Abstract. For many years the Shenmu watershed has been heavily impacted by landslides induced by extreme rainfall events, with an even greater impact in recent years due to climate change in addition to the Chi-Chi Earthquake after-effects. This study utilizes remote sensing technology to spatially and temporally interpret landslide processes in the Chushui and Aiyuzih sub-watersheds. An event-based landslide dataset is constructed which consists of 11 historical disaster events including 17 satellite images spanning the past 14 years. The study explores the contribution of causative environmental factors and other factors, which are based on the physiographic conditions and geographic locations of the landslides induced, on landslide potential. These factors are utilized to construct a logical reason-based rule set to build a framework of procedures for semi-automated image interpretation and artificial image identification. Spatial relationships show that landslides are frequently found in areas at 1500–2000 m of elevation with slope gradient over 55 % W-SE orientation, and within the Nanchuang Formation adjacent to a 25 m buffer zone of a river course in the Chushui and Aizuyih sub-watersheds. Landslide occurrences are prevalent on both sides of the river course and are the direct suppliers of sediments to the river, leading to sediment related disasters. Temporally, it is found that the typhoon-induced landslides can be subdivided into three distinct time intervals and the event that caused the greatest increase in landslide area can be recognized. These intervals and their greatest impact events are as follows: (1) Before the 1999 Chi-Chi earthquake, the 1996 typhoon Herb; (2) From the 1999 Chi-Chi earthquake to 2009 typhoon Morakot, the 2004 typhoon Mindulle; (3) After this time period, the 2009 typhoon Morakot. After comparisons were made of the total landslide areas and the new landslide areas before and after the 1999 Chi-Chi earthquake in the two sub-watersheds, it was found that the earthquake amplification effect of the quantified magnification was estimated to be at least a doubling effect. This is an estimate that agrees well with the previous studies. The methodology of our extensive study can be utilized to improve the dataset accuracy in similar research, to classify and differentiate the contribution of environmental factors to landslide occurrences and to build landslide occurrence potential maps for sub-watersheds. These results are important to decision-makers to improve the reference information basis for preliminary evaluation of the exposure of elements at risk. This, in turn, is important for improving early warning systems, rapid response mechanisms, evacuation protocols and long term mitigation solutions. The results may also influence the recommendations for remediation of slope areas and construction of preventative engineering solutions in the two sub-watersheds analyzed using the landslide potential map to prioritize urgencies by comparing the necessity in one area to the next. Keywords: Shenmu watershed, environmental factors, landslide potential map.
The discussion paper was formally withdrawn.
Fu, K.-L., Lin, B.-S., Thomas, K., Chen, C.-K., and Ho, H.-C.: Evaluation of Environmental Factors in Landslide Prone Areas of Central Taiwan using Spatial Analysis of Landslide Inventory Maps, Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2016-127, 2016.