¹Graduate Student, Department of Civil Engineering, Tohoku University, Sendai, Japan

²Associate professor, Department of Civil Engineering, Tohoku University, Japan

³Professor, Department of Civil Engineering, Tohoku University, Japan

Aoba-yama, 06, Sendai, 980-8579, Japan

Tel.: +81-22-217-7515, Facsimile +81-22-217-7453

E-mail: asaoka@kaigan.civil.tohoku.ac.jp

Abstract: NOAA/AVHRR satellite data were utilized for temporal and spatial analysis of snow covered area. The relationship between snow depth and altitude distribution was investigated to estimate snow distribution in East Japan (250,000km²). At first, the gradients of snow depth to elevation (snow increasing coefficient) were estimated at AMeDAS observation points during 1998 winter season. Secondly, the distribution of retreat rates of snow line was evaluated with the images of snow covered area, which are estimated with NOAA/AVHRR satellite data. Thirdly, the relationship between the snow increasing coefficient and the retreat rate was investigated.

As the results, the followings are obtained; the movement of snow line in snowmelt season is understood, and snow increasing coefficient is inverse proportion to retreat rate of snow line. It was possibly estimate snow depth in retreat region during snowmelt season.

 

Keywords: altitude distribution of snow depth, snow increasing coefficient, snow line, snow retreat rate

1  INTRODUCTION

The winter monsoon from Japan Sea brings heavy snow to East Japan, which depth is sometimes over 5m. Therefore, east Japan is famous for one of the heaviest snow area all over the world. Heavy snow causes traffic jam, cutting electric wires, prevention of commercial transportation and disasters such as house collapsing, snow avalanche and snowmelt floods. On the other hand, snowfall in winter is stored in the mountains and becomes useful as water resources for electric power, irrigation and water supply. Therefore, estimating snow volume in a wide area is important for water resources planning. However, the characteristics of snow distribution have not been understood clearly because of difficulty of snow survey at high elevation region and difference of annual amount of snowfall. Many researchers reported that snow depth is related to elevation. Koike^[1] estimated snow water equivalence from satellite images. Lu^[2] retrieved regional character of snow distribution of snow distribution with snowfall simulation. These study areas are limited only one watersheds. Kazama et al^[3] estimated snow distribution in the Tohoku district, Japan. But this research supposed snow depth is proportion to the difference of elevation from snow line and detailed character of snow distribution cannot be estimated. Matsuyama^[4] arranged snow surveys conducted in mountain regions. Every study periods of these snow surveys vary according to snow survey.

Fig.1  Snow map, April 21,1999

The purpose of this study is to understand the relationships between snow depth and altitude distribution in wide area (250,000km²).

2  STUDY AREA AND DATASET

East Japan (250,000km²) was selected as a study area. It is located between 35°N and 45°N. This area includes Hokkaido Island, Tohoku district, Hokuriku district and the Japanese Alps which are heavy snow area in Japan. Winter monsoon brings high humidity from the Japan Sea to the mountains and results heavy snowfall which depth exceeds over 5m. Almost all regions are steep and covered with the natural dense forest. Study period is from January to April 1999.

There are about 200 points AMeDAS (Automated Meteorological Data Acquisition System, Japan Meteorological Agency) snow depth observation stations at study area. Those data were used as ground truth data of snow depth. Geographical Survey Institute, Japan publishes various kinds of digital maps. Ks-110 among them is used as DEM (digital elevation data).

One Snow covered area (Snow map) images per one month were prepared. These are composed with NOAA/AVHRR satellite data and multi-spectral analysis. Figure1 shows an example of snow map, April 21, 1999.

3  THE EVALUATION OF ALTITUDE DISTRIBUTION OF SNOW DEPTH

As the index to evaluate the altitude distribution of snow depth, snow increasing coefficient is utilized. This is the gradients of snow depth to elevation and is presented following,

                                                    (1)

where, a is snow increasing coefficient (cm/m), SD is snow depth (cm) at AMeDAS snow observation point and ΔH is the difference of elevation from snow line.

Figure2 show the distributions of snow increasing coefficients at AMeDAS snow depth observation points. The values of snow increasing coefficients at the region near the snow line are very high. Specially, these are very high in the low elevation area where snow depth is high because the difference of elevation from snow line is very small. On the other hand, the values of coefficient are low in high mountain area. Main Island of Japan can divided into Japan Sea side from Ohu mountains range and Pacific Ocean side from this. Winter monsoon from Japan Sea brings to Japan Sea side. Low elevation area in Japan Sea side is covered with snow. The difference of elevation from snow line is large and snow increasing coefficient is low. Comprising this area, in Pacific Ocean side is not affected by winter monsoon.

Low elevation area is not covered with snow, therefore the difference of elevation from snow line is small and snow increasing coefficient is high.

Fig.2  Distribution of snow increasing coefficient

Figure3 shows temporal variation of snow increasing coefficients under 5.0. The abscissa indicates cumulative day from 1^st December, 1998. In many regions, maximum coefficients are appeared at February or March. This reason is that maximum snow depth is appeared at February and March. The changes of snow increasing coefficient of which maximum is under 1.0 are stable. These are in high elevation area of which the condition of snowfall is stable. On the contrary, the others is not stable. These are the one in low elevation region. These depend on condition of snowfall.

Fig.3  Temporal variation of snow increasing coefficient

Snow line retreats from low elevation area to high elevation area in snow melt season. The retreat of snow line influences the altitude distribution of snow depth. Therefore, the distribution of retreat rates of snow line is estimated from February to March and from March to April 1999. The method of estimating retreat rate is following. Comparing snow maps of current month and previous month, if the area, where previous month is covered with snow, is not covered with snow, the distance from nearest snow line at previous month, in short retreat rate as one month, is calculated. Figure4 shows the images of distribution of retreat rate at March and February 1999. Noticeable retreats of snow line are appeared at March. These also make sure snow map on February and March. This snow season is little snowfall. Kazama⁵⁾ reported that low elevation areas are not covered with snow in little snowfall year, on the contrary, these areas are covered with snow in heavy snow year. This snow season is little snowfall and low elevation areas are not so covered with snow, therefore snow lines don’t retreat. The areas where snow line moves are valley and near the coastline. Retreat of snow line appeared on April. Specially, southwest at Hokkaido Island and Kitakami mountain range is notable. These area are influenced by not winter monsoon from Japan Sea but the movement of low pressure, therefore there are little snowfall and low temperature and amount of solar radiation in these area and snow line moves at high rate on April.

Fig.4  Snow retreat rate in snowmelt season

The distribution of snow increasing coefficient and the distribution of snow retreat rate is estimated in previous sections. The movement of snow line in snowmelt season is evaluated with these databases. Figure5 shows the relationship between snow retreat rate and snow increasing coefficient at previous month. Many plots concentrate in the neighborhood of origin of coordinates. These snow depths are about zero, therefore snow retreat rate and snow increasing coefficient is small. Omitting these plots, snow retreat rate is fast in region the gradient of snow depth to elevation is low and contrary snow retreat rate is slow in region it is high, in other words, the relationship between snow retreat and snow increasing coefficient is inverse proportion. The reason is that if the gradient of snow depth to elevation is high, the snow volume in the range which snow line retreat is fast and snow retreat rate is slow. On the contrary, snow volume is low and snow retreat rate is fast, if it is low.

In Figure5, symbol A indicates the value at Iwamizawa, (43.395°N, elevation 6m) and B does at Ishikari, (43.21°N, elevation 31 m). These points are at Ishikari plain in Hokkaido Island. Snow retreat rate is fast and snow increasing coefficient is low because snow depth is low but climate is warm. C does at Aterazawa, (38.36°N, elevation 140 m) and D does at Inawashiro, (37.56°N, elevation 527 m). These points are in Yamagata valley and Inawashiro valley. Snow retreat rate is slow and snow increasing coefficient is high because snow depth is high but climate is cold.

Fig.5  Relationship between snow increasing coefficient and snow retreat rate

4  SUMMARY

The summary of this report is followings. At first, the altitude distribution of snow depth at about 200 AMeDAS snow depth point is evaluated. As results, the gradient of snow depth to elevation (snow increasing coefficient) is lower in Japan Sea side than in Pacific Ocean side. In addition, it is higher on low elevation area than in high elevation area. Secondly, snow retreat rate is estimated and relationship with snow increasing coefficient is evaluated. After all, it is possibly that snow increasing coefficient is inversely proportional to snow retreat rate in snowmelt season.

This work was conducted with the support of the Sendai Construction Office of the Ministry of Construction. It is gratefully acknowledged that Dr. J. Kudo, Computer center in Tohoku University helped us data collection of NOAA/AVHRR.

[1]  T. Koike, Y. Takahasi and S. Yosino (1985), Estimation of Basin-wide Snow Water Equivalent Using Snow-covered Area, Journal of Japan Hydraulic, coastal and Environmental Engineering, No.357/Ⅱ-3, pp.159-165.

[2]  M. Lu, N. Hayakawa and Y. Yoshioka (1997), Approach to retrieve Regional Snowfall Distribution Using Remotely Sensed Snow Covered Area and Distribution Hydrological Model, Annual Journal of Hydraulic Engineering, Vol. 41, pp. 239-244.

[3]  S. Kazama and M. Sawamoto (1996), Estimation of the snow depth distribution and snow water resources in wide area, Int. Conference. on Water Resources. & Environment Research. Towards the 21st Century, Vol.1, pp.659-666.

[4]  H. Matsuyama (1998), A Review on the Snow Surveys Conducted in Mountains Regions in Japan to Determine Distribution Factors, J. Japan Soc. Hydrology. & Water Resources, Vol. 11, No. 2, pp. 164-174.

[5]  S. Kazama and M. Sawamoto (1994), On Water Balance in a Basin with Heavy Snow Area –Case Study in the Taki Dam Basin, the Tadami River-, Annual Journal of Hydraulic Engineering, Vol. 38, pp. 113-118.