サテライトリモートセンシングによる土壌含水比分布の地図化の試み 九十九里地盤沈下地域を例として

抄録

Using satellite remote sensing data, the system for predicting soil moisture contents and mapping of its distribution was investigated in this paper.
It was necessary to observe environments and to collect the data in the field for the numerical analysis of remote sensing technique. Since a collection period of ground truth data (landuse and soil samples) was desirable to coinside with satellite (LANDSAT-3) overpass (Jan. 1980), the collection was carried out for 4 days (Jan. 17_??_20, 1980) in the Kujukuri coastal plain.
Procedures of this system were based on multivariate statistical analysis and computer mapping techniques. The data were processed by means of a multiple discriminant analysis for classification of landuse and a multiple regression analysis for estimation of soil moisture contents, where the soil moisture contents is a dependent variable and the remote sensing parameters are independent variables.
Above mentioned processes and automatical mapping with LANDSAT data brought the following conclusions.
1. Some problems (e. g. performance of classification and efficiency of modelling) to be solved in the processing of the future data were indicated.
2. Distribution of soil moisture contents showed NNE-SSW linear pattern subjected to the developments of the micro-topographies such as emerged bar, sand spit and dunes in the Kujukuri coastal plain.
3. The region such as Hasunuma, Naruto, Ohamishirasato-Honno and Shirako were under higher soil moisture contents. In particular, at Shirako and its vicinity on the lower Nabaki which was major subsiding area, submerged paddy field and exposed ground were observed.
4. Soil moisture contents reduced from Shirako and its vicinity to Kujukuri.
5. On the other hand, the soil with higher moisture contents were originated with shade behind the valley wall or restricted drainage with barrier, whichever artificial or not, in the valley plain and the rear of the coastal plain.
6. It is expected that these soil moisture contents models had to be revised in the future analysis and satellite remote sensing might offer practical potential for the thematic mapping and the monitoring of the environments.