Abstract: 【Objective】Recently, researches on identifications and dynamic changes of landscapes and land use types by using remote sensing techniques have been a hot topic. However, the vast majority of studies have taken farmland as a “single” land type; the spatial distribution and variation of different crops inside the farmland have been neglected. This paper aims to explore the extraction methods of large scale crop acreage and distribution pattern by using remote sensing and the application of landscape pattern indices in crop pattern dynamics. 【Method】 Based on the full coverage MODIS images and NDVI data during the crop growing periods of 2005 and 2010, by analyzing the planting structure, phenology calendar and NDVI time series curve characteristics, different area extracting models were established and were used to extract the spatial distribution of main crops (spring maize, soybean and paddy) by using RS and GIS techniques in Northeast China. Meanwhile, some landscape pattern indices were used to describe the characteristics and rules of crop pattern dynamic changes.【Result】Compared with the average statistical data of several years, the overall areas extraction accuracies of these two years were more than 90%. The main crop planting structure changed a lot from 2005 to 2010 in Northeast China. Soybean area decreased obviously, its dynamic degree reached -4.47%, and the average patch area reduced by 0.05 km2. Change range of paddy and spring maize reached 22.37% and 22.82%, respectively, during the 5 years. And the average patch area also increased. 【Conclusion】 Increasing planting costs and decreasing relative benefits were main reasons for these changes. It is technically feasible for large scale crop acreage extraction by using medium resolution remote sensing data. Landscape ecology pattern index can be used to analyze crop pattern dynamic changes.

Key words: remote sensing , NDVI , pattern change , paddy , spring maize , soybean , Northeast China