AGROMETEOROLOGICAL MODELING FOR COTTON YIELD ESTIMATION Leonidas Toulios 1 , Alexia Tournaviti 2 , Georgia Zerva 1 and Theodore Karacostas 2 1 National Agricultural Research Foundation (NAGREF), Larissa, Greece 2 Aristotle University of Thessaloniki, Department of Meteorology and Climatology, Greece Abstract Early estimates of agricultural production are of great importance for agricultural policy and trade. Cotton is very important crop worldwide and especially in Greece, and there is a great need for good estimates of yield and total biomass production. Agrometeorological models are used worldwide with, unfortunately, in most cases, only local value, and generalization is one of the research issues today. Relatively simple models using the photosynthetically active range of the solar radiation are available from the literature, although they have not been verified extensively. These models require information about the climatic efficiency and the radiation use efficiency of the crop. Assuming the above, field studies with cotton were conducted to verify these models under Greek conditions. Based on the linear connection between the interception efficiency of photosynthetically active radiation (PAR) and the normalized difference vegetation index (NDVI), spectral reflectance was measured by means of hand-held radiometer during the whole cultivation season, in a three year experiment in the main cotton zone in Greece, expressed as NDVI. Total dry matter was measured in a weekly basis and it was compared to the findings of the model. The results show that the estimated dry matter and yield agree well with the field measurements and there is a potential in applying this approach in an operational basis with multitemporal remote sensing data. Limitations concerning the radiation use efficiency (phytomass production per unit of energy received) and the spectral data collection are also discussed. Introduction During the last decades, the study of the weather conditions and their connection to the plant growth and the crop yield has been very important in agricultural research. In the same time, remote sensing technology has been developed and its products have many applications in agriculture, especially in crop identification, monitoring of crop growth, land cover estimation and yield prediction. In most cases, the empirical methods for yield estimation include multiple regression models having the yield as dependent-predictor variable and various meteorological variables as independent-predictant variables. The last ones are selected among others using statistical and physical criteria (Delecolle et al., 1998; Tournaviti et al., 1998). Although this kind of models can often give satisfactory results, they have an intensively local character and it is difficult for them to be generalized. On the other hand, there are more complicated models simulating the physiological processes of the plant growth, taking into account any factor affecting the crop and containing even thousand of equations. In that case, the models are more probable to be used