The purpose of this study is to evaluate the influence of the atmospheric correction errors on ASTER TIR data, especially in humid atmosphere such as in Japanese summer. To understand the characteristics of humid atmosphere, we originally derived typical new model atmospheres of Japan from Aerological Data of Japan. These models are the average seasonal profiles of temperature and relative humidity in North, Middle and South areas of Japan. Then these were compared with the generic model atmospheres in LOWTRAN 7 for transmittance. These results indicate that the transmittance of Japan Middle area Summer model and Japan South area Summer model is smaller than that of Tropical model in LOWTRAN 7, which reflects the difference of total water vapor content for each atmospheric model. The influence of the error of water vapor content (WVC) on the surface brightness temperature (SBT) was evaluated by the computer simulations using LOWTRAN 7 from a viewpoint of atmospheric correction. The results are summarized as follows : The SBT errors are more sensitive to the WVC errors in humid model atmospheres. Thing to be emphasized is that the SBT errors in Japanese summer atmosphere are relatively large. These results indicate that the summer data observed in east or southeast Asia would need careful atmospheric correction. The general tendency was also found that the SBT error of Band 10 and 14 are larger than those of the other ASTER TIR bands. Moreover, the sensitivity of the SBT errors to each WVC error was evaluated in several atmospheric layers, which were equally divided multi-layers in troposphere. It is shown as a general trend that the sensitivity of the verical profiles on SBT decreases as the altitude of atmosphere increases. In detail, however, it is also indicated that the SBT errors of Band 10 are highly influenced by the WVC errors in the middle and upper layers, and that the sensitivity of each layer varies with real surface temperature.
The diffusion state of SO2 in the atmosphere have been evaluated by digital simulation method using the Air Diffusion Model. However, it is a few possibility to find the affected site at with low (high) SO2 concentration by only using the Air Diffusion Model as existing method. In this case, from several findings by our past researching, it can be solved by using the growing index vegetation which shows obvious reaction against SO2 concentration change. Comparing with the vegetation which normally growing condition, it is shown a tendency that the spectrum of the vegetation with bad growing condition rises at the visible red band (R) and falls at the near infrared band (NIR). If evaluating the vegetation condition grown at the SO2 polluted area by the reflectance characteristics (NIR/R) as an index, the affected site can be distinguished between low concentration site and high concentration site. In this study, it is proposed the making process for the SO2 distribution map and the evaluation method of the air polluted condition around urban area by using that map, and also reported the verified results of the analyzing accuracy by mentioned method for the real urban area.
In this study, effective spectral bands were selected using several methods to produce a wetland vegetation classification map with airborn MSS data. In the process of band selection, we have determined the optimal bands combinations using the both standards of maximizing Jeffries-Matusita (JM) distance and maximizing the classification accuracy of the test data. Then, in order to check how the classification method affect the band selection, we have compared the results of band selection by maximum likelihood method and minimum distance method. Band selection was conducted for 20 combinations of training and test data set, and the results were evaluated to rank the effectiveness of each band. The results shows that the firstly selected 3 bands are common for the case of JM distance and the classification accuracy of test data. These 3 bands are, 1: near infra red band (0.82-0.90μm) which is sensitive to the biomass of plants, 2: mid infra red band (1.52-1.72μm) which is sensitive to water content of the surface, 3: green band (0.57-0.59μm). This order of the first 3 selected bands was same for both maximum likelihood and minimum distance methods. However, after the 4th band selection, the selected bands have changed both by selection standards and classification methods. It was observed that the increasing rate of classification accuracy saturated in the case of using test data accuracy standard. Especially, by the maximum classification method, the classification accuracy have decreased after the 5 th band. Finally, the wetland vegetation map was produced using the 7 bands of the MSS data which have attained the maximum classification accuracy of the test data. While the conventional air-photograph interpretation method to produce wetland vegetation map have failed to classify the detailed vegetation types in the bog area, we have succeeded in classifing the bog vegetation into the classes of community dominated by Isotutuji (Ledum palustre), Suge (C. lyngbyei), Yoshi (Phragmites communis), Mizugoke (Sphagnaceae) and Hannoki (Alnus japonica).
The computer graphical method for analysis of remote sensing multi-spectral image data is proposed. Three-dimensional histogram is constructed in three-dimensional feature space from the image data and represented by three-dimensional computer graphics. On the histogram fine structure of distribution of data in the feature space can be observed. Therefore the categories of remote sensing multi-spectral image data can be identified easily on the computer screen. Each distribution of water, vegetation and soil etc. has its charasteristic features in the histogram. One of the most distinguished features is that distribution of data of each category is long and narrow, and the principle component axes of the categories concentrate on a point near the origin of the feature space.
A thermal inertia model based on the heat balance and one dimensional heat transfer of the earth's surface for the remote sensing was established by the analytic and mathematical solution techniques. The thermal inertia model for determining soil moisture was applied to airborne multispectral scanner (MSS) data to produce soil moisture map in the Kujukuri coastal plain. Our mapping procedure consists of the formulating the equation, the discriminant analysis of land cover by sample data from field observation and MSS data, and computer mapping technique. The equation for determining soil moisture was developed incorporating airborne MSS data and heat balance terms obtained at the micro-meteorological observation field at a time. One of these terms is the relative humidity of the soil surface. This parameter cannot be estimated by the remote sensing data itself because of unknown parameters in the estimate equation based on heat balance. Therefore, it was derived from the normalization of albedo based on the empirical model using the data obtained from the MSS measurement, the micro-meteorological measurements of the observation field of National Institute for Environmental Studies (NIES) and in the Kujukuri coastal plain. The other parameters were substituted by the heat balance terms such as diffusion velocity (coefficient), downward longwave radiation, global solar radiation at the observation field of micrometeorology. Thus obtined, our model for soil moisture enabled estimate of the soil moisture of bare soil and sparsely vegetated surface. The result of the experiment seems rather good. However, some parameters such as diffusion coefficeint, downward longwave radiation, ratio of vegetation cover in the model has to be examined in further analyses. Also, the model's application to the hilly country needs the information of the inclination of the ground surface and its processing system in this procedure.
The land coverage changes on the Mizushima and Kojima districts, Okayama Prefecture have been observed by using Landsat TM image data and digital map information to trace the process of industrialization and predict the capability of future development in these districts. The geographical changes in the past some decades have been found by overlaying old maps onto the TM image data. The environmental conditions in these coastal areas were discussed from a point of view of human activities. The industrialization of the Mizushima district was done from 1960's through 70's but little geographical change has been made in the past near decade. On the other hand, that of the Kojima district have not been made, but the wide-reclaimed salt pans are still left without using as industrial site. We compared two TM images (1984, 1990) of the Mizushima district with each other by using tasselled cap transformation (TCT) to examine the change of land coverage by time elapssing. It has been found that the Mizushima industrial complex is grown up as a fully developed area.
Measuring the velocity of river-surface fluid running is necessary to get the basic river information for the management of it. Therefore, it is important for this kind of research field-work to establish the simple and useful measuring method. At model channel, it has been used the various measuring methods and equipments, and also its result of the velocity distribution makes the hydraulics phenomenon clear with high precision. However, recently, it is reported from several field-works that those methods and devices are overly unfit for measurement at the real river. In fact, the typical measuring method needs many floats and a lot of time to get the velocity information, and also requires a great deal of labor and cost. This study proposes the newly method using the remotely sensed technique for many information of river velocity with widespread surface. The several conditions on this featuring measuring method as follows; (1) Video camera is used in the case that can be observed by the naked eyes. Using the video camera on the balloon, it is possible to provide the surface distribution of velocity with wide area. (2) Thermal infrared sensor is used in the case that can be not observed by the naked eyes. Using the thermal infrared sensor, it is possible to observe the surface velocity state with invisible information by the naked eyes. Furthermore, both method were conducted at the model channel and real river, and it is clearly found the advantage and the disadvantage for above mentioned new method.