Kuan-Tsung Chang

Automatic Case-Based Reasoning Approach for Landslide Detection: Integration of Object-Oriented Image Analysis and a Genetic Algorithm

This paper proposes an automatic method for detecting landslides by using an integrated approach comprising object-oriented image analysis (OOIA), a genetic algorithm (GA), and a case-based reasoning (CBR) technique. It consists of three main phases: (1) image processing and multi-image segmentation; (2) feature optimization; and (3) detecting landslides. The proposed approach was employed in a fast-growing urban region, the Pearl River Delta in South China. The results of detection were validated with the help of field surveys. The experimental results indicated that the proposed OOIA-GA-CBR (0.87) demonstrates higher classification performance than the stand-alone OOIA (0.75) method for detecting landslides. The area under curve (AUC) value was also higher than that of the simple OOIA, indicating the high efficiency of the proposed landslide detection approach. The case library created using the integrated model can be reused for time-independent analysis, thus rendering our approach superior in comparison to other traditional methods, such as the maximum likelihood classifier. The results of this study thus facilitate fast generation of accurate landslide inventory maps, which will eventually extend our understanding of the evolution of landscapes shaped by landslide processes.

Observing Land Subsidence and Revealing the Factors That Influence It Using a Multi-Sensor Approach in Yunlin County, Taiwan

Land subsidence is a worldwide problem that is typically caused by human activities, primarily the removal of groundwater. In Western Taiwan, groundwater has been pumped for industrial, residential, agricultural, and aquacultural uses for over 40 years. In this study, a multisensor monitoring system comprising GPS stations, leveling surveys, monitoring wells, and Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) was employed to monitor land subsidence in Western Taiwan. The results indicate that land subsidence in Yunlin County was mainly affected by the compaction of subsurface soils and over-pumping of groundwater from deep soils. The study area comprised western foothills, characterized by sediments containing predominantly gravel, and coastal areas, where clay was predominant. The subsidence in coastal areas was more severe than that in the western foothills, as a result of groundwater removal. An additional factor affecting subsidence was the compaction of deep layers caused by deep groundwater removal and the deep-layer compaction was difficult to recover. Based on multisensor monitoring results, severe subsidence is mainly affected by compaction of subsurface soils, over-pumping of groundwater from deep soils, and deep soil compaction.

AN OBJECT-ORIENTED ANALYSIS FOR CHARACTERIZING THE RAINFALL-INDUCED SHALLOW LANDSLIDE

Landslides are natural phenomena for the dynamic balance of the earths surface. Because of frequent occurrences of typhoons and earthquakes in Taiwan, mass movements are common threats to peoples lives. In this paper, the interpretation of knowledge is quantified as recognition criteria. Multisource high-resolution data, for example, a SPOT satellite image, 20m×20m Digital Terrain Model (DTM) reduced from Light Detection And Ranging (LiDAR) data, and aerial orthophotos were used to construct the feature space for landslide analysis. Landslides were recognized by an object-oriented method combining edge-based segmentation and a Supported Vector Machine (SVM) method. The classification results are evaluated in comparison with those by manual interpretation. Two cases from northern and central Taiwan are tested. Both cases show that the object-based SVM method is better than a pixel-based method in classification accuracy. The commission error of the proposed method is also smaller than that of the pixel-based method. Moreover, except for the spectral features, the slope and Object Height Model (OHM) characteristics are also important factors for improving landslide classification accuracy. Further study is required for assessing the mixed pixel effect when the resolution is as large as 20 m and for characterizing the effects of sampling rates or scaling caused by changes in resolution.

Accuracy assessment of land use classification using hybrid methods

Hillside region accounts for 73.6% of the land in Taiwan. The mountain region consists of high mountain valley of deep and faults-knit environment, fragile geological, abrupt slopes, and steep rivers. With the rapid development in recent years, there has been not only great change in land use, but the destruction of the natural environment, the improper use of soil and water resources also. It is prudent to effectively build and renew the existing land use information as soon as possible. Among various land use status investigation and monitoring technology, the remote sensing has the advantages in getting data covering wide-range and in richness of spectral and spatial information. In this study, hybrid land use classification methods combining with an edge-based segmentation and three kinds of supervised classification methods, means Maximum Likelihood, Decision Tree, and Support Vector Machine, were conducted to automatically recognize land use types for Yi-Lan area using multi-resource data, e.g. satellite images and DTM. The second land use investigation result of Taiwan in 2006 by the Ministry of the Interior is assumed as the ground truth. The higher classification accuracy results indicate that the proposed methods can be used to automatic classify agricultural and forest land use types. Moreover, the results of object-based DT and object-based SVM are better than the ones for the object-based ML methods. However, adequate training is not easy to select the appropriate samples for the transportation, hydrology, and built-up land classes.

Estimation of carbon sequestration by using vegetation indices

The Global warming forces countries in the world to pay efforts reducing the amount of carbon emission. Forests play an important role in terms of absorbing and storing carbon dioxide. Many studies indicated that carbon sequestration can be efficiently estimated via a method combining with forest inventories and remote sensing. Therefore, a SPOT5 and a FORMOSAT-2 satellite image are used to extract different kinds of vegetation indices, e.g. the Normalized Difference Vegetation Index (NDVI) etc., incorporated with field data to calculate amount of carbon sequestration via four regression analysis methods. The results show that the estimation is workable with biomass indices and proposed regression methods. Moreover, the proposed regression models are better suitable to estimate the woods volume of two forest types, e.g. pure artificial coniferous and artificial broadleaf mixed forest. However, the variants of MRA or BPNN can more accurately estimate the amount of carbon sequestration than original methods.

Accuracy evaluation of ALOS DEM with airborne LiDAR data in Southern Taiwan

Recently, some global-scale DEM products, such as GTOPO30, ETOPO1, SRTM and ASTER GDEM have been published for geoscience applications. The latest product, ALOS DEM was announced to be available for a global coverage in 2016. This study examined the performance of ALOS-DEM in describing accurate morphometric and volumetric measurement of land features. A comparison was made on basis of DEM and DSM data of airborne full-waveform LiDAR data. Results showed that ALOS DEM is more approximately in reality an ALOS DSM which reveals the ground envelop surface rather than the ground bare surface. The differences between ALOS DEM and LiDAR DSM are mainly from 0 to 2.75 m with a standard deviation of 1.58 m. The differences between ALOS DEM and LiDAR DEM give a bias of as large as 20m, mostly located at the areas with abrupt change of relief and mainly in the north-facing slopes. This is probably due to ALOS sensors geometry in corresponding to its looking-direction. The stream networks derived from both ALOS DEM and LiDAR DEM are in good agreement. It is suggested that further studies on methods for assessing geomorphometric changes in landform structures should be developed and compared.

Land subsidence monitoring with a network of continuously operating GPS stations in Yunlin County, Middle Taiwan

Land subsidence in western Taiwan has been an issue since 1970s due to over extraction of ground water for aquaculture. This problem became serious in Yunlin County because high-speed rail way transects the center of land subsidence which may cause safety problem. Leveling survey is applied annually. A more efficient method with 6 continuous GPS stations is used in this study to observe the subsidence. The effectiveness and advantages are compared with leveling survey and underground monitoring wells in the period from 2007 to 2012. It is concluded that the trend of the subsidence can be fully observed by the GPS network and the observation frequency of GPS prevails the annual survey of traditional approach which is an important factor for early warning purposes.

A statistical analysis for characterizing landslide caused by heavy rainfall and severe earthquake

Heavy rainfall and earthquakes are the two major factors inducing landslides in Taiwan. The distribution of area size is the most basic quantitative parameter of landslides. Therefore, the purpose of this study is to characterize the scale and spatial difference of rainfall-induced as compared with those of earthquake-induced landslides. Two representative landslide cases, Toraji typhoon occurred in 2001 and 921 earthquake occurred in 1999, are used to analyze the causes of different kinds of landslide disasters in the paper. The test area for the 921 earthquake case is 3700 km2, the recognized number of landslide in the event is 7279, and total area for the landslides is 14766 Ha. Moreover, the maximum area of one landslide is 532 Ha, the average area for the landslides is 2 Ha, and its corresponding standard deviation is 13 Ha. In the test case of Toraji typhoon, the total study area is 8847 km2, the number of landslides is 10359, and total area for the landslide is 22305 Ha. The maximum area of a landslide is 232 Ha, the average area for the landslides is 2.2 Ha, and its corresponding standard deviation is 6.2 Ha.

Combining multi-temporal satellite images and a cloud platform to develop new evaluating procedures for landslide vulnerability study

Taiwan is a mountainous and geologically active country with but land resource for residential development is limited. Under a crowded population pressure, the use of land resources is often overrun. Many hillside had been improperly developed. It causes significant impact on natural environment and the peoples lives and property. This paper proposed to use the archival images of the Formosat-II satellite in the last nine years to explore the high landslide vulnerability areas. Susequently, the interpretation results are published on a dedicated three-dimensional digital platform in the cloud for quick share with peoples concerned. The new procedures are an innovation of a stream-line of image platform and results of image interpretation on an innovated 3d platform. This study shows that landslide prone areas can be effectively detected with simple automated procedures with multi-temporal images. The areas with high activity which requires long-term observation or monitoring can be enhanced and published to the public by the on-line platform.

Estimating surface temperature and land cover change in Hsinchu City

Changes in land use and land cover (LULC) are the critical driving forces of change in atmospheric, climatic and ecological systems. The purposes of this study are thus aiming to understand the relationship between land cover changes and thermal properties of urban heat island effects using Landsat images spanning from 1991 to 2007. On basis of the images of 17 years, the temperature differences between urban and rural of Hsinchu had been raised from 2–3 degrees to 4–5 degrees. Moreover, local hot spots showed up in downtown, Nan-Liao fishing port, and the traffic arteries surrounding residential area. It shows that urban expansion and heat discharged by traffic vehicles are major reasons for the temperature anomaly in the study area. Further study is required for understanding the sources of these features.