Aquatic weeds prediction: A comparative study

Aquatic weeds are the greatest generator of biomass in aquatic environment which motivates using intelligent methods for the prediction and estimation of indicators that affect the growth of such weeds. In this study a set of new interpolation methods are used and assessed over the study area for predicting a set of chemical indicators that can predict and affect the growth of weeds. The used methods are bi-harmonic, regularized spline with tension, Barnes, tri-scatter, and kriging. The different interpolants are used to create thematic maps representing the different chemical indicators that are sensed at discrete positions for supporting decision making. The performance of individual interpolants is assessed using mean square error over a set of test sites. Results prove that the Tri-scatter interpolant is the one with best performance for all the sensed indicators while the regularized spline performs well when the number of points for interpolation is large enough.