iversity of 
iculture, is 
1 be used to 
uction as a 
1994). The 
nts which 
ig, a yield 
stantaneous 
for salinity 
termination 
GPS / yield 
ler normal 
| C/A code 
GPSCard™ 
to provide 
sts using a 
the code 
1992). Once 
analysed, 
quent years 
> used as a 
ol for this 
e variety of 
1s layers of 
surfacing 
the DGPS 
crop yields 
'onclusions 
h salinity 
ons can be 
when it is 
can also be 
h as aerial 
Current 
linity, soil 
llected for 
oject is to 
p for any 
t season of 
indication 
this year, 
in treated 
is used to 
t shown in 
The figure 
geneous in 
ds allows 
blem areas 
ing results 
rner of the 
| by high 
ns a false 
  
  
  
= DE LE 
er adea. 
en SET 
£0) Alaa 
Va EIS 
Fig. 5: Yield Map (3rd Dimension is Yield) 
Hussar Site Subset 
Once all of the necessary data has been sorted into 
their respective layers, relationships and effects 
between each layer can be determined. This 
information can be used to optimize the field 
potential by treating the field based on the 
specific sub-class of different sections. A field may 
have several distinct classes of soil that should 
have different quantities and mixtures of 
fertilizers applied to it in order to get maximum 
yield. The variable rate application would also 
take into account other variables such as salinity, 
topography and history of previous crops and 
applications. 
The test field consists of gently rolling hills with a 
steep north facing hill in the middle. The GPS 
monitor station was installed near the field and 
the moving platform was operating within a few 
km from the reference station. The crop was 
harvested on September 20 and 21 1993. On 
November 9, soil samples were taken at various 
locations for cross-referencing the first dataset. 
Table 5: RMS Agreement Between Carrier Phase 
Smoothing and OTF Solutions at Crossover Points 
  
  
  
  
RMS of Differences 
Date East North | Height 
(m) (m) (m) 
Sept. 21 0.14 0.21 0.51 
Nov. 9 0.14 0.26 0.66 
  
  
  
  
  
  
The positioning accuracy requirements in this 
project are 0.5 m horizontally and 1.0 m vertically. 
Two techniques were used to reduce the DGPS data, 
namely a carrier-smoothing of the code technique, 
and a on-the-fly ambiguity resolution procedure 
(OTF). The achievable height accuracy was 
verified by comparing the estimated positions 
using these two techniques, i.e. the OTF solution 
was used as a reference trajectory. Table 5 shows 
169 
the results for each of the two test days and 
illustrates that the positioning requirements are 
being met using the current configuration. 
4.4 Highway Inventory 
Over the past several years, developments into 
the combination of land-based precise positioning 
with imagery to form a 'highway inventory 
system’ have been ongoing. Early systems, such as 
the Alberta Mobile Highway Inventory System 
(MHIS), used dead-reckoning sensors such as gyros, 
accelerometers and compasses along with video 
imagery to continuously record visual information 
which was tagged with a positional locator. 
Information on pavement condition as well as the 
highway infrastructure (e.g. signs, guardrails) 
could then be used by engineers and planners for in- 
house reconnaissance which would minimize field 
inspections. 
One of the major drawbacks of the system was the 
error from the positioning module caused by drifts 
in the sensor output. To circumvent this problem, 
the sensors required frequent calibration which 
greatly affected productivity. In 1988, the 
feasibility of using GPS combined with an inertial 
navigation system (INS) to improve the 
positioning accuracy to the level of 0.2-0.3 m was 
demonstrated (Schwarz et al. ,1990). 
     
   
CCD camera 
  
  
  
  
  
GPS 
Positio 
  
  
  
Ati de 
  
  
  
  
  
  
  
  
Fig.6: VISAT Concept (Schwarz et al.,1993)