This research analyse the pattern of rice field productivity. It was identified from the characteristics of the land component based on physiographic condition through landscape perspective. Identification of productivity pattern has been done partially based on each typology of land components into several segment of the Citarum watershed, West Java Province, Indonesia. Spatial autocorrelation through GIS tool was used as the method in this research. By using moran’s I (index) measurement, degree of dependency of these variables are generated to find the spatial pattern. The result of this study is separated the value of productivity based on segments of watershed, the values of the average of productivity are upstream (6.39 ton/Ha), middle stream (6.52 ton/Ha), and downstream (7.17 ton/Ha), sequentially. The highest productivity is in the downstream area (9.83 ton/Ha) and the lowest is in the upstream area (4.55 ton/Ha). In accordance with physiographic typology showed the rice field in the middle stream has more variation than the upstream or teh downstream area. The highest of average rice field productivity is on alluvial plain. Overall, the rice field productivity on the hills is higher rather that other types of landform the stuctural formation is more dominant, in addition. The spatial pattern shows the distribution of rice field productivity most likely to clustered based on the similarity of physiographic type. Statistically, it has p-value <0.01 and z-score >2.58 (239.26) correspond to Spatial Autocorrelation (Moron’s I). This positive value means a less than 1% likelihood that this clustered pattern could be result of random choice, which the rice field productivity value has similar pattern to others. Thus, it can be generated that the pattern of rice field productivity has a very close relation with the physical characteristics which associated of each typology of land components.