The object of this study is to make sure of the availability of aerial photographs in investigating the damages the typhoon 7513 did to the Island Hachijo. From the result of the photo interpretation, a multiple regression equation to induce the damage rates from combinations of the topographical factors was worked out by multiple regression analysis. As the result of this, aerial photos were recognized as useful to the investigation of storm damages as well as to the recognition of the damage distribution in view of the fact that as the photos show, the damages were mostly primary ones due to the sheer verocity of the wind and that the indirect damages from secondary causes, such as pebbles, scraps of sheet zinc, etc. were comparatively few. Hence their usefulness to the analysis of the damage distribution. Moreover it became clear that the damage distribution chart based on the photo interpretation and the multiple regression equation show the damage distribution have been influenced by the differences of the topographical factors. Among factors which control the damage distribution, the sea rate, i. e. the percentage of the sea space in an area is most dominating in the case of the Island of Hachijo. The regional discrepancy between the multiple regression equation and the results of the aerial photographs is supposed to be due to some local terrain factors, such as a windbreak forest, hillocks, etc.
VHRR data of NOAA have been received and used for weather forecast at Meteorological Agency of Japan. Those NOAA data are recorded in magnetic tape. They are reproduced and converted to digital signals and processed by an image processing system, IMAGE-100. VHRR-IR image of sea surrounding east-northern part of Japan was produced and sea surface temerature was sliced into eight grades at 2°C interval in the range from 7°C to 25°C.
The study deals with two types of mathematical models for estimation of vegetation cover in Tokyo Districts from LANDSAT MSS digital data. Two types of models, multi-regression model and parametric model were applied to the test area with ground surveyed data of vegetation cover collected in 1973 by the Ministry of Construction and geographically corrected LANDSAT MSS digital data, taken on 26th Nov. 1972. Multi-regression model is based on the average data in the unit area of so called“500 meters mesh”, while parametric model is based on each of pixels in the unit area. From the results of the study, accuracies of the estimation for these two model were not remarkably different. Vegetation cover could be well estimated with the accuracy of 0.87 in multi-correlation coefficient, and ±15% in standard deviation.