Monitoring nitrogen species in the soil solution is essential to optimize agricultural fertilization, but their in situ measurement is costly and time-consuming. Ion-selective electrodes (ISEs) represent a cost-effective alternative, although they have limitations: they cannot detect all species, and they introduce noticeable errors in the ones they do measure. To overcome this, this work proposes an Extended Kalman Filter (EKF) aimed at the joint estimation of concentrations (states) and kinetic rates (parameters) of the nitrification process. The algorithm relies solely on noisy measurements of two states (ammonium and nitrate). Its validation on experimental data demonstrates that the estimator robustly reconstructs the system dynamics against different parameter initializations and the presentation order of the experiments. Furthermore, this methodology successfully infers the intermediate state of the process, for which empirical measurements are unavailable.