Eliminating uncertainty in soil N supply could reduce fertilizer input, but the amount of N mineralized during plant growth is usually still unknown. We aimed to test the relatively simple two-pool net N mineralization model NET N that uses site-specific temperature and soil water functions as well as pedotransfer functions for deriving the pool sizes and was developed for NW Germany. The objectives were to (1) evaluate, if field net N mineralization under unfertilized winter wheat could be satisfactorily simulated, and to (2) examine the variation in time patterns of net N mineralization within years and sites and from two functional N pools: a rather small, fast mineralizable N pool (N-fast) and a much greater, slowly mineralizable N pool (N-slow). NET N simulations for 36 site-year-combinations and up to five dates within the growing season were evaluated with detailed N balance approaches (calculated from: soil mineral N contents, plant N uptake using estimates of green area index, simulated N leaching). Simulated net N mineralization was highly significantly correlated (r(2) = 0.58; root mean square error = 24.2 kg N ha(-1)) to estimations from the most detailed balance approach, with total simulated net N mineralization until mid August ranging from 62.1 to 196.5 kg N ha(-1). It also became evident that N mineralization from pool N-slow-in contrast to pool N-fast-was considerably higher for loess soils than for sandy or loamy soils. The results suggest that NET N was adequate for simulations in unfertilized winter wheat. However, further field studies are necessary for proving its applicability under fertilized conditions.
