－ 53 －

GIS－理論と応用Theory and Applications of GIS, 2009, Vol. 17, No.1, pp.53-62

【研究・技術ノート】

Using GIS for assessing stream water chemistryin a forested watershed

Akiyuki KAWASAKI, Reiji FUJIMAKI, Nobuhiro KANEKO and Satoru SADOHARA

Abstract: Systematic survey preparation and data analysis are required for conducting efficient

research on forested mountains because investigating many streams in such an environment

is difficult. A methodology of GIS utilization for surveying and analyzing nitrogen leaching in

headwaters was demonstrated in this paper. Firstly, support maps for the Tanzawa Mountains,

Southern Kanto District, Japan, were created for effective data collection by field surveys.

Then, contributions of catchment properties to N-leaching were investigated by spatial analysis

using fine-scale terrain data. Finally, N-leaching prediction maps were created for decision-

support.

Keywords: forest ecosystem, catchment, water sampling point, nitrogen

1 . Introduction1 -1 Background

Nitrogen (N) plays a key role in regulating

primary production in ecosystems, especially in land-

based systems. Increased human socioeconomic

activity has changed nitrogen circulation drastically

on a global level. Increased reactive nitrogen from

humans through synthetic fer tilizers, industrial

use, and fossil fuel consumption has been exerting

strong influences on the ecosystem. High levels

of anthropogenic nitrogen, coming primarily from

atmospheric deposition, are recognized as a crucial

environmental issue in natural ecosystems on a

par with global warming and acid rain (Aber et al.,

1998; Galloway et al., 2003). Problems associated

with high nitrogen levels were also featured in the

United Nations’ Millennium Ecosystem Assessment

(Millennium Ecosystem Assessment, 2005). A

major symptom caused by increased nitrogen input

to forested catchments is increased nitrate (NO3-)

concentrations in stream waters, which can cause

eutrophication and acidification in freshwater and

coastal waters, which deteriorates water quality while

threatening human health (Vitousek et al., 1997).

There is an increasing need to understand stream

water quality.

In general, researching streams in an entire

watershed is difficult due to its large area. Therefore,

systematic monitoring location planning and

analysis are crucial for effective research. Recent

developments in information technology such as

Geographic Information Systems (GIS) allow the

rapid arrangement of a spatial data infrastructure and

comprehensive analyses of various kinds of spatial

data. Consequently, GIS of fer promising tools for

planning and analyzing the study of stream water

quality in large-scale forested watersheds. This paper

provides a case study on applying GIS to streamwater

research in the Tanzawa Mountains of northwestern

Kanagawa Prefecture, central Japan. The Tanzawa

Mountains are an impor tant area because most

川崎：〒150 -8925 東京都渋谷区神宮前5丁目53 -70 国際連合大学サステイナビリティと平和研究所 United Nations University, Institute of Sustainability and Peace 5 -53 -70 , Jingumae, Shibuya-ku, Tokyo 150 -8925 , Japan E-mail：kawasaki@unu.edu

－ 54 －

Kanagawa Prefecture residents depend on the area

as a source of water. For that reason, deterioration

of water quality in this area is a critical issue for

Kanagawa water resources. This area, located

in Japan’s southern Kanto district, is most likely

affected by anthropogenic disturbances. Okochi and

Igawa (2001) reported high nitrate concentrations in

a stream in the Tanzawa Mountains, but their study

was limited to only one stream; so water chemistry

patterns on a broader scale within the Tanzawa

Mountains remains unclear. Additionally, nitrogen

leaching from forest ecosystems varies significantly

because of many factors like differences in geology

(Willard et al., 2005), hydrology (Shibata et al., 2001),

watershed topography (Creed and Band, 1998), and

tree species composition (Lovett et al., 2000).

1 -2 ObjectiveThe research team for this study consists of

geologists, pedologists, and vegetation scientists.

The team has been tr ying to explain nitrogen

leaching mechanisms from forest ecosystems and

proposed a future strategy of forest management

(Kaneko and Fujimaki, 2006; Sakai et al., 2007). The

research project involved conducting a general field

survey to assess the current nitrogen leaching from

forest ecosystems in the Tanzawa Mountains. The

study described in this paper specifically examines

the utilization of GIS for survey and analysis. This

paper proposes and describes implementation of

spatial information utilization for promoting efficient

nitrogen leaching assessments in forested mountain

headwaters.

G I S h a v e b e e n u s e d t o i n v e s t i g a t e t h e

characteristics of watersheds for many purposes;

for example, Luzio et al. (2004) developed a model

to assess agricultural non-point and point pollution

loading at the watershed scale; Downs and Priestnall

(1999) studied river channel adjustments; and

Kaur et al. (2004) analyzed the impact of land-

use planning on soil conser vation by using a

watershed-scale hydrologic model. Specifically as a

preceding study, Jarvie et al. (2002) investigated the

relationship between river chemistry and watershed

characteristics, including landuse, topography and

geology, using GIS and statistical analysis for the

Humber catchment in England. In addition, nitrogen

leaching factors at the catchment scale in forested

mountains in Japan were analyzed using a 50 meter

mesh digital elevation model (DEM) (Ito et al., 2004;

Ogawa et al., 2006). However the priority and process

of selecting water sampling points remain unclear,

and spatial patterns of nitrogen loss in streams and

the cause of the pattern were analyzed within the

watershed delineated from these sampling points.

A methodology for designing sampling points are

necessary to consider forest management in the

Tanzawa Mountains, which consists of precipitous

terrain combined with a variety of geology and

vegetation over a broad area.

In this study, sur vey suppor ting maps were

prepared for effective data collection in the field by

considering some working hypotheses. Then, factors

contributing to water chemistry were analyzed using

fine-scale spatial data in an individual catchment.

Finally a “nitrogen leaching prediction map” was

created for decision support concerning the type and

location of actions to properly manage future forest

ecosystems in the Tanzawa Mountains.

Fig. 1 Research site

－ 55 －

2 . Developing working hypotheses and allocating sampling points

2 -1 Study areaThe Tanzawa Mountains are located in the

northwestern part of Kanagawa Prefecture, in an

area extending 50 km east and west and 30 km north

and south. The mountain range is located where the

Philippine, Eurasian and North American continental

plates collide, and it was formed by the collision of

the Honshu Arc and the Izu Arc. Japan’s highest

rates of orogenic uplift and denudation by the Izu-Arc

are still observed here (Arima and Kaneko, 2006)

and the topography is composed of steep slopes.

The impact of atmospheric pollution is apparent

in the Tanzawa Mountains. The distance from the

center of the Tokyo metropolitan area, where Japan’s

socio-economical activities are concentrated, is

approximately 50 km. In the mountains, 11.3 kg-N/

ha/year of nitrogen deposition was measured in

Ooyama and 13.3 kg-N/ha/year was measured in

Fudagake (Asou et al., 2001).

Six river systems in the mountains were studied

(Fig. 1): the Hayato River and Shimizu River flowing

to Miyagase Lake in the eastern par t; and the

Kurokura River, Nakagawa River, Oomatazawa River,

and Yozuku River flowing to Tanzawa Lake in the

western part. The Tanzawa region straddles both

Kanagawa and Yamanashi Prefectures, but only

the Kanagawa side was chosen due to personnel

limitations. The total research area was 21,000 ha:

95.7% of the area was covered by forest, 2.5% by waste

land, and 1.6% by water (National-Land Information

Office, 1997). The elevation of the area ranges from

290 to 1,673 m with an average of 851 m.

2 -2 Preparing datasets and characterizing catchments

To assess the influences of topography, geology

and vegetation on water quality, headwater outlets

discharged from various small catchments were

targeted as sampling points. The tentative size of

the catchments was limited to less than 80 ha so

that the results could be applied to a selection of

future experimental sites for long term monitoring.

For example, the Hubbard Brook Ecosystem Study

for a US site of the Long Term Ecological Research

(LTER), on ecosystem patterns and processes

in a watershed, selected nine watersheds of less

than 80 ha for observation (Hubbard Brook LTER,

2008). Fur ther consideration of the catchment

size is required if this unit is suitable for complex

topography in Japan, or for considering topography,

geology, vegetation and soil as a unit, and how

heterogeneous characteristics inside the catchment

is changed as the catchment size increases.

Several GIS datasets were collected and analyzed

to find appropriate sampling points (Table 1). First,

as an interpolation-conversion from elevation point

to raster dataset, catchments were delineated using

a DEM with a 12 m cell size using ESRI ArcGIS

Desktop 9.1 and Spatial Analysis. The DEM was

generated from vector elevation data called GISMAP

Terrain. Although the nominal resolution of the

data is 10 m, we produced the 12 m DEM due to the

Table 1 GIS data sources

Dataset name Format Scale Product name Data provider Category used in this study

Elevation Point 1:25,000 GISMAP Terrain(10 m mesh)

Hokkaido-chizu Co.Ltd. Elevation: 800 m> or 800 m<;Slope gradient: <30°, 30-35° or >35°

Bedrock geology Polygon 1:50,000National land

classification surveys

Ministry of Land,Infrastructure and Transpor t;

Ishikawa et al . (2005)

Tuff bedrock, Plutonic bedrock,Pyroclastic rock, Mudstone,

Pudding stone, or Loam

Vegetation Polygon 1:50,000 Vegetation inventory Kanagawa Pref. Natural forest, Plantation forestor Mixed vegetation

－ 56 －

parameter setting of the interpolation-conversion

module of the software. The Tanzawa Natural

Environment Information Station by Kanagawa

Prefecture also adopted a 12 m cell size for deriving

a DEM from the same data (Kanagawa Prefecture

Natural Environment Conservation Center, 2006).

With a flow accumulation grid delineated from

the DEM, streams were defined through a flow

accumulation value of 1,500 cells. A threshold flow

accumulation of 1,500 cells with a 12 m cell size

correspond to a drainage area of 216,000 m2 or

21.6 ha. The reasons for defining a threshold value

and verification of water flow status are described

in section 3-2. Next, stream junctions were treated

as catchment outlets. As a result, 139 headwater-

catchment sampling points were selected from the

study area.

The watershed characteristics of the selected

sampling points were classified according to the

following properties (Fig. 2).

Elevation: Catchments were categorized into two

elevation groups: one with mean elevation greater

than 800 m, the other less than 800 m. At this latitude

(35°24' – 35°37'N), native vegetation growing at

elevations higher than 800 m is different from that

growing in lower areas. Cool-temperate natural mixed

forests with deciduous broadleaf trees (beech) and

conifers (fir and hemlock) prevail in upper elevations,

whereas temperate evergreen broadleaf forests

prevail in the lower areas (oak). Nowadays, the

latter areas are dominated by secondary vegetation

and conifer plantations such as cedar and Japanese

cypress (Sakai, 2006). These changes in vegetation

type, along with elevation, might alter the nitrogen

retention in catchment areas.

Slope: Slope gradient was assumed to influence

water transfer speeds on the sur face and inside

the soil, as well as to affect soil depth. Mean slope

gradient was used to categorize catchments into

three types: <30°, 30–35°, and >35°, since thick loamy

volcanic soils tend to cover moderate slopes less than

30°, and granitic tonalite soils cover a broad range of

slopes but the soil depth tends to decline inversely

with the gradient in the Tanzawa Mountains (Kaneko

et al., 2007; Sakai et al., 2007). Granitic tonalite soils

have higher productive capacity of nitrates per

organic substance than volcanic ash soils, because

organic matter in granitic tonalite soils tends to have

higher mineral content. Slopes around 30°and higher

were considered to have high nitrate production rates

because such slopes are mainly covered by granitic

tonalite soils.

Geology: The eastern part of the research site is

covered by tuff bedrock developed in the Miocene

Fig. 2 GIS datasets for categorizing the watershed characteristics of selected sampling points

Elevation Slope

Geology Vegetation Candidate points

Candidate point

Watershed

Elevation Slope

Geology Vegetation Candidate points

Candidate point

Watershed

－ 57 －

(Ter tiar y) period, and the western par t on the

other side is mainly covered by plutonic bedrock

(granitic tonalite). This geological distribution was

assumed to have a significant influence on stream

water chemistry and ecosystem material flow by

affecting vegetation and chemical weathering of the

soil. Furthermore, such mechanisms might influence

the buffering capacity against nitrogen deposition in

forest water systems.

Vegetation: The catchments are dominated by

degraded trees due to a lack of forest management,

and this may strongly influence nitrate leaching.

Consequently, the catchments were categorized as

“natural forest, plantation forest or mixed vegetation”

by calculating the dominant vegetation (>70% of

the area) of each catchment. Catchments were

categorized as having mixed vegetation when either

natural or plantation forest covered less than 70%

of the total area. The value of 70% was a tentative

threshold.

3 . Implementation of water sampling3 -1 Survey outline

A general survey by six teams with 25 members

was conducted to assess nitrogen leaching in the

Tanzawa Mountains headwaters on July 13th and

14th, 2006. Each team was supplied with two types of

the river system map highlighting sampling points

and catchment properties (Figures 3 and 4), 1:25,000

Geographical Survey Institute (GSI) topographical

maps, longitude and latitude of sampling points, and

GPS units. Some members were familiar with the

study site, while some were not. Team members

were assigned to collect samples from watersheds

that have various characteristics of elevation, slope,

geology and vegetation in a balanced manner using

survey support maps.

Despite initially aiming to acquire 100 samples

out of 139 candidate points in Fig. 2, the survey

ultimately yielded 51 samples of water, soil and

leaves (bamboo grass and Polygonum cuspidatum),

inc lud ing severa l unp lanned po ints due to

topographic difficulty in accessing several planned

points, time restrictions, and heavy rainfall on July

14th (Figures 5 and 6). The points not surveyed were

either too far from main road, too dangerous for

data collection (rough landscape and swollenness of

stream), or unsuitable in terms of the balance of the

watershed characteristics. For supporting a more

systematic sur vey, focusing on priority areas by

preliminarily considering catchment properties and

displaying optimal routes might be useful. Achieving

over half of the intended water samples in this short

period under imperfect conditions was considered

successful.

Additionally, the GPS units including Trimble

Fig. 4 Survey support map – example 2Fig. 3 Survey support map – example 1

－ 58 －

Pathfinder Pro XR and ProXH provided accurate

location information (accuracy of less than 1 m using

real time differential correction), which allowed the

effective surveying and collection of sample points.

However, GPS signals were often insufficient due

to the forested mountainous terrain, so the GSI

topographical maps were used to supplement the

searching tasks.

3 -2 Verifying the status of headwater flow extracted from the DEM

No standard processes exist regarding what

methodologies and parameters to use for extracting

streams using a DEM, and it is possible to delineate

a wide range of stream features using GIS depending

on the data used, analysis resolution, extraction

parameters and methodologies, etc (Kawasaki, 2006;

Maidment, 2002). In this study, sampling points

were designed from catchments and stream data

delineated from the DEM, as described in section

2-2. In general, water chemistry analysis gets more

complicated with larger catchment areas, and it is

ideal to collect samples from the upper reaches of

a catchment to assess the effects of topographical

features, geology, and vegetation on water chemistry.

Therefore, it was necessar y to capture accurate

locations of headwaters not described even in the GSI

topographical map. For that reason, headwaters that

regularly maintain at least a minimum level of flow

despite seasonal and conditional variability had to be

extracted using GIS.

Accordingly, a flow accumulation value of 1,500

cells was defined as the stream delineation threshold

based on the knowledge of experts familiar with a

part of the study area (Nishizawa catchment, 306

ha). The experts compared various kinds of stream

data extracted using a range of thresholds, and

determined the most suitable threshold to describe

actual water flow including small streams.

During the survey, stream flow at the planned

points was verified including locations at which actual

water samplings were not implemented. Inventory

forms were distributed to each team and the status of

water flow was identified as one of the four categories

in Table 2 by visual observation. The table shows

that actual water flow in various statuses was verified

at most locations checked during the survey. Total

rainfall was recorded as 2 mm for one week prior to

the survey (July 7th - 13th), and 94 mm for the past

two weeks (June 30th - July 13th) at the neighboring

Tanzawa Lake observation site (Japan Meteorological

Agency, 2006). The results shown in Table 2 might

be biased since the results were limited to only those

locations accessed, and some field survey members

might have selected easier locations for obtaining

water.

4 . Analysis and modeling of nitrogen leaching

4 -1 Chemical analysis and mapping the resultStream water, soil, and leaf samples were used to

assess water chemistry such as pH balance, water

temperature, nitrate (NO3-), NH4

+, total nitrogen (TN),

and dissolved organic carbon (DOC). The result

was outlined in this paper and utilized to create the

nitrogen leaching prediction map in section 4-2.

The mean concentration of TN of all stream water

was 0.74 mg-N/L, ranging 0.18 mg/L to 1.30 mg/

L with over 95% containing nitrate (Figures 5 and

6). The factors shown in Fig. 2 were almost equally

distributed because of the use of the survey support

maps (Figures 3 and 4). Nitrate concentrations in

headwaters of the eastern Tanzawa Mountains tend

to be higher than those of western areas (Fig. 6),

which corresponds to dif ferences in geology: the

eastern part consists of tuff bedrock, whereas the

western part consists of plutonic bedrock (Fig. 2).

Chemical analysis methods and equipments, as well

as the results and mechanisms of Nitrogen leaching

in the Tanzawa Mountains are described in Fujimaki

－ 59 －

et al. (2008)

4 -2 Predicting model of nitrogen leachingA general linear model was used to predict the

nitrogen concentrations in streams using topographic

characteristics in the catchment areas. The following

parameters were assessed.

a ) Distance from the Yokohama city center (km)

b ) Distance from the Sagami Bay coast line (km)

c ) Total catchment area (ha)

d ) Mean catchment elevation (m)

e ) Catchment elevation standard deviation (m)

f ) Catchment elevation range (maximum −

minimum, m)

g ) Catchment mean slope (deg.)

h ) Catchment slope gradient standard deviation

(deg.)

i ) Proportion of steep slopes (>30°) area to total

catchment area (%)

j ) Most frequently observed slope aspect category

k ) Mean of the cosine-transformed value of the

slope aspect azimuth: The cosine of the slope

aspect azimuth represents the degree of north

facing, ranging from -1 to 1.

l ) Mean catchment Laplacian value

m) Standard deviation of catchment Laplacian

value

From the geological map and the vegetation

inventor y, each catchment was categorized as

follows:

n ) Most frequently observed bedrock geology in

three categories (plutonic, tuff, and volcanic

tuff)

o ) Dominant vegetation (>70% of area) in three

categories (natural forest, conifer plantation,

and mixed vegetation);

Model selection was per formed according to

Akaike’s information criterion (AIC), a statistical

reference for determining the suitability of statistical

models among several options (Venables and Ripley,

1999). Lower AIC values indicate a better model

based on the lowest squares of residual sum but

with a penalty for the greater number of parameters.

The software program R 2.6.0 (R Development Core

Team, 2007) was used for the statistical analysis. The

selected model was;

TN = 2.5913 - 0.2389 k - 0.1541 h - 0.0071 a ,

where TN is total nitrogen concentration (mg-N/L)

in stream waters, k is mean of the cosine-transformed

value of the slope aspect azimuth, h is catchment

slope gradient standard deviation (deg.), and a is

distance from Yokohama city center (km) (R2 = 0.343,

AIC = -5.19, F = 8.17, p < 0.001).

TN was negatively related to k, showing that

stream TN tends to be high when the proportion of

south-facing slopes increases. This is because soils

Fig. 5 Total nitrogen [mg-N/L]

Fig. 6 Nitrate (NO3--N [mg-N/L])

Table 2 Stream flow status (n=61)Stream flow status Number of points (ratio)

)%58( 25wolf retaw tnadnubA)%2( 1wolf ot eganaM)%8( 5retaw pmuS)%5( 3wolf retaw oN

－ 60 －

in south-facing slopes tend to be warmer, probably

promoting nitrogen mineralization and nitrification.

TN was also negatively correlated to h, indicating that

nitrogen is less likely to leach with more complex

terrain, but this exact mechanism requires further

investigation. Additionally, the Tanzawa Mountains,

directly facing the Kanto Plane where Yokohama

is located and where industry and agriculture are

highly developed, receive reduced nitrogen loads

that are inversely propor tional to a. Additional

interpretation of individual parameters is described

in Fujimaki et al. (2008). AIC analysis also yielded the

following model:

TN = 2.1577 - 0.1954 k - 0.1473 h(R2 = 0.313, AIC = -4.91, F = 10.91, p < 0.001).

This second model had fewer parameters than

the first model with little dif ference in AIC value,

although the coefficient of determination was less.

Fig. 7 shows the results of the first model fitted

to a topographical map. Continuing sur veys and

analyses with additional consideration are required

for improving the scientific credibility of this study

because this map was extrapolated from spatial

nitrogen leaching patterns observed on July 13th

to 14th, 2006. This map, however, can contribute

to effective and advanced surveys in the future by

illustrating where to collect water in an ef ficient

manner.

5 . ConclusionsThis study used GIS for supporting sampling point

selection to analyze nitrogen leaching in headwaters

and explained the available methodologies. The

following results were obtained:

1. Sur vey suppor t maps were created to locate

sampling points and field data collection.

Headwaters in small catchments with dif ferent

properties were successfully extracted.

2. Spatial analysis based on fine-scale terrain data

was used to investigate the effects of catchment

topography like slope gradient and slope aspect

on nitrogen leaching.

3. A nitrogen leaching prediction map was created

for the entire Tanzawa Mountains based on data

from 51 sampling points. Estimation of the status

of the broad range of mountains and considering

further desirable countermeasures are possible,

by expanding the analysis results in small

catchments to the broad range of all mountains

using GIS, where capturing aerial ecosystem data

was difficult.

Further debate is required on how to use GIS for

creating a material flow model to predict nitrogen

status in mountain headwaters and estimating

effective countermeasures using scenario analysis.

AcknowledgmentsWe are especially grateful to the valuable support

and suggestions from the referees and editorial

committee of GISA, and also Prof. Arima, Prof.

Nakamura, Prof. Ohno, Prof. Ito, Prof. Ishikawa,

Dr. Kaneko and Dr. Sakai at Yokohama National

University. This study was financially supported by

the 21st Century COE Program of the Japan Society

for the Promotion of Sciences, the scientific research

fund of the Ministry of Education, Culture, Sports,

Science and Technology (No. 18710035), and the

fund for Enhancement of Education and Research in

Yokohama National University.

Fig. 7 N-leaching prediction map (TN)

－ 61 －

ReferencesAber, J.D., McDowell, W., Nadelhof fer, K., Magill, A.,

Berntson, G., Kamakea, M., McNulty, S., Currie, W.,

Rustad, L., Fernandez, I. ( 1998) Nitrogen saturation in

temperate forest ecosystems. BioScience, 48, 921 -933 .

Arima, M., Kaneko, Y. (2006) Geoscientific characteristics

and ecological risk assessment in the Tanzawa

Mountains. Proceedings of the fifth symposium on

“Environmental Risk Management for Bio/Eco-

systems”, Yokohama National University Graduate

School of Environment and Information Sciences, 7 -15

(in Japanese).

Asou, T., Takeda, M., Aihara, K. ( 2001 ) Observational

study of acid rain and air pollution in the western part

of Mt. Tanzawa ( 1995 - 2000 ). Bulletin of Kanagawa

Environmental Research Center, 24, 82 -89 .

Creed, I., Band, L. ( 1998 ) Expor t of nitrogen from

catchments within a temperate forest: evidence for a

unifying mechanism regulated by variable source area

dynamics. Water Resour. Res., 34, 3105 -3120 .

Downs, P.W., Priestnall, G. ( 1999 ) System design for

catchment-scale approaches to studying river

channel adjustments using a GIS. Int. J. Geographical

Information Science, 13 ( 3 ), 247 -266 .

Fujimaki, R., Kawasaki, A., Fujii, Y., Kaneko, N. ( 2008 )

The influence on the stream N concentration in the

Tanzawa Mountains, Southern Kanto District, Japan.

DOI 10 . 1007 /s 10310 - 008 - 0089 - 1 , Journal of Forest

Research.

Galloway, J.N., Aber, J.D., Erisman, J.W., Seitzinger, S.P.,

Howarth, R.W., Cowling, E.B., Cosby, B.J. (2003) The

nitrogen cascade. BioScience, 53, 341 -356 .

Hubbard Brook LTER (2008) Watershed.

　　<http://www.hubbardbrook.org/watersheds/

watersheds.htm>

Ishikawa, M., Komito, Y., Nishimoto, T. ( 2005 ) Slope

failures triggered by the 1923 Kanto Earthquake.

GISA Conf. Proc., 14, 199 -222 (in Japanese).

Ito, E., Ono, K., Imaya, A., Kanomata, H. ( 2004 ) Regional

estimation of potential nitrogen loss from forest soils

at the catchment scale. Jpn. J. For. Environment, 46,

77 -84 (in Japanese with English summary).

Japan Meteorological Agency ( 2006 ) Meteorological

records archives. < http://www.data.jma.go.jp/obd/

stats/etrn/index.php >

Jarvie, H.P., Oguchi, T., Neal, C. ( 2002 ) Exploring the

linkages between river water chemistry and watershed

characteristics using GIS-based catchment and locality

analysis. Reg. Environ. Change, 3, 36 -50 .

Kanagawa Prefecture Natural Environment Conservation

Center ( 2006 ) Natural environment information:

GIS data list.<http://www.e-tanzawa.jp/contents/

gis_02sizen.html>

Kaneko, N., Fujimaki, R. ( 2006 ) Ecosystem management

in the Tanzawa Mountains and the relationship with

mega city Yokohama as its river system. Gekkan

Jokasou (Monthly Sanitation) , 376 , 12 - 16 (in

Japanese).

Kaneko, Y., Arima, M., Sato R., Ono, Y., Iwagaki, T.,

Kawasaki, A. ( 2007 ) Distribution of tonalitic soil

structure and geochemistry in the western part of

Tanzawa Mountains: Construction of basic data for

long-term monitoring study. Journal of the Geological

Society of Japan, 113, 611 - 627 (in Japanese with

English summary).

Kaur, R., Srivastava, R., Betne, R., Mishra, K., Dutta, D.

( 2004 ) Integration of linear programming and a

watershed-scale hydrologic model for proposing an

optimized land-use plan and assessing its impact

on soil conservation: A case study of the Nagwan

watershed in the Hazaribagh district of Jharkhand,

India. Int. J. Geographical Information Science, 18 ( 1 ),

73 -98 .

Kawasaki, A. ( 2006 ) Delineating watersheds from DEM

data. In Kawasaki, A. and Yoshida, S. eds. Introduction

to ArcGIS Part 2 - Step up for GIS Utilization, Tokyo:

Kokon Shoin Ltd., 41 -64 (in Japanese).

Lovett, G.M., Weathers, K.C., Sobczak, W. (2000) Nitrogen

saturation and retention in forested watersheds of the

Catskill Mountains, NY. Ecol. Appl., 10, 73 -84 .

Luzio, M.D., Srinivasan, R., Arnold, J.G. ( 2004 ) A GIS-

coupled hydrological model system for the watershed

assessment of agricultural nonpoint and point sources

of pollution. Transactions in GIS, 8 ( 1 ), 113 -136 .

Maidment, D.R. (2002) Arc Hydro: GIS for Water Resources,

Redlands, USA: ESRI Press.

Millennium Ecosystem Assessment (2005) Ecosystems and

Hunan Well-being: Synthesis. Washington DC: Island

Press.

National-Land Information Office ( 1997 ) Digital national

land information.

　　< http://w3 land.mlit.go.jp/ksj2/index.html >.

－ 62 －

Ogawa, A., Shibata, H., Suzuki, K., Mitchell, M.J., Ikegami,

Y. (2006) Relationship of topography to surface water

chemistr y with par ticular focus on nitrogen and

organic carbon solutes within a forested watershed in

Hokkaido, Japan. Hydrol. Process, 20, 251 -265 .

Okochi, H., Igawa, M. ( 2001 ) Elevational patterns of acid

deposition into a forest and nitrogen saturation on Mt.

Oyama, Japan. Water Air Soil Poll., 130, 1091 -1096 .

R Development Core Team ( 2007 ) R: A language and

environment for statistical computing. R Foundation for

Statistical Computing, Vienna, Austria. ISBN 3-900051-

07 -0 , URL http://www.R-project.org.

Sakai, A. ( 2006 ) Status of natural environmental and

ecological risk in the Tanzawa Mountains, Gekkan

Jokasou (Monthly Sanitation), 376 , 5 - 11 (in Japanese).

Sakai, A., Fujimaki, R., Kawasaki, A., Tomita, M., Kaneko,

N. ( 2007 ) Topographical distribution of forest trees

in a watershed on the actively denuding mountains,

Tanzawa. Report on “Environmental Risk Management

for Bio/Eco-systems”, Yokohama National University

Graduate School of Environment and Information

Sciences, 71 -77 (in Japanese with English summary).

Shibata, H., Kuraji, K., Toda, H., Sasa, K. ( 2001 ) Regional

comparison of nitrogen export to Japanese forest

streams. Scientific World, 1, 572 -580 .

Venables WN, Ripley BD (1999 ) Modern Applied Statistics

with S-Plus. Springer-Verlag, New York.

Vitousek, P.M., Aber, J.D., Howarth, R.W., Likens, G.E.,

Matson, P.A., Schindler, D.W., Schlesinger, W.H.,

Tilman, G.D. ( 1997 ) Human alteration of the global

nitrogen cycle: Causes and consequences. Ecol. Appl.,

7, 737 -750 .

Williard, K.W.J., Dewalle, D.R., Edwards, P.J. ( 2005 )

Influence of bedrock geology and tree species

composition on stream nitrate concentrations in mid-

appalachian forested watersheds. Water Air Soil Poll.,

160, 55 -76 .

（2008年4月15日原稿受理，2009年3月13日採用決定，2009年4月27日デジタルライブラリ掲載）