The main suffering area of the Wenchuan Earthquake locates in the mountainous area of western Sichuan

an area which is very prone to mountain hazards such as landslides and debris flows because of the mountains and deep valleys

complex geological structure as well as well developed fault zones there.The Wenchuan Earthquake had directly induced massive secondary disasters such as collapse and landslides which destructed large area of farmlands.At the same time

the Wenchuan Earthquake also induced more than 30 barrier lakes making some farmlands be inundated and destructed.In order to make a scientific evaluation of the disaster situation and provide scientific basis for reconstruction and rational settlement of people in the suffering area

it is necessary to acquire the information of the destructed farmlands.In this paper

satellite remote sensing data as well as appropriate information extraction methods were used to acquire the distribution information of farmlands before the Wenchuan Earthquake and information of collapse

landslides

barrier lakes after Wenchuan Earthquake based on information such as landform features in mountainous area

landuse information and imagery characteristics of seasonal aspect.Spatial analysis was applied to evaluate the farmland loss to provide reference for the reconstruction planning.In this paper

Tangjiashan area of Beichuan county was taken as a case study area for evaluating the farmland loss caused by secondary disasters.NDVI was derived from the SPOT image data of November 10th

2007.Based on NDVI and landform information

the farmlands were identified and the distribution information of farmlands was obtained.Manual corrections were then used to eliminate possible errors in the previous farmland identification.The FORMOSAT-II image of May 14th

2008 and the ALOS image of June 4th

2008 were used to acquire the information of these secondary disasters such as collapse

landslides as well as barrier lakes induced by the Wenchuan Earthquake based on man-computer interactive interpretation.Based on GIS technology

the rapid evaluation of farmland destruction was carried out according to the comprehensive analyses of farmland distribution before the Wenchuan Earthquake and the distribution information of these secondary disasters induced by the Wenchuan Earthquake.Then the temporal and spatial characteristics of farmland loss were analyzed.The results show that the secondary disasters such as collapse

landslides as well as barrier lakes were very serious in the study area.There were 533 sites of collapse and landslides which had a total area of 1408.20 hm2

as well as barrier lakes with an area of 604.69 hm2 in the study area.The area of the farmlands destructed by secondary disasters were 86.88 hm2 accounting for 8.84% of the total farmland area.The distribution of collapse

landslides and destructed farmlands was consistent with the distribution of fault zone.And the destruction rate of the farmland on the fault zone was 18.74%

accounting for 74.53% of the farmlands destructed in the study area.The secondary disaster induced by the main shock had seriously destructed the farmlands while the aftershocks furthered the destruction.With the advent of rain season

the secondary disaster will still be of great threats to the farmlands so it is necessary to strengthen the dynamic monitoring of the farmland in the study area.As there is good forest coverage in this area

suitable seasonal aspects with vegetation index were used to extract the distribution information of farmlands before the Wenchuan Earthquake with the assist of landform information.In the FORMOSAT-II and ALOS imagery the distribution information of collapse and landslides induced by the earthquake was quickly extracted by visual interpretation according to the hue

shape information in the imagery

landform features and reference to the satellite imagery before the earthquake as well

then the farmland loss information was acquired using GIS.However

the rapid acquirement of the collapse and landslides information by automated classification needs further study.