233 
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
TURTLE and HARE, two detailed crop reflection models 
J.A.den Dulk 
Department of Theoretical Production Ecology, Agricultural University, Wageningen, Netherlands 
ABSTRACT: Yield prediction can be done by means of dynamic simulation, based on crop properties and global 
environmental conditions. The use of remote-sensing data, interpreted by means of a crop reflection model 
will improve the quality of the prediction. 
Two crop reflection models are presented: both models can be used to predict the reflection of a wide range 
of crops, even when crop properties vary with crop height. One of them (TURTLE) consumes much computertime 
and yields a complete flux profile within the canopy, whereas the other one (HARE) is much faster, but 
produces only the reflection properties of the complete crop. 
RESUME: Le rendement des cultures arables peut être prédit par simulation dynamique, basé à les 
caractéristiques de la culture et les conditions environnementales. L'usage des données obtenues par 
télédétection et analysées au moyen d'un modèle de rèflection du couvert végétale, peut amélorier la qualité 
de prédiction. 
Deux modèles de ce type sont ici présentés. Tous les deux peut être utilisé pour prédire la rèflection du 
couvert. L'un deux (TURTLE), exigeant un temps prolongé d'exécution sur ordinateur, produit un profil complet 
des flux à 1'intérieur du couvert végétal, alors que l'autre (HARE) exige un temps d'exécution beaucoup plus 
court, mais ne produit les caractéristiques de reflection que pour la culture entière. 
1 YIELD PREDICTION BY DYNAMIC SIMULATION 
At the Department of Theoretical Production Ecology 
of the Wageningen University, and in cooperation with 
the Centre for Agrobiological Research and other 
institutes, a number of dynamic simulation models for 
crop growth have been developed during the last 
decennia (Penning de Vries & van Laar, 1982). Some 
of these models are especially meant for detailed 
simulation of the processes in a single plant during 
one diurnal cycle, whereas other models are used to 
simulate growth and development of a complete plant 
stand during a complete growing season. 
If the modelled crop grows under well-known 
circumstances, as in a green house or in a region 
where weather conditions are fairly stable, then a 
good correspondence between the real crop and the 
model results is achieved. When however the 
knowledge about the actual growing conditions 
(weather, soil, nutrients) is poor, serious 
deviations between model results and measured data 
can occur. This means that the quality of crop yield 
simulation based on mean environmental conditions 
hardly exceeds the quality of a prediction that is 
based on intelligent extrapolation of the yields as 
obtained in the past under comparable circumstances. 
2 GENERAL CROP PRODUCTION THEORY 
Light interception is an important factor in biomass 
production rate, and on the other hand the coverage 
of a crop is directly related to the intercepted 
radiation. In the first phase of the growth of a 
crop, when the crop does not cover the soil totally, 
interception, and therefore also growth rate, 
increases with biomass production. This phase is 
called the phase of exponential growth. During this 
phase there is a strong positive feedback of biomass 
on biomass production rate. Therefore relatively 
small differences between model assumptions and 
reality will cause serious errors in the simulation. 
In the second phase of the growth, when the canopy 
becomes more and more closed, light interception does 
not increase further with biomass production. This 
phase is called the linear phase, where feedback is 
less important than in the first phase. But during 
this phase stress factors like water or nutrient 
shortage or pests and diseases may play their role In 
the growth of the crop, so also in this phase the 
simulation can show serious deviations between the 
modelled situation at harvest time and the real crop. 
3 ENHANCING THE QUALITY OF YIELD PREDICTION 
The purpose of this work was to enhance the quality 
of yield prediction obtained by dynamic simulation by 
including data gathered during the growing season 
from the crop itself. Combination of the actual 
weather data (in stead of mean climatological 
knowledge) with the deviations between the actual and 
the predicted state of the crop caused by stress 
factors will improve yield prediction. 
Data collection on the ground can be a problem. A 
good way to observe the crops themselves seems to be 
the use of remote sensing techniques and among these, 
multiband spectral scanning has been proven to be an 
applicable technique for the detection of crop 
properties (Bunnik, 1978). 
4 DYNAMIC GROWTH MODELS 
Before continuing on the topic of the interpretation 
of reflection data to adjust simulations carried out 
with dynamic growth models, we will give a brief 
explanation of the technique of dynamic simulation. 
A dynamic crop simulation model can be considered 
as a set of simultaneous ordinary or partial 
differential equations, in which each equation 
represents the in- or decrease in one quantity of 
interest, like leaf weight or dry matter stored in 
the kernels. The state of the system at the start of 
the simulation (for instance the beginning of the 
growing season) defines the initial values, whereas 
the environmental conditions such as rainfall, 
radiation and temperature are the boundary conditions