fertilizer application rates for 1931 are not 
available, data from 1943 were considered to be 
approximations for those existed in 1931. These 
attributes, along with polygon area parameters, 
were employed to derive the nutrient load from 
every discrete land-use polygon. Such loadings 
were accumulated over a sub-catchment to obtain 
total loading. This involves a large number of 
operations to be repeated for each sub-catchment 
and for all six land-use coverages. Therefore, 
special routines using the Arc Macro Language 
(AML) were developed to carryout these 
operations automatically. 
Validation of the model was carried out by 
comparing predicted nitrogen loadings with 
observed loadings. Since the export coefficient 
model predicts annual loading, it was necessary to 
extract both runoff and water quality data to 
compute annual observed loading. Both water 
quality and runoff data were examined and 
records of specific dates having both water 
quality and runoff data were extracted in order 
that nutrient ratings could be established. The 
TON concentrations and discharge data were 
available only for Burton Coggles and Holywell 
Brook during 1984-1987, and for Kates Bridge 
during 1984, 1987 and 1989. These data were 
analyzed to develop rating relationships between 
nitrogen loading and runoff, since daily records 
of the former were not available. Rating 
relationships developed for 1984, 1987 and 1989 
are presented in Table 3. It is clear from this table 
that nitrogen loading have strong positive 
relationships with discharge, and the high r2 
values indicate that loading can be calculated with 
a fair degree of accuracy. It was possible to 
employ these relationships, along with the daily 
runoff records, to compute the annual loading. 
This restricted the validation of the model to these 
specific sub-catchments for 1984, 1987 and 
1989. 
5. RESULTS AND DISCUSSION 
Nitrogen loadings in the surface water were 
calculated using the export coefficient model. 
Results were validated by comparing estimated 
loads with observed loads recorded at water 
quality sampling stations. Since the export 
coefficient model estimates annual loadings, the 
observed loads were also calculated for a period 
of one year. Daily discharge data for an entire 
year were used in conjunction with the rating 
relationships presented in Table 3 to calculate the 
observed yearly nitrogen load. Observed nitrogen 
loads were compared with those estimated by the 
export coefficient model, and Table 4 shows that 
they are of a similar order. 
Table 3. Relationships between nitrogen loading 
and discharge for Burton Coggles, Holywell 
Brook, and Kates Bridge. Models are of the form: 
Y = Bo + B1 X where Y is the calculated nitrogen 
load in kg/day; X is the observed discharge in 
m?/sec; Bo and B4 are the regression coefficients. 
  
Sampling Year Model 
station Bo B1 r2 
  
Burton Coggles 1984 - 34.724 1883.5 97.6 
1987 .- 1.534 1130.5 97.6 
Holywell Brook 1984 - 14.224 914.3 95.3 
1987 - 14.334 1059.8 94.3 
Kates Bridge 1984 -141.0+ 1267.0 93.8 
1987 -204.0+ 1152.0 85.3 
1989 -34.6+ 873.0 95.1 
  
Table 4. Comparison of nitrogen loading 
estimated by the export coefficient model with 
the observed loads. 
  
Sampling Year Nitrogen loss 
station (tonnes/year) 
Observed Predicted 
  
Burton Coggles 1984 106.35 128.33 
1987 83.76 124.36 
Holywell Brook 1984 84.36 92.39 
1987 91:61 84.87 
Kates Bridge 1984 612.87 533.68 
1987 336.73 470.50 
1989 862.07 603.80 
  
This demonstrated a generally successful 
application of the model, and the export 
coefficients used, as a basis to estimate nitrogen 
loading. However, some variations can be seen 
between the observed and estimated values which 
may be attributed to the use of rating 
relationships to calculate the observed loads and 
the application of export coefficients selected 
from the literature (Mattikalli et al., 19953). 
Nevertheless, the results demonstrate that the 
model estimates nitrogen loading with a fair 
degree of accuracy. 
The predicted losses of nitrogen for the 
entire study area for all six years are presented in 
figure 2. It is clear from this figure that nitrogen 
loading has increased during 1931-89. Nitrogen 
loading increased from 67.80 tonnes/year in 
1931 to 603.80 tonnes/year in 1989. This 
increase i 
due to th 
land and 
nitrogen 
nitrogen 
attributec 
fertilizer 
compared 
388888 
Nitrogen loading, tonnes/year 
3 
1 
  
© 
Figure 2. 
the entire 
In 1 
loading i 
fertilizer : 
of nitroge 
combined 
fertilizer 
identify t 
and fertili 
achieved 
different 
historical 
The result 
were use: 
loading (f 
Using 
100 4 
80 4 
Nitrogen loading, tonnes/year