Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
Sugar beet biomass estimation using spectral data 
derived from colour infrared slides 
Robert R.De Wulf & Roland E.Goossens 
State University Gent, Belgium 
ABSTRACT: Sugar beet spectral reflectance data were extracted from multitemporal ground - 
based colour infrared photography. Measured agronomic parameters include green leaf area 
index (GLAI), cover percentage, fresh root weight and sugar content. 
Statistical relationships between spectral reflectance and biomass were determined for 
one growing season and applied for prediction of biomass in a subsequent growing season. 
1 INTRODUCTION 
Crop-weather relationships numerically expressed in 
prediction models are being applied for yield fore 
casting in a number of countries. 
However no crop growth model has been able to 
perfectly simulate the synergistic effects of 
environmental conditions, meteorological factors and 
agro-cultural factors (Colwell 1977 et al). 
Moreover , a number of abiotic and biotic hazards 
which cannot be predicted by agromet models or 
visually assessed in the field, can have conside 
rable negative effects on crop yield. 
Remote sensing may serve to complement traditional 
methods for crop growth monitoring and production fo 
recasting on local and on global scale. 
Ground-based radiometry has established empirical 
relations between spectral and agronomic parameters 
for a range of economically important crops. Research 
efforts have established the basics for operational 
large scale applications as global wheat yield fore 
casting in the LACIE project (MacDonald and Hall 
1978) and biomass estimation in arid zones using 
multitemporal NOAA AVHRR data (Tucker et al 1985). 
Crop production can be viewed as a product of 
acreage and yield (Colwell et al 1977), A colour 
infrared (CIR) photograph allows extraction of both 
factors. Using visual interpretation methods or 
interactive image classification preceded by 
digitizing, parcels of land covered by a particular 
crop can be measured fairly precisely. This aspect 
is not elaborated here. 
The numerical assessment of the yield factor requires 
a statistically-based relationship between remotely 
sensed data and agronomic features. This relationship 
has been studied for sugar beets during the 1984 and 
1985 growing seasons. 
The choice of a 35mm colour infrared emulsion is 
tied to its planned use in an automatic SLR camera on 
board a remotely piloted aircraft (RPA). 
This delta-winged RPA, custom-built for the Centre 
for Remote Sensing of Vegetation (CEVA) , made 
successful test flights in March 1986 and is 
scheduled to be fully operational for low altitude 
crop monitoring in the second half of the 1986 
growing season. 
The procedure for extraction of spectral parameters 
from 35mm CIR slides, in a preparatory phase taken 
from a tripod, became a priority research issue at 
CEVA. 
The selection of sugar beet as crop of interest is 
determined by its importance in Belgian agriculture. 
In 1984 sugar beets covered 117000 ha or 8.4 % of the 
country's agricultural surface (N.I.S. 1985). 
2 REMOTE SENSING OF SUGAR BEET BIOMASS : PRINCIPLES 
According to Thorne (1971) arable crops can be 
regarded as machines for converting C02 and water 
into carbohydrate, using the sun's energy. 
Photosynthate can be transferred to physiological 
sinks which consitute the harvestable parts of a 
number of economically important crops : potato 
tubers, wheat grain kernels and sugar beet roots. 
White sugar is the most important part of the sugar 
beet from the economical point of view. However in 
terms of biomass quantity (15%) it is clearly sur 
passed by a sizeable amount of processed by-products 
including pulp and brown molasse, both used as 
livestock feed. 
Remote sensing of sugar beet biomass requires 
efficient and economical data aquisition and know 
ledge of yield-related crucial moments on the 
physiological time scale. 
Sugar beet productivity is strongly favoured by long 
growing seasons provided that meteorological and soil 
conditions are not limiting (Analogides 1979). For’ 
sugar beets (Steven et al 1982), potatoes (Allen and) 
Scott 1980) as for cereals ( Gossse et al 1986) it 
has been shown that total dry matter accumulation is 
proportional to intercepted solar radiation. 
The amount of photosynthetically active biomass, 
usually expressed as Green Leaf Area Index (GLAI) and 
its time of duration (Green Leaf Area Duration,GLAD ) 
have become critical parameters in the majority of 
plant growth analysis. The LAD concept is illustrated 
by Evans (1972), who compared dry matter at harvest 
with GLAD for potatoes and sugar beets, and found 
both factors to be 57% higher for sugar beets. 
From a prediction point of view radiation alone 
may not be a particular useful variable. In the case 
of sugar beet, and specifically when its root sugar 
content is considered meteorological factors are 
known to play a decisive role which varies with the 
physiological time scale. 
Soil moisture deficit at the 4-leaf stage can have 
disastrous consequences for harvest, whereas too much 
rains in July and August favour leaf development at 
the expense of root weight and sugar content (Leblon 
1983). Cool days and frosty nights in late September 
and October check leaf and root growth but increase 
photosynthetic activity and sugar storage (Whyte 
1960). 
Scammel's general yield prediction model for 
Belgium is based on sugar beet physiology and 
meteorological parameters but is said to perform 
poorly on local basis (Leblon 1983). This is mainly 
due to varying climatological conditions and soil