Evaluation a Neural Net Work Program for Predicting Soil Moisture Retention Curve from in Situ Measurements Using Neutron Scattering Method

This work aims to evaluate Rosetta program for predicting soil moisture retention curve (SMRC) in clay loam soil. This evaluation was carried out in the farm of the Soil and Water Research Department of Nuclear Research Center, Atomic Energy Authority. This evaluation was done by two methods: The first called Tensiometic method (T.M), which depends on some in situ measurements using neutron probe for determining soil moisture contents and tensiometers for determining soil matric potential in the tensiometric range. The second one is van Genuchtens method (v.G), which depends on values of soil matric potential up to 15 bar and the next of soil moisture contents. As for Rosetta program, which needs to sand, silt, clay, soil bulk density and soil moisture contents at soil matric suction 0.33 and 15 bar. The output of these methods point to the prediction of SMRC from Rosetta program was less accurate than van Genuchten and tensiometric methods. Tensiometric method was the highest one for determining and predicting SMRC, The combination work between neutron scattering technique and tensiometers as in Tensiometric method helps to obtain in situ SMRC, which expresses the natural situation in the field