Development of a Model for Predicting Water melon yield in a Semi-Arid region of Nigeria

A mathematical model for predicting the yield of watermelon using soil and plant properties was developed from a two year field experiment which was conducted in 2013 and 2014 to investigate the effect of different tillage methods on the physical properties of sandy loam soil, growth and yield of water melon (Citrullus vulgaris) on a sandy-loam soil in a semi-arid environment. The Tillage treatments were disc ploughing plus disc harrowing (DP+DH) as conventional tillage (CT), double disc ploughing (DDP), double disc harrowing (DDH), disc ploughing (DP) and disc harrowing (DH) as minimum tillage (MT) and zero tillage (ZT) as direct drilling method (control). The model was developed from soil properties like bulk density, moisture content and penetration resistance. The plant properties were leave area index, plant length, number of fruits per plant, fruit length, fruit weight, fruit diameter, and yield per hectare. The assumptions were derived from empirical results. It is a linear function with some non linear variables. It is therefore reasonable to take the model as non linear function that was solved using non linear regression procedure using SPSS Version 16 computer program. The model was validated by comparing the predicted yields by the model with the actual obtained yields from the experiments. This procedure minimized the error due to deviations and therefore rendered the solution more accurate as could be seen from the improved R2.The model gave a good agreement of calculated watermelon yield with linear relationship of (r2=0.98).