1048 
Figure 2. Schematic diagram showing the kinds of data 
included in the modelling of supply and demand of 
fuel-wood in semi-arid Sudan. Source: Olsson et.al. 
(1985) 
7.4 Combined supply/demand modelling 
When quantifying e.g. vegetation amount by use of 
remote sensing, the result is normally presented as 
amount per area unit. When dealing with supply and 
demand of resources, the aim is to translate the area 
based measure into a per capita based measure, and 
then compare, in one way or another, the supply with 
the demand. I will briefly outline two different 
attempts to model supply and demand, both carried out 
in semi-arid parts of the Sudan, dealing with fuel- 
wood (Olsson et.al. 1985 and Olsson 1985) and millet 
production (Olsson 1985). 
In the case of wood resources, the modelling was 
based on Landsat MSS derived data on supply of woody 
biomass (tons/ha) and estimated, by interviews, 
annual consumption of wood, in Kordofan province of 
the Sudan. The modelling compared the demand for each 
village with the resources within a certain walking 
distance, a schematic description of the different 
kinds of data and analyses is presented in Figure 2. 
As a result the walking distance needed to satisfy 
the demand or the actual wood deficit was calculated 
for each village. The methodology is readily appli 
cable to studies on future development of the wood 
situation, through simulation of different variables, 
population increase, migration, afforestation, land 
management etc. 
In the case of millet production, the modelling was 
based on population distribution and statistical 
prediction of millet productivity. Spatial modelling 
yielded detailed information on land availability 
(Ha/capita). This was achieved by, following the rule 
that a piece of land belongs to the nearest settle 
ment, applying an algorithm delineating polygons 
around every settlement (Thiessen polygons). 
The number of hectars available per capita could 
then be calculated and represented as a raster image. 
The yield of millet was predicted from climtic data, 
represented as a trend surface, and combined with the 
land availability. The combination of productivity 
estimates and land availability made it possible to 
project the per capita production. When combined with 
more accurate measurements of actual crop return, 
based on remote sensing methodology, this approach 
will be a powerful instrument for studies of spatial 
distribution of food production. 
7.5 Soil erosion modelling and prediction 
Much research has been undertaken to establish 
mathematical relationships between various landscape 
features and rate of soil erosion. The most wellknown 
example is perhaps the Universal Soil Loss Equation 
(USLE) by Wishmeier & Smith (1960) , primarily 
developed for North American conditions, but 
subsequently adjusted for applications in other 
areas. The traditional methodology for mapping of 
erosion hazards is based on measurements in limited 
catchments, from which approximations have then been 
done. It is, however, today possible to apply a 
mathematical erosion model in a GIS, in order to 
carry out spatially full-covering mapping of poten 
tial soil erosion. The most important factors 
controlling erosion are topography, vegetation cover 
and land use. All these factors can with high 
accuracy and speed be treated using the combination 
of remote sensing and GIS. 
REFERENCES 
Bartlett D.S., R.W.Johnson, M.A.Hardisky & V.Klemas 
1986: Assessing impacts of off-nadir observation on 
remote sensing of vegetation: use of the Suits 
model. Int. J. Remote Sensing, vol. 7, pp 247-264. 
Bauer M.E., J.E.Cipra, P.E.Anuta & J.B.Etheridge 
1979: Identification and area estimation of 
agricultural crops by computer classification of 
Landsat MSS data. Remote Sensing of Environment 
Vol:8 pp.72-92 
Bush & Ulaby 1978: An evaluation of radar as a crop 
classifier. Remote Sensing of Environment Vol: 7, 
pp. 15-36 
Cihlar J., R.J.Brown & B.Guindon, 1986: Microwave 
remote sensing of agricultural crops in Canada. 
International Journal of Remote Sensing, Vol 7, pp 
195-212. 
Cohen J. 1960: A coefficient of agreementof nominal 
scales. Educational and Psychological Measurement 
Vol:20, pp. 37-46. (in Rosenfield & Fitzpatrick- 
Lins 1986) 
Cohen J. 1968: Weighted kappa: Nominal scale 
agreement with provision for scaled disagreement or 
partial credit. Psychological Bulletin Vol: 70, 
pp.213-220 (in Rosenfield & Fitzpatrick-Lins 1986) 
Coiner J.C. 1980: Using Landsat to monitor changes in 
vegetation cover induced by desertification 
processes. Proc. 14th International Symposium on 
Remote Sensing of Environment, pp.1341-1351. 
Colwell R.N. 1956: Determining the prevalence of 
certain cereal crop diseases by means of aerial 
photography. Hilgardia Vol: 26 pp. 223-286. (in 
Reeves 1975) 
Crist E.P. & R.J.Kauth 1986:The tasseled cap de 
mystified. Photogrammetric Engineering & Remote 
Sensing Vol:52, pp 81-86. 
Curran P.J, 1980: Multispectral remote sensing of 
vegetation amount. Progress in Physical Geography, 
Vol: 4, No: 3, pp. 315-341. 
Curran P.J, 
Longman, I 
Duggin M.J. 
Evaluation 
Applied Op 
Ewalt D. 19 
study Vol. 
Report 15 
Townshend 
Graetz R.D., 
1982: Th 
resource 
applicatic 
Australian 
Symposium 
Lands, Cai 
Guignard J. 
and operat 
Journal Vc 
Haralick R.M 
and spatia 
Sensing of 
Hay A.M. 19 
accuracy. 
Sensing Vc 
Hellden U. 
digital da 
an enviro 
University 
& Notiser- 
Hellden U., 
remotely 
the Sudan. 
Rome 2-6 D 
Hoffer R.M 
consider at 
techniques 
Swain & Da 
Holben B.N. 
assessment 
Photogramm 
46 pp. 651 
Hord R.M. 
criteria. 
Sensing Vo 
Hugli H. & 
reflectanc 
Engineerin 
Hutchinson 
Landsat a 
cation im 
Remote Sen 
Justice C.O. 
1985: Ana 
tion Using 
Journal of 
Kauth R.J. 
graphic 
developmen 
Landsat. 
Remote Sen 
Ch. 1103-1 
Kettig R.L. 
multispect