, 56(11), 
cum. 1995. 
Sensing for 
rthern 
tal Remote 
isconsin-Madison. 
D. Severinghaus. 
pical Effects of 
Arm 
Army 
lo. N-142). 
  
82. The 
ce in the Landsat 
of an Australian 
otogrammetric 
, 48(11), 
Digital Image 
J: Prentice Hall. 
1994. Remote 
| (Third ed.). New 
caused Wind 
New Mexico. In 
nghampton 
of the Basin and 
Survey 
ada Agricultural 
da. 
90. Tracked 
t the Pinyon 
0. Journal of 
243. 
lams, & A.R. 
sserts: I. À 
from 
nsing of 
oil Erosion and 
at Three State 
ifornia. 
er Science, 10(2), 
  
ESTIMATION OF NITROGEN LOADING IN SURFACE WATER 
USING A GIS AND THE EXPORT COEFFICIENT MODEL 
N. M. Mattikalli 
NASA/ Goddard Space Flight Center, Laboratory for Hydrospheric Processes 
Hydrological Sciences Branch, Code 974, Greenbelt, MD 20771, USA 
B. J. Devereux, and K. S. Richards 
University of Cambridge, Department of Geography 
Downing Place, Cambridge CB2 3EN, England 
ABSTRACT 
A methodology is presented to estimate surface water quality as indexed by nitrogen loading 
using the export coefficient model operating within a vector-based GIS. The model requires information 
on land-use and fertilizer application as the input parameters, and it employs export coefficients to 
estimate solute loading. Historical land-use data, for the River Glen watershed in eastern England, are 
derived from a variety of sources including remotely sensed satellite images, and are georeferenced in a 
GIS. Export coefficients are compiled from literature. Special GIS routines have been developed to 
carryout the model computations automatically using the data stored in vector format. Results 
demonstrate that the model predicts nitrogen loading with an acceptable degree of accuracy. Analysis 
shows an overall increase of nitrogen loading from 1931-84, after which it remains steady. The model 
and the GIS methodology have a significant potential to hindcast and forecast changes in water quality 
in response to historical/ proposed land-use changes. 
1. INTRODUCTION 
The quality of surface water has 
deteriorated in many countries in the past few 
decades. Meybeck et al. (1989) note that whilst 
the global median nitrate concentration in surface 
waters excluding Europe is 0.25 mg NO3-N per 
litre, the European median level is 4.5 mg NO3-N 
per litre, which was attributed to the greater 
anthropogenic loading of nitrogen on surface 
waters. In Europe, data show a distinct increase in 
nitrate concentrations in regions where there is a 
significant agricultural input (Meybeck et al., 
1989). Nitrate levels in the U.K. rivers have risen 
by 50-400% over the past twenty years (Royal 
Society, 1983). Analyses of water quality data for 
a number of rivers in the southern and eastern 
England have indicated significant and rapid 
increases in nitrate content, to levels exceeding 
the European Community/ World Health 
Organization (WHO) (Roberts and Marsh, 1987; 
Jose, 1989). 
In rural watersheds, water quality is 
mainly affected by soil erosion and the resulting 
suspended sediment load. However, leaching of 
fertilizers is a key process which derives solutes 
from agricultural fields. Cirello et al. (1979) 
related increases in nitrate concentrations to land- 
use intensification. The eastern parts of England 
have experienced major changes in agricultural 
activity, with increased arable farming in the past 
133 
few decades (Robinson and Armstrong, 1988). 
Such increases in arable agriculture have been 
reported in the MAFF statistics (MAFF, 1981), 
which demonstrate that the land-use in the eastern 
parts of England is dynamic. Parallel to these 
land-use changes, there have been increases in 
nitrogen fertilizer usage (Mattikalli, 1993). Table 
1 shows significant increases in the average rates 
of nitrogen fertilizer application on arable crops 
and grassland, between 1943-89. Analyses 
conducted by a number of authors (e.g. Kauppi, 
1984; Roberts and Marsh, 1987) indicate a strong 
correlation between nitrate content and the 
increased use of nitrogen fertilizers. 
Therefore, observed changes in land-use 
and fertilizer application rates may be expected to 
have caused changes in water quality. This paper 
presents a vector-based Geographical Information 
System (GIS) approach to estimate nitrogen 
loading in surface water using data concerning 
land-use and fertilizer application. GIS is a 
necessary tool to collect, store, manipulate, and to 
analyze large amounts land-use data derived from 
maps, air photographs and remotely-sensed 
satellite images. The GIS has been employed to 
integrate land-use data of various scales, and 
analyze historical land-use changes. Our 
approach employs the Export Coefficient Model 
recently pioneered by Burt and Johnes (1994).