Temperature functions of the rate coefficients of net N mineralization in sandy arable soils. Part II. Evaluation via field mineralization measurements

The aim of this study was to evaluate experimentally derived temperature functions for the rate coefficients of net N mineralization in sandy arable soils from NW Germany via field measurements. In part I of this paper (Heumann and Böttcher, 2004), different temperature functions for the rate coefficients of a two‐pool first‐order kinetic equation were derived by long‐term laboratory incubations at 3°C to 35°C. In this paper, field net N mineralization during winter of 25 plots was measured in undisturbed soil columns with a diameter of 20 cm to the depth of the A_p horizon. Mean simulated net N mineralization with the most adequate multiple functions corresponded also best with the mean of the measured values despite of an overestimation of about 10%. Distinctly larger deviations under use of other temperature functions (Arrhenius, Q₁₀) were directly related to their deviations from mean, experimentally derived rate coefficients. Simulated net N mineralization in the soil columns was significantly correlated with measured values, regardless of the temperature functions. Yet the goodness of fit was generally relatively low due to the spatial variability of measured net N mineralization within replicate soil columns, although the mean CV (38%) was by far not extraordinary. The pool of slowly mineralizable N contributed considerably to net N mineralization during four to five winter months, on an average 10.0 kg N ha^–1, about one third of total simulated N mineralization. Sometimes, it contributed even 21.3 kg N ha^–1, which is almost sufficient to reach the EU drinking‐water limit for nitrate in these soils. Simulations with widely used functions that were once derived from loess soils overestimated mineralization from pool N_slow in the studied sandy arable soils by a factor of two.