Validation of the DNDC-Rice model to discover problems in evaluating the nitrogen balance at a paddy-field scale for single-cropping of rice

Abstract

The DNDC (DeNitrification–DeComposition)-Rice model is one of the most advanced process-based models for estimating greenhouse gas emissions from paddy fields, and can be used to simulate the N balance of a paddy field. In this study, we validated DNDC-Rice using field observation data, including N balance data, and revealed problems when using the model to evaluate a paddy field’s N balance. Using a modified process submodel for fixation of ammonium (NH4+) ions by clay, DNDC-Rice simulated the dry weight of roots, stems, and grains well, but overestimated leaf dry weight. The normalized root-mean-square errors (nRMSEs) for the root, stem, grain, and leaf dry weights were 13, 16, 7, and 60 %, respectively. DNDC-Rice also overestimated the leaf area index (LAI) and leaf N content, with nRMSEs of 125 and 37 %, respectively. The overestimation of leaf dry weight and LAI resulted from overestimation of N uptake by rice and of N allocation to leaves. The high N uptake might have been caused by a high available soil N content. At harvesting, the simulated N balance (=N input − N output) was −38.8 kg N ha−1, which was much lower than the N balance determined by observations and from relevant literature (12.8 kg N ha−1). The underestimation of the N balance resulted mostly from the model’s inability to calculate dry N deposition and N fixation as inputs and from overestimation of grain N uptake as outputs.