1 Introduction The purpose of this study is to verify the applicability of the four-factor prediction model constructed from Shincheon river data and verify the validity of the model by comparing different data based on the Ota river landscape photographs and GIS-based CG pictures.<br> 2 Outline of the study In this study, we made a pectination of the physical and psychological evaluation data of the above three photos and pictures to grasp the characteristics of the river landscape. Then we used these three data for multiple regression analysis and compared the results. Particularly, we selected the same psychological questionnaire items, and compared the two stages of 「PHe→PSe(items)」 and 「PSe(items)→PSe(satisfaction)」 to test the validity of the multi-level prediction model.<br> 3 Outline of the subjects Three types of data are used in this study. There are 28 photos of Shincheon river landscape, 48 photos of Ota river landscape, and 48 Ota river CG pictures made by GIS. We extracted physical evaluation values and psychological evaluation values for these three types of data. The physical evaluation used three indicators such as OA, OG, OO, and four elements such as building, green, mountain and sky. The psychological evaluation is based on a questionnaire corresponding to the 4 factors, such as "F2. Openness ", "F3. Complexity ", "F4. Constructiveness ", "F5. Green-visibility ".<br> 4 Conclusion Major findings are as follows:<br> 「PSe(items)→PSe(satisfaction)」<br> 1) At first, representative items of four factors were selected to construct predicting models of satisfaction of three data. The results indicated that it is feasible that satisfaction can be sufficiently explained. About effectiveness of the predicting model of satisfaction, particularly F4 and F5, even have relatively larger contributions of Ota River's two data, which can be inferred by p value of these two explanatory variables. It shows the validity of the predicting model of satisfaction from four factors, and also it is recognized that the CG image has a very high effectiveness in the predicted structure.<br> 2) Regarding the relationship between representative items and satisfaction, F5, F4 and F2 have similar trends in the three data, but F3 tends to be different. Also, from the multiple regression model, the weights of F4 and F2 are almost the same in the three data, whereas the weight of F5 is different between Shincheon and Ota river, which is shown that the relationship between these two factors and satisfaction more depends on the feature of different cities and landscape types, direction of photography, etc. On the other hand, the weights of F3 are all different in the three data.<br> 「PHe→PSe(items)」<br> 3) We examined each of the three kinds of data and showed extremely high commonality regardless of data in F5, and also in previous studies, with no contradiction. On the other hand, different explanatory variables in other factors may give relatively high explanatory power, but also shows that dependence on data due to differences in feature of different cities and landscape types is recognized to some extent.<br><br> In the future, we will try to improve the accuracy of the predicting models with high commonality, and for those with low commonality, we will study the relationship between the features of the evaluation object group and the strong influence factors, to propose new physical indicators with high explanatory power.