• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.1
• /
• pp.133-142
• /
• 2010

To monitor composition and change of the national land, intelligent topographical classifier which enables accurate classification of land-cover types from airborne LiDAR data is highly required. We developed a topographical classifier development support system cooperating with da1a mining tool WEKA to help users to construct accurate topographical classification systems. The topographical classifier development support system has the following functions; superposing LiDAR data upon corresponding aerial images, dividing LiDAR data into tiles for efficient processing, 3D visualization of partial LiDAR data, feature from tiles, automatic WEKA input generation, and automatic C++ program generation from the classification rule set. In addition, with dam mining tool WEKA, we can choose highly distinguishable features by attribute selection function and choose the best classification model as the result topographical classifier. Therefore, users can easily develop intelligent topographical classifier which is well fitted to the developing objectives by using the topographical classifier development support system.

• Proceedings of the KSRS Conference
• /
• 1999.11a
• /
• pp.271-280
• /
• 1999

The topographical effect resulted from rugged terrains and inhomogeneous spectral characteristics due to the complexly mixed land cover condition of Korea substantially lower the remotely sensed land cover classification accuracy In this study, a topographic correction method using digital elevation model to alleviate the topographic effects. To deal with inhomogeneous spectral characteristic, a hybrid classifier with inclusion of prior probabilities was introduced. This investigation concluded that the topographical normalization and hybrid classification with prior probabilities are effective on rugged landscape. The overall and average classification accuracies were improved by 0.92％ and 1.016％ respectively. The most substantial and noticeable accuracy improvement was observed in forest areas.

• Journal of Korea Water Resources Association
• /
• v.51 no.6
• /
• pp.481-490
• /
• 2018

Recently, the spatiotemporal patterns of flood disasters have become more complex and unpredictable due to climate change. Flood hazard map including information on flood risk level has been widely used as an unstructured measure against flooding damages. In order to product a high-precision flood hazard map by combination of hydrologic and hydraulic modeling, huge digital information such as topography, geology, climate, landuse and various database related to social economic are required. However, in some areas, especially in developing countries, flood hazard mapping is difficult or impossible and its accuracy is insufficient because such data is lacking or inaccessible. Therefore, this study suggests a method to delineate large scale flood-prone area based on topographic factors produced by linear binary classifier and ROC (Receiver Operation Characteristics) using globally-available geographic data such as ASTER or SRTM. We applied the proposed methodology to five different countries: North Korea Bangladesh, Indonesia, Thailand and Myanmar. The results show that model performances on flood area detection ranges from 38% (Bangladesh) to 78% (Thailand). The flood-prone area detection based on the topographical factors has a great advantage in order to easily distinguish the large-scale inundation-potent area using only digital elevation model (DEM) for ungauged watersheds.

• Korean Journal of Soil Science and Fertilizer
• /
• v.34 no.5
• /
• pp.342-349
• /
• 2001

This study was carried out to classify and evaluate the land cover map using Landsat TM data in Pyeongtaeg City. DGPS data, aerial photography, topographical map were used for selection the training sets and accuracy assessment. The overall accuracy and Kappa coefficient of the land cover classification map(using supervised classification with 13 classes) with Landsat TM data(16 June. 1997) were respectively, 86.8%, 85.4%, but the user's accuracy of urban/village and vinyl-house was below 60%, and the producer's accuracy of read and vinyl-house below 70%. Maybe it was caused the spectral reflectance characteristics, heterogeneity and small distribution area on the artificial things such as urban/village, vinyl_house and road, etc. And then, the agricultural land cover classification system using remote sensing data in Korea was to classify level I and II. Level I consisted of 5 classes such as agricultural land, forest land, water, barren land, urban and built-up land.