This study took place in the city of Semarang. The purpose of this study were: (1) to compare the accuracy of the map of land use decision tree classification results with maps of land use classification results maximum likeness integrated with geographic information systems, (2) an inventory of Semarang land use classification methods that have the highest accuracy in the data Landsat multiwaktu, (3) assess changes in land use classification results Semarang city that has a higher accuracy rate.
The method used in this study using the classification decision tree for mapping land use that incorporates 6 channels spectral Landsat TM / ETM + path / row 120/65 recording in 1994, 2002 and 2006 with layer support, namely: the maps results Transformation Kauth and Thomas, NDVI (Normalized Difference Vegetation Index), NDBI (Normalized Difference Building Index), vegetation index, and spatial data such as maps of landforms, soil maps, elevation maps and map slopes. For comparison selected the maximum similarity classification, integrated with a geographic information system to be reduced to the land use map. Land use classification used has two different levels of detail for the scale of 1: 100,000 and 1: 250,000. The results of the two methods above then compared the level of overall accuracy, user's accuracy and producer accuracy and Kappa index. Highest level of accuracy will serve as an inventory of land use data for the city of Semarang. The next method is to assess changes in land use of Semarang visually using the road network and a map of Semarang RTRW Year 2000-2010.
In this study show that the classification results of land use map decision tree has an overall accuracy rate (overall accuracy) and Kappa index were higher than the maximum similarity classification results are integrated with geographic infromasi system. The results of land use classification level I have better accuracy than the land use classification level II. For classification level on the map in 1994 I obtained 54.14% accuracy with a Kappa index of 0.4822 for maximum similarity and classification accuracy of 66.34% with a Kappa index of 0.6256 for the classification decision tree. In the 2002 map obtained results with an overall accuracy of 75.12% Kappa index of 0.713 for maximum similarity and classification accuracy of 81.46% with a Kappa index of 0.787 for the classification decision tree. On the map in 2006 obtained an overall accuracy of 78.05% with a Kappa index of 0.7641 for maximum similarity and classification accuracy of 82.45% with a Kappa index of 0.805 to map land use decision tree classification results. Changes in land use in the city of Semarang lead to a decrease in agricultural land and plantations and the growing residential and industrial.