A new empirical model (SNAP) combines a simple laboratory measurement of the basal rate of N mineralisation with the modifying effects of daily temperature and water content to predict seasonal and annual rates of mineralisation of forest soils. Short-term (20-60-day) aerobic incubations of either undisturbed or bulked and mixed soil were found suitable for prediction of the basal rate of N mineralisation. Data from laboratory incubations of a range of soils were used to calibrate empirical relationships describing the effects of temperature (Tm) and water (Wm) on rates of N mineralisation. Submodels for predicting daily average temperature (STUF) and water content (SWUF) for up to 3 surface soil layers were developed and used to provide inputs to the Tm and Wm functions, respectively. Inputs required for SNAP are restricted to variables whose values are easily obtained. In addition to the amount of N mineralised during a short aerobic laboratory incubation, other soil properties required are bulk density, gravel and clay content, and upper and lower limits of soil water content. Climatic data required included daily air temperature, rainfall, and solar radiation. Other inputs are slope, leaf area index of the stand, and approximate mass and height of litter. Predicted rates of N mineralisation have been verified using data from 9 native forests, 12 radiata pine plantations, and 12 eucalypt plantations from across southern Australia. Despite the wide range of forest types, soil types, climatic regions, and management systems, predicted annual rates of N mineralisation were in close agreement with those observed in the field, regardless of whether daily soil temperature and water content were predicted ( R 2 = 0.76, P < 0.001, n = 127) or observed ( R 2 = 0.78, P < 0.001, n = 68). Sensitivity analysis showed that it was most important to minimise analytical error in inputs used to calculate the basal rate of N mineralisation (i.e. soil temperature, water content, and N mineralised during laboratory incubation). The model was more sensitive to daily soil temperature than to daily soil water content.

The FullCAM model was developed for full carbon accounting in agriculture and forests at project and national scales. For forest systems, FullCAM links the empirical CAMFor model to models of tree growth (3PG), litter decomposition (GENDEC), and soil carbon turnover (RothC). Our objective was to calibrate RothC within the FullCAM framework using 2 long-term forestry experiments where productivity had been manipulated and archived and new soil samples were available for analysis of carbon within the various pools described by RothC. Inputs of carbon to soil at these trials were estimated by calibrating FullCAM to temporal data on above-ground growth, litterfall, and accumulation of litter. Two alternative submodels are available in FullCAM (CAMFor and GENDEC) for predicting decomposition of litter, and thus the input of carbon into the soil. Calibration of RothC was most sensitive to the partitioning of carbon during decomposition of debris between that lost as CO2 and that transferred to soil. Turnover of soil carbon was best simulated when the proportion of carbon lost to CO2 from relatively labile pools of debris was 77% (when simulated by CAMFor) and 95% (when simulated by GENDEC), whereas resistant pools of debris lost about 40% to CO2 during decomposition. Although rates of decomposition of pools of soil carbon were originally developed in RothC for agricultural soils, these constants were found to be also suitable for soils under plantation systems.