Table 4. Surface derived parameters for Tambacounda of visible reflectance 
(/Visl, near-IR reflectance (ftnir), solar albedo (/> so ]), surface 
temperature (T e ) , and M1V1. Also included is the two week and four week 
antecedent precipitation for each date. Standard deviations are Riven 
within parentheses. 
Table 5. Surface derived parameters for Podor of visible reflectance 
(Pvie) » near IR reflectance (/? n ir) > solar albedo (/Jgol) , surface 
temperature (T s ) , and NDVI. Also included is the two week and four week 
antecedent precipitation for each date. Standard deviations are given 
within parentheses. 
Date 
0/3/81 
P 
(vis) 
9.0 (.4) 
P p 
(nir) (sol) 
39.5 (.6) 30.0 
mvi 
0.63 
T(s) 
305.3 (.8) 
Pp’t 
2-week 4-week 
180.0 410.0 
Date 
9/3/81 
p 
(vis) 
21.8 (.5) 
P 
(nir) 
51.2 (.7) 
P 
(sol) 
34.1 
fOVT 
0.40 
Pp’t 
T(s) 2-week 4—week 
308.4 (.5) 63.0 71.4 
3/22/82 
6/4/82 
7/8/82 
10/25/82 
28.4 (1.0) 
28.8 (.8) 
27.2 (.6) 
33.8 (.5) 
35.1 (1.5) 32.4 
36.3 (.9) 31.7 
38.2 (.6) 32.7 
49.8 (.5) 41.6 
0.11 
0.15 
0.17 
0.19 
321.5 (xx) 
328.5 (1.2) 
317.1 (.8) 
305.2 (2.5) 
0 
0 
0 
0 
0 
0 
0 
19.0 
3/22/82 
6/4/82 
7/8/82 
10/25/82 
49.5 (.6) 
41.3 (.4) 
46.4 (.2) 
37.3 (.5) 
62.4 
51.3 
61.0 
41.0 
(.9) 
(.5) 
(.3) 
(.8) 
56.9 
47.2 
54.3 
42.9 
0.11 
0.11 
0.13 
0.11 
321.7 (.8) 
326.9 (.3) 
319.8 (.6) 
323.7 (.4) 
0 
0 
l 
0 
0 
0 
1 
6 
3/31/83 
4/24/83 
8/17/83 
26.2 (.9) 
44.5 (.4) 
26.8 (1.9) 
36.6 (1.7) 31.5 
57.0 (.7) 51.5 
50.7 (2.3) 37.2 
0.17 
0.13 
0.31 
326.1 (1.0) 
314.9 (.7) 
307.2 (5.2) 
0 
0 
34.4 
0 
0 
34.4 
3/31/83 
4/24/83 
33.1 (.5) 
43.9 (.5) 
48.9 
53.4 
(.5) 
(.6) 
40.8 
49.7 
0.19 
0.09 
325.5 (.6) 
321.4 (.6) 
0 
0 
0 
0 
3/1/84 
3/9/84 
3/16/84 
6/30/84 
28.1 (.4) 
24.7 (.5) 
40.1 (.5) 
27.3 (2.3) 
40.3 (.5) 34.1 
44.0 (.9) 33.0 
53.5 (.6) 42.7 
50.5 (2.1) 37.5 
0.18 
0.14 
0.14 
0.30 
318.3 (1.6) 
322.9 (.6) 
324.1 (.5) 
304.9 (1.3) 
0 
0 
0 
26.8 
0 
0 
0 
52.5 
3/1/84 
3/9/84 
3/16/84 
6/30/84 
34.7 (.4) 
37.8 (1.0) 
40.0 (.6) 
41.2 (.6) 
59.4 
46.8 
55.0 
59.2 
(.5) 
(.9) 
(.8) 
(.6) 
45.9 
43.2 
44.1 
50.1 
0.26 
0.11 
0.16 
0.18 
313.5 (3.0) 
317.2 (0.9) 
323.1 (0.4) 
318.2 (.6) 
0 
0 
0 
36.0 
0 
0 
0 
36.0 
7/17/85 
8/3/85 
8/25/85 
9/10/85 
10/27/85 
27.9 (1.0) 
21.8 (.9) 
15.9 (1.0) 
19.3 (1.4) 
20.7 (.8) 
53.0 (0.8) 38.9 
54.0 (.9) 35.1 
48.7 (1.1) 29.2 
60.1 (1.3) 35.7 
45.3 (1.3) 31.2 
0.31 
0.42 
0.43 
0.51 
0.13 
299.7 (.6) 
295.5 (1.2) 
299.8 (0.6) 
304.9 (0.7) 
321.1 (0.8) 
18.8 
44.1 
26.0 
84.4 
0 
45.6 
52.1 
79.0 
105.3 
0 
7/17/85 
8/3/85 
8/25/85 
9/10/85 
10/27/85 
40.4 (.5) 
42.0 (1.0) 
42.5 (.9) 
38.0 (1.0) 
27.2 (.4) 
58.3 
61.0 
51.3 
61.2 
40.0 
(.5) 49.3 
(1.3) 51.3 
(1.1) 50.3 
(1.2) 48.7 
(.6) 35.9 
0.18 
0.18 
0.15 
0.23 
0.26 
316.1 (.3) 
305.2 (.4) 
314.7 (.1) 
320.3 (.7) 
317.8 (.5) 
7.0 
9.0 
0.8 
5.7 
7.0 
45.0 
47.0 
1.8 
6.5 
7.0 
can be 
seen that 
approximately 
50% of 
the 
cases 
relationship to 
the near- 
-IR 
reflectance. 
The 
had antecedent precipitation, and the remaining 
50% did not. 
Inspection of Tables 4 and 5 indicates that 
the average visible and near—IR reflectance, 
albedo, and ground temperature at Podor were 
larger than that at Tambacounda. This is likely- 
due to the greater density of vegetation at 
Tambacounda (as reflected by the larger value of 
the NDVI at Tambacounda) . During the dry period, 
the visible reflectance and the ground 
temperature are high, and the NDVI and near—XR 
reflectance are low, compared to the average 
values during the wet period (existing antecedent 
precipitation), which is indicative of a bright 
substrate having little or no green vegetation. 
The range for the visible and near-IR 
reflectance, the NDVI, and the ground temperature 
between the wet and dry periods for Tambacounda 
are larger than that at Podor, reflecting a 
greater variance in vegetation and antecedent 
precipitation from dry to wet periods at 
Tambacounda. The results at Tambacounda indicate 
changes in the solar albedo as a result of the 
vegetation changes were not substantial. This 
was due to the compensating effects of the 
reflectance of visible and near—IR bands. For 
low vegetative densities a larger proportion of 
the solar radiation is reflected by the bright 
substrate. In comparison, for higher vegetative 
densities there is a high reflectance in the 
near-IR from green leaf reflectance, compensating 
for a reduction in the reflection of radiation 
from the ground surface. At Podor where there 
was only sparse vegetation, there was only a 
small decrease in albedo with a change in 
vegetation amount. The small albedo change is 
likely frcm increased absorption of solar 
radiation in the visible and middle-IR and a 
reduced contribution frcm the reflecting 
substrate covered by the vegetative canopy. In 
either case, the presence or absence of 
vegetation did not substantially change the solar 
albedo frcm its overall mean value. This 
suggests that the underlying ground conditions or 
substrate has a compensating or a controlling 
influence on the albedo at these two locations, 
and unless major ground or canopy changes occur, 
changes in the albedo due to the presence or 
absence of vegetation does not occur. 
Trends in the behavior of the surface derived 
parameters are shown in the correlation parameter 
matrixes given in Tables 6 and 7. The NDVI is 
strongly and inversely correlated with the 
visible reflectance, but shows very little 
inverse relationship 
visible reflectance 
absorption in the 
between the NDVI and the 
is caused by the increased 
visible by the vegetation, 
having a higher chlorophyll pigment 
concentration, that is produced during the rainy 
season. In the dry season, the highly reflective 
substrate controls surface radiation processes. 
Since vegetation should be correlated with 
precipitation, the NDVI and the visible 
reflection should be reasonably correlated with 
precipitation also, for the reasons given above. 
However, the correlation between the near-infared 
reflection and precipitation was low. In the 
case of Tambacounda, if the dates with no 
precipitation were removed frcm the data set, the 
correlation increased markedly (r=.906), which 
indicated that the spectral variation frcm then 
non-green background was highly variable as 
indicated by the reflectance data 
(0.35<p n i r <0.60), contributing to the scatter and 
poor relationship to not only the precipitation 
but also the NDVI. Further, a bright substrate 
worsens the relation of NDVI to green leaf 
amounts due to a confusion between a highly 
reflective ground surface and a highly reflective 
vegetative canopy in the near-IR. The existence 
of a woody scrub also reduces the strength of the 
relation between NDVI and vegetation density 
because of increased absorption of solar 
radiation from the woody components. 
The NDVI was reasonably well correlated with 
the albedo at Podor, but not so well correlated 
with the albedo at Tambacounda. The relationship 
of ground temperature to the antecedent rainfall 
and the NDVI (vegetation) was, for the most part, 
moderate. However the data would suggest that 
the relationship should be more substantial, 
especially at Tambacounda. A natural logarithm 
transformation of the NDVI improved the linear 
correlation from r=-0.67 to r=-0.72 for 
Tambacounda. The correlation between ground 
temperature and solar albedo was very low, 
suggesting that the albedo has little control on 
the ground temperature. The ground temperatures 
more than likely are controlled by the presence 
or absence of clouds, vegetation, and moisture. 
5. SUMMARY 
NOAA AVHRR spectral data were converted to 
biophysical estimates of surface parameters. The 
estimated parameters were normalized difference 
vegetation index, solar albedo, spectral visible 
and near-IR reflectance, and ground temperature. 
261