAL SAR TIME SERIES ON GOOGLE COLAB: CASE STUDY IN CAMAU CITY, MEKONG DELTA, VIETNAM”
Ha Trung Khien;
GIS-IDEAS Poster Presentation
The phenomenon of land subsidence is widespread and complex in Ca Mau Province, stemming from both natural factors and human activities. Radar remote sensing technology has become an effective tool for monitoring and surveying land subsidence over large areas. The process of processing radar images to identify land subsidence not only requires specialized knowledge and experience but also necessitates powerful hardware systems to handle multi-temporal data. Therefore, this process can be costly and time-consuming. This study focuses on evaluating the application of the Google Colab online platform using the MT-SAR (Multi-Temporal Synthetic Aperture Radar) processing method to analyze a series of 24 Synthetic Aperture Radar images from January 2022 to December 2023 for Ca Mau City and surrounding areas. By comparing the results from Google Colab with field survey data, the study will assess the effectiveness and accuracy of land subsidence detection using the SBAS technique on the Google Colab platform, with results showing a correlation coefficient R2 of 0.80 and RMSE of 4mm.
“Investigation of Machine Learning Models for Slope Failure Suceptibility Zonation in parts of Yen Bai Province, Vietnam”
Tran Tung Lam, Tatsuya Nemoto, Venkatesh Raghavan, Xuan Quang Truong;
GIS-IDEAS Oral Presentation
Yen Bai Province in northern Vietnam, especially Mu Cang Chai (MCC) and Van Yen (VY) districts, are highly susceptible to slope failure due to rugged terrain, high rainfall and anthropogenic activities . In this research MCC was used as an area for training and testing the machine learning models, while VY serves for model validation due to similar topographic and geological conditions.
The methodology treats the slope failure prediction as a binary classification task (landslide/no-landslide). A balanced dataset of 286 landslide and 286 non-landslide points in MCC, along with 16 contributing factors, including topographic, geologic, hydrologic, anthropogenic and vegetation factors calculated from open data sources and made use from existing databases and from previous research on the area. Principal Component Analysis (PCA) and Pearson Correlation Coefficients refine the dataset by evaluating correlated factors and removing the least important ones, the size of the training dataset can be reduced while ensuring the performance of the ML models. Four ML models: Random Forest (RF), Support Vector Machine (SVM), Logistic Regression (LR), and Extreme Gradient Boosting (XGBoost) are trained and evaluated to select the best hyperparameter tuning for each model. Model accuracy is assessed via confusion matrices, accuracy score, ROC (Receiver operating characteristic) curves and AUC (Area under the ROC Curve).
Results show the models perform effectively in MCC with the average accuracy of all models being 0.74. The trained ML models with tuned hyper-parameters after running on MCC data, was validated on datasets for VY. The VY data also consists of 16 factors, with a data set of 308 landslide/non-landslide points. RF and XGBoost have the highest accuracy for both training and testing area (MCC) and Validation area (VY), with XGBoost showing a slightly higher accuracy score of 0.83 while RF scores 0.80.
The XGBoost model produces good results and could be further optimized to achieve even better zonation in future studies. The machine learning workflow can be applied on other areas that are prone to slope failures. Other geologic and weathering factors could be included in the analysis to further improve the model.
“Landslide Risk Mapping and development of landslide database and mobile app for Bhutan”
Younten Tshering;
GIS-IDEAS Oral Presentation
Bhutan experiences several landslides during the monsoon season. Although there are a few studies, they do not cover the entire country. Additionally, these maps are published in hardcopy, making them less accessible for reference. There is also no comprehensive landslide database in Bhutan. Therefore, this study aims to produce a digital landslide risk map of Bhutan and create a database through a mobile app known as Bhutan Landslide Information (BLI). The landslide map was produced using twelve influencing factors and a landslide inventory collected through field visits and from relevant offices. A total of 2,324 landslide locations were documented, with 70% (1,620) used for training and 30% (704) for testing. The analysis was conducted using logistic regression. Prior to the regression model, tolerance (TOL) and Variance Inflation Factor (VIF) were used to check for collinearity. The analysis showed all factors had a VIF less than 10 and tolerance above 0.1, indicating none of the independent factors were redundant. According to the logistic regression model, Land Use Land Cover (LULC) had the highest coefficient of 0.8027, making it the most influential factor for landslide events. The landslide risk map was validated using sensitivity, accuracy, and area under the curve (AUC), with values of 0.8852 and 0.9457, respectively, indicating high accuracy. The landslide map was then exported to QGIS2Web to be made available online and incorporated into the BLI mobile app. Users can locate their geolocation and view the landslide risk zone. Additionally, a landslide database was developed, allowing users to take photographs of landslide events through the BLI app. The app automatically records the landslide location (latitude and longitude) and the time the photograph was taken. Users can download the landslide database as a CSV file for research and decision-making purposes.
“MAPPING SOIL POLLUTION IN CAMAU CITY, VIETNAM”
Ho Dinh Duan;
GIS-IDEAS Oral Presentation
Camau city is located in far south of Vietnam, and recently experienced a fast social-economic development, in the same rhythm of the Mekong delta region. As a result, an environmental degradation has been seen. This paper focuses on the heavy metal soil pollution, which comes from various factors, including industrialization, as well as agricultural and fishery activities. Soil and water samples from surveys carried out by the provincial Department of Natural Science and Environment (Camau DONRE) in 2021 and 2022 were used for assessment of typical heavy metal contamination: Cu, Pb, Zn, Cd and As. The IDW spatial interpolation was employed for building the pollution map. In addition, we have based on the point source of pollution map to verify the contamination map just created and classify the level of harmful affects caused by these potential sources. For this purpose, the Dice coefficient was used to assess the dissimilarity between two subsets in an image. The findings of this study may help the local government regulate and mitigate the environmental degradation in this area.
Keywords: Heavy metal, soil pollution, IDW interpolation, point source pollution, Dice coefficient, Camau city.
“Monitoring the changes in vegetation area in Bangkok metropolis through NDVI analysis from Sentinel-2 images”
Pattamaporn Lerdlimchalalai;
GIS-IDEAS Oral Presentation
Vegetation area are vital for urban areas, particularly in large cities with high population densities such as Bangkok. This study monitors vegetation area changes in Bangkok metropolis by using the NDVI analysis from Sentinel 2 image on 2018 to 2024 applied knowledge in Geographic Information Systems (GIS) and remote sensing, including satellite image extraction to analyze the Normalized Difference Vegetation Index (NDVI) The aforementioned steps will use the raster calculator function in ArcGIS software for the calculations. Then we will transform raster to polygon so that we can have and attribute data to calculate the area of each range of NDVI and plot a graph to visualize the trends in vegetation area changes each year. Based on the analysis of the obtained NDVI values, the classification for each NDVI range is -1.00 to -0.050 = Water area -0.051 to 0.000 = Built-up areas and water area 0.001 to 0.150 = Built-up area 0.151 to 0.450 = bare land 0.451 to 0.550 = Vegetation near built-up area 0.551 to 0.750 = Vegetation-covered area and 0.751 to 1.000 = Densely vegetated area
the findings from the longitudinal monitoring of changes across different years reveal significant trends and patterns that provide valuable insights into the dynamics and transformations occurring within the studied parameters.
The result show that the green area (NDVI=0.045 – 1) are decreasing every single year accept in 2022 while the urban area (NDVI = -0.051-0.150)
“Oil Spill in the Southern Marine Regions of Vietnam: Models and Simulations”
Ho Dinh Duan;
GIS-IDEAS Poster Presentation
The marine economy plays a vital role in the era of economic development in Vietnam. However, it also causes various incidents for natural resources and the environment among which oil spills have caused significant environmental and ecological harms. Therefore, rapid forecasting of the spreading process of oil makes an important influence to supporting the decision of oil spill prevention and response, as well as minimizing economic losses and negative impacts on the environment and ecology. In this paper, by integrating the Euler-Lagrange method into the 3-D hydrodynamic model-based finite element method, we build a model on simulating the spread of oil spills with scenarios occurring in different hydrometeorological conditions in the southern waters of Vietnam.
The model exploited in this paper describes the horizontal composition (x,y) to reflect the preservation of heat and salt of the tangled current dynamics, and the mixing distance. Dirichlet boundary conditions are derived from tidal oscillations at the open boundary nodes, the sliding stress at the bottom and the air temperature at the surface.
Our case study of oil spill was taken from the collision of the Lisa Auerbach with the My An 1 in Ba Ria - Vung Tau sea, and the My An 1 ship containing 40 tons of oil on September 14, 2021, being a severe environmental risk. Simulation results of oil spread showed that oil spills had affected ecosystems of mangroves and wetlands in the Mekong Delta.
“Predicting land-use change using GIS and machine learning - a case study in Tuyen Quang city, Vietnam”
Bui Ngoc Tu;
GIS-IDEAS Poster Presentation
Bui Ngoc Tu, Le Phuong Thuy, Pham Le Tuan, Nguyen Xuan Linh, Tran Quoc Binh
University of Science, Vietnam National University, 334 Nguyen Trai, Thanh Xuan, Hanoi, Vietnam
Email: buingoctu@hus.edu.vn
The prediction of land-use change helps identify trends in land conversion and supports urban land-use planning. In this study, we utilized the Artificial neuron network-Cellular automata (ANN-CA) model and random forests (RF) algorithm to predict future land-use change and analyze the contributions of each driving factor in Tuyen Quang ciy, Vietnam. The model incorporated land-use maps in the years 2010, 2015, 2020 and 12 driving factors, categorized into three groups: natural, transportation, public services and population factors. A comparison of the simulated and actual land-use maps for 2020 achieving a strong overall accuracy of 87.58% and a Kappa coefficient of 0.83, thereby validating the dependability of the model’s predictions. Subsequently, this model was applied to predict land-use change for the years 2025 and 2030. Additionally, this study identified elevation, slope, distance to river, distance to main road and distance to trade center as the most significant factors influencing land-use change in Tuyen Quang ciy. This study offers urban planners useful insights to develop land use strategies and support the area's sustainable development goals.
Keywords: Machine learning, ANN-CA, Random forest, GIS, Land-use change, Driving factors.
“PRESENTING THE FLOOD HAZARD ANALYSIS IN WANGTONG DISTRICT, LOWER NORTH THAILAND USING GEO-INFORMATICS TECHNIQUES WITH OPEN ACCESS PROGRAM: HAZMAPPER.”
Kittituch Naksri | Chaiwiwat Vansarochana;
GIS-IDEAS Oral Presentation
Flooding remains a pervasive issue in Thailand, causing significant economic losses and affecting community well-being. This study evaluates the effectiveness of the Hazmapper program in assessing flood-induced vegetation loss, in Wang Thong District, Phitsanulok, Thailand, following the flood event of October 8, 2023. The primary objective is to assess the capability of Hazmapper in detecting and quantifying vegetation changes using the relative difference NDVI (rdNDVI) approach.
The Hazmapper program utilizes Sentinel-2 satellite imagery, with a pre-event observation period of 15 days and a post-event period of 9 days, to compute rdNDVI values. This methodology involves comparing vegetation indices before and after the flood to quantify vegetation loss. Geographic data from sources such as Google Earth and Google Maps were integrated to refine spatial accuracy and context.
The rdNDVI values indicated negative values in regions where vegetation had diminished due to flooding. These results were corroborated by visual comparisons of satellite imagery taken before and after the event, demonstrating a noticeable decline in green vegetation in inundated regions.
This study also highlighted several limitations associated with the Hazmapper program. First, the rdNDVI values are computed for entire areas, including regions not directly affected by flooding, such as harvested fields and deforested lands. This broad coverage can dilute the accuracy of flood impact assessments. Second, Hazmapper's reliance on vegetation indices limits its effectiveness in detecting flooding in non-vegetated or barren areas. In short-duration floods, minimal vegetation loss may result in negligible rdNDVI values, reducing the program’s sensitivity to such events.
In conclusion, this study demonstrates that Hazmapper is an effective tool for evaluating flood-induced vegetation loss, offering valuable information for disaster response and recovery efforts. The program’s application of satellite imagery and vegetation indices provides significant benefits for understanding and managing the impacts of flooding on vegetation and the environment.
“Processing and analyzing multi-source, multi-resolution geospatial data on cloud platforms for some environmental and disaster applications”
Van Anh Tran;
Keynote Talk
Recently, our earth has been facing increasingly severe conditions due to various climate change phenomena. Natural disasters such as storms, floods, and landslides have become more frequent. Monitoring changes or deformations in the land surface has thus become more crucial. While processing data over large areas often provides a comprehensive view for assessing the extent of changes and damage caused by natural and human factors, it requires handling vast amounts of data. Fortunately, cloud computing platforms have become more powerful and are now supporting numerous studies in earth science and environmental monitoring. This article provides an overview of how two cloud computing platforms, Google Earth Engine (GEE) and Google Colab (GC), are utilized to process large datasets for monitoring and forecasting land surface changes in various regions of Vietnam. Four examples of applications using GEE and Google Colab include: analyzing open-pit coal mine changes using Sentinel-1 and Sentinel-2 satellite images on the GEE platform; predicting landslides using Support Vector Machine (SVM), Random Forest (RF), and Gradient Boosting (GB ) machine learning models on the GEE platform; monitoring landslides from Sentinel-1 Radar image series using PyGMTSAR on Colab, and last is forecasting land subsidence using a subsidence value series from 2015 to 2023 showcasing the capabilities of these platforms.
“Relationship between Compound Extreme Indices and Atmospheric Circulation Patterns in Thailand”
Teerachai Amnuaylojaroen;
GIS-IDEAS Poster Presentation
This study examines the correlation between compound extreme indices and atmospheric circulation patterns in Thailand from 1970 to 2023. The aim of this study is to examine the consequences of various atmospheric indices, including the Pacific Decadal Oscillation (PDO), Southern Oscillation Index (SOI), Niño indices (3.4, 1+2, 3), Global Mean Sea Surface Temperature (GMSST), Quasi-Biennial Oscillation (QBO), Pacific Warm Pool (PACWARM), and the Trans-Niño Index (TNI), on compound extreme weather events. Composite analyses were performed to compare the average values, as well as the high and low extreme values, during periods of compound extreme heat and precipitation, as well as extreme conditions of Tmax35-SPI5day. The results suggest that the PDO and Niño indices typically have negative mean values during low extremes, indicating an outbreak of stronger La Niña conditions. During periods of extreme conditions, the Southern Oscillation Index (SOI) experiences an increase, which suggests the existence of stronger La Niña signals. The Global Mean Surface Temperature (GMST) exhibits a consistent and continuous upward trend during periods of high extremes, thus emphasizing the ongoing phenomenon of global warming. During periods of high extremes, the Quasi-Biennial Oscillation (QBO) experiences substantial declines, whereas the Pacific Warm Pool (PACWARM) index shows an increase. Conversely, during periods of low extremes, the PACWARM index decreases. The TNI exhibits variations in ENSO phases, declining during periods of high extremes and rising during periods of low extremes. The results reveal the dynamic characteristics of atmospheric circulation patterns and their vital effect on climate patterns and compound extreme events in Thailand.
“The systematic effectiveness comparison of spatiotemporal carbon dioxide emission with spatial interpolation methods: case study in Thailand”
Suriyawate Boonthalarath;
GIS-IDEAS Oral Presentation
The rapid increase in greenhouse gases, especially the high concentration of carbon dioxide in the atmosphere, has aggravated the effects of the worsening global warming crisis. There is a growing focus on managing and tracking spatio-temporal carbon emissions to achieve carbon neutrality in academic research. However, it is challenging to measure and assess the dynamic release of human-made and natural carbon dioxide across broader spatial resolutions due to numerous missing data points. The interpolation method is an essential tool for predicting missing data in datasets that monitor variations in carbon dioxide from sampling points, satellite measurements, or ground-based observations. Various interpolation techniques have recently been used to analyze and evaluate the continuous increase in carbon dioxide pollution at different spatial scales across various disciplines. These techniques have proven to be valuable in decision-making and problem-solving. Unfortunately, there is a lack of research explaining the specific use of these systematic interpolation techniques. This study aims to compare the performance of four interpolation techniques in evaluating carbon dioxide emissions. The objective is to analyze the spatial and temporal dynamics in atmospheric carbon dioxide concentration by comparing the geo-statistics of the satellite OCO-2 dataset and the ground-based TCCON dataset within Thailand's boundaries from 2015 to 2020. Furthermore, each interpolation technique was assessed for accuracy using Root Mean Square Error (RMSE) in cross-validation studies as the criteria for selecting the most appropriate spatial interpolation techniques for specific scenarios. It was found that the four interpolation techniques demonstrated strong predictive performance in visualizing and analyzing Thailand's boundaries, as well as revealing the spatiotemporal patterns of atmospheric carbon dioxide concentrations. Although the results indicated that IDW and Spline interpolation outperformed the Natural Neighbor and Kriging methods, achieving a total RMSE of less than 0.01 in cross-validation. When comparing the IDW and Spline interpolation methods, it was discovered that the Spline algorithm showed overfitting for both datasets in comparison to IDW. As a result, IDW demonstrates a more suitable sensitivity variation for improving atmospheric carbon dioxide concentration in response to spatiotemporal human-made emissions.