were lowest for that eq. 3b. The underlying CanCov equations 
have been shown to be relatively insensitive to CH of more than 
7m (results not shown). Since the vCf estimates from eq. 2, 3a 
and b did not vary substantially the vCf estimates from eq. 2 are 
used in the following. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Figure 1. Time series of vCf predictions for four vegetation 
structural categories (VSC) (a to d) in the study area using 
MODIS FPAR to approximate the SurfCov subfactor (eq. 2; 
blue line), as the product of eq. 2 and the CanCov subfactor 
calculation from eq. 3a (green line) and eq. 3b (red line). Those 
three vCf predictions are named Scheme I, IIa, and IIb above. 
Dry seasons are symbolised by light yellow bars. The mean 
number of MODIS FPAR observations per observation date 
were 2548, 713, 4191, 1326, 380, and 129 for the ROIs of the 
VSC 1 to 6 respectively for the time period of 02/2000 to 
12/2006. 
3.2 High temporal frequency soil loss predictions 
Figure 2 shows a time series of high temporal remotely sensed 
vCf (MODIS FPAR vCf) (eq. 2), annual vCf (BGI vCf), 
predicted soil erosion (t) (MODIS FPAR E and BGI E) and 
daily rainfall (mm) for a randomly selected coordinate in the 
study. 
120 2500000 
100 
2000000 
80 
1500000 
60 
Erosion (t) 
Daily rainfall (mm) 
and vegetative cover factors (times 1000) 
1000000 
40 
500000 
20 
   
0 
20/02/2000 20/02/2001 20/02/2002 20/02/2003 20/02/2004 20/02/2005 20/02/2006 
  
| Serata MODIS FPAR vCf (times 1000) BGI_vCf (times 1000) «Daily rainfall + MODIS FPAR_E > BGI_E ] 
  
Figure 2. Time series of high temporal remotely sensed vCf 
(MODIS FPAR vCf) (eq. 2; green line), annual vCf (BGI_vCf, 
bright green line), predicted soil erosion (t) (MODIS FPAR E 
and BGI E; dotted brown and orange lines respectively) and 
daily rainfall (mm; blue line) for a randomly selected coordinate 
in the study. Daily rainfall data from SILO (Jeffrey, Carter et al. 
2001). 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
Soil loss predictions from the MODIS FPAR vCf (eq. 2) 
coincide well with those from the annual BGI vCf, while the 
soil loss predictions from eq. 2 are much higher. This is 
regarded as an indication of the dominance of the rainfall 
erosivity factor in the USLE. Generally, soil losses are mostly 
predicted to occur at the end of the dry seasons with high vCf 
values. The wet season 2004/2005 has the highest predicted soil 
losses from eq. 2. 
Whether the patterns of predicted soil loss for the study area 
have some agreement with events of streamflow and changed 
water quality at the outlet of the catchment (such as increased 
suspended solids) can be judged in comparison to an 
independent data set: in-stream measurements of cumulative 
daily total suspended solids (Figure 3). 
Plots of time series of average daily streamflow at Myuna 
station, predicted soil loss (FparErosion) (t), and TSS measures 
(mg/l) show some similarities between the onset of events and 
the shape of event trajectories but other inconsistencies are 
quite prominent (Figure 3). 
  
  
  
  
  
  
Cod 
300 
£3 
200 St 
£8 
xo sz 
35 
200 SE 
i 
100 
o 
PERI 43722 39/12 120% 36/01 *a 2302 9/03 23/03 £04 
1 ENSÉCUMTSS wat season 2003/2004 | —=mean sireamiiou jcamacs) —Fpartroson 2003/2008 : 
fab wed season 200072004 
140000 3500 
120000 
wo 
- = 
j^ i 
— 1500 a Ë 
2 20000 2 & 
x + 
: à 
t 20000 28 
+ & a 
300 
20000 
o o 
ve 15/12 20:12 12/01 26005 9/02 23/02 2/03 23/03 404 
EHO TES vet season 2004/2005 mm raa streamfiov, | cumecs; -——partrosiom 20/2008 — : 
14} wet season 2004 2DU5. 
23006 450 
70000 200 
£s 80000 334 gt 
= m9 
Ë 30000 8 
ë m0 28 
= o tes gi 
É 33000 on ii 
S 20000 100 CR 
13000 za 
8 a 
  
172 152 29/12 12/01 26/01 9/02 23/02 9/03 23,33 604 
  
gH cam 788 25 2005/100& 3 row toumecst Fzarfrosion 2005/31008 — : 
  
  
(ej wet season 2005 2000. 
Figure 3. Cumulative TSS (mg/l) (grey shade) and streamflow 
(cumecs/s) (blue line) at the Myuna station and predicted soil 
loss for the study area (FparErosion; in t*10 000 for scaling 
purposes) (brown line) for three wet seasons (a) 2003/2004, (b) 
2004/2005, and (c) 2005/2006. Predicted soil loss was 
calculated from the USLE using Scheme I in Figure lor eq. 2 
with MODIS FPAR as approximation for the GC. Data sources: 
TSS data by David Post, CSIRO; streamflow from QDERM. 
Note different scale for y-axes in a, b, and c. 
Discharge events and wet seasons show characteristically 
different trajectories of all three variables. The wet season 
2004/2005 seems to have the closest fit between all three 
trajectories in comparison to the other wet seasons. 
  
   
   
    
     
    
     
     
    
    
   
    
     
   
   
  
   
    
    
    
    
   
   
   
   
    
   
   
   
Similar 
predicti 
regressi 
96% of 
over th 
MODIS 
betwee 
Integra! 
is sugg 
tempor: 
predicti 
betweei 
tempor: 
FPAR 
(Figure 
  
from RUSLE using MODIS 
daily predicted soil loss (t) 
  
Figur 
made 
Zu 
In cor 
using | 
lies at 
taken 
depen 
daily | 
freque 
Limita 
the us 
and tl 
transp 
factor: 
overes 
ackno 
not th 
predic