246 
From the plot of near infrared to red (Thematic 
Mapper band 4 to 3; wavelength respectively 776 to 
905 nm and 624 to 693 nm) it can be concluded that 
most units with the exception of a small group show 
spectral curves representative for soils with a low 
to absent vegetation cover. Comparing feature space 
plots of the first four bands (band 1 452-518 nm 
and band 2 529 to 610 nm) in the visible and near 
infrared part, three main types can be 
discriminated: 
1. a small group with vegetation (palm trees in 
the oasis) with a low reflectance for blue, green 
and red and intermediate reflectance in the near 
infrared and a high 4/3 ratio. 
2. a group of (almost) bare soils changing to a 
range of intermediate to very low reflectance 
values in band 4. 
3. a group of bare soils, having an intermediate 
to very high reflectance in all 4 bands. 
Introducing also band 7 (2097 to 2347 nm) the 
above mentioned groups can be characterised as 
follows: 
1. the reflectance of the oasis is very low in 
band 7, as can be expected for vegetation. 
2. this group has like in band 4 an intermediate 
to very low reflectance. 
3. two extremes develop within this group: 
1. one extreme has both in band 7 and in band 
4 the intermediate to very high reflectance. 
2. the other extreme has in band 7 low values 
in stead of intermediate values and intermediate 
values in stead of very high values. 
Band 5 (1568 to 1784 nm) can be considered as in 
termediate between the first four bands and band 7. 
The following characteristics can be given of the 
3 groups. The oases, covered with palm trees in the 
upper layer, are group 1. Group 2 consists out of 
units of different parts of the playa and playa 
border zone. The relatively dry footslope and dune 
areas are the third group. The large decrease of 
reflectance in band 7 in 3.2 is due to the presence 
of gypsiferous sand or gypsum crust at the surface. 
Differences in reflectance within the footslope and 
playa group are due to different amounts of stones, 
crust, gypsiferous or non-gypsiferous sand, 
vegetation, surface roughness, exposition, 
moisture, salt content and mineralogy of the top 
surface. 
3. CAUSES OF DIFFERENCE BETWEEN JANUARY AND MAY 
DATA 
The differences between the two days can be divided 
into differences of the surface itself and those 
due to the effect of sun elevation at Landsat 
overpasstime and turbidity of the atmosphere. 
3.1 Difference of the surface 
The most probable differences of the surface may be 
caused by the effect of rain, wind and vegetation. 
Due to the effect of rain the surface layer may 
be moist. An increase in moisture content in the 
surface layer leads to a decrease in reflectance. 
The cause and relation between reflectance and 
moisture content are described a.o. by Angstrom 
(1925), Bowers and Hanks (1965) and Planet (1970). 
The decrease in reflectance will last for a longer 
time if the groundwater table is close to the 
surface. Therefore, if the moisture effect of rain 
is visible; this will be in the playa and playa 
border zone. Field measurements showed that this 
moisture effect in the top surface layer has 
disappeared in the footslope area within a day. 
A second effect of rainfall is the redistribution 
of £>alt. Salt efflorescence occurs in the playa and 
playa border zone. It was observed in May 1985 that 
even in the relatively high parts of the playa 
border zone (up to 1 meter) this phenomenon occured 
and lasted at least for a week after rainfall. 
A third effect to be mentioned is the occurence 
of slaking after rainfall. This effect is also 
important in the relatively dry areas even in the 
dunes and may last for long periods. 
Erosion and sedimentation is a fourth 
possibility, which may affect reflectance if 
rainfall occurs. 
Due to the wind in sand areas the roughness of 
the surface may change as a function of the 
windvelocity before and at the time of Landsat 
overpass. Moreover dunes may migrate. Since sand 
grains or sheets are often an important part of the 
surface, the effect on reflectance may be present. 
The vegetation in May in this area will in 
general be more green and healthy than in January. 
Moreover annuals may be present on places, where in 
January no plants are present. In order to estimate 
the effect of vegetation the spectral reflectance 
curves in relation with the soil surface have to be 
taken into account. 
All effects mentioned above may differ in place 
not only due to the characteristics of the surface 
but also due to irregular distribution of rain, 
runoff and wind within the area. 
3.2 Other differences 
The solar zenith angle at the acquisition dates 
differs significantly (29 degrees in May and 60 
degrees in January). Moreover differences in 
turbidity of the atmosphere at the two days may 
exist. The turbidity and solar zenith angle 
influence the relation between reflectance at the 
top of the atmosphere and the surface reflectance. 
Different approaches and formulas relating these 
two types of reflectance can be found (Otterman and 
Fraser 1976, Nack and Curran 1978, Chen and Ohring 
1984, Koepke et al 1985). The relationship between 
surface albedo (a ) and planetary albedo (ax; 
albedo at the top of the atmosphere) can^ be 
approximated according to Koepke et al (1985) by: 
a=a+ba (1) 
where a^= path raâiance 
b = two way transmittance 
A nomogram is presented by Koepke et al (1985) to 
find values for a and b at different solar zenith 
angles and turbidity factors. In table 1 a and b 
values are presented for 29 and 60 degrees solar 
zenith angles with albedo expressed in reflectance 
percentage. As can be expected path radiance will 
increase with increasing solar zenith angle and 
tubidity of the atmosphere, while the two-way 
transmittance will decrease with these two factors. 
It may be clear that transmittance is wavelength 
dependent. Hence the correction for albedo bands 
cannot be applied directly on the different bands. 
Table 1. Path radiance (a) and two-way 
transmittance (b) values for different turbidity 
factors (T at 550 nm) and solar zenith angles of 29 
and 60 degrees (after nomogram presented by Koepke 
et al. 1985). 
zenith angle 29° 60° 
a 
b 
a 
b 
3.5 
.79 
5.7 
.73 
3.6 
.77 
5.9 
.71 
5.7 
.62 
10.3 
.53 
A soil 
Therefor 
function 
direct 
(Makarov 
1981). 
(1977) 
reflecta 
converte 
implies 
zenith i 
natural 
surface 
would gi 
sun dep 
degrees 
probably 
waveleng 
With 1 
shadow v 
areas w: 
higher a 
will lea 
will be 
solar ze 
January, 
will re. 
Finally 
parts oi 
function 
time of 
4. C0NVE 
The ref] 
the atmc 
shown, t 
at the 
waveleng 
even in 
correcte 
for both 
Since 
calculat 
These mi 
lower li 
are equa 
the feat 
is the 
constant 
All ap 
- Upper 
necessar: 
lower or 
due to 1 
surfaces 
advantagi 
that t 
minimize* 
- The as 
space ma; 
- Ref lei 
constant 
In ord 
ref lectai 
comparab' 
Ultimai 
which a 
this wii; 
the foot 
been se 
reflectai 
this obj 
bare pla; 
way in a 
approach 
May have 
0.882 an 
with Ja