Comparison between observed and DeNitrification-DeComposition model-based nitrous oxide fluxes and maize yields under selected soil fertility management technologies in Kenya

Abstract

Process-based biogeochemical models can be used to simulate soil nitrous oxide (N2O) fluxes and maize yields and draw insights on yields improvement and climate change mitigation options. We compared both observed and DeNitrification-DeComposition (DNDC) simulated soil N2O fluxes and maize yields. We used DNDC model to simulate soil N2O fluxes and maize yields under four soil fertility treatments (inorganic fertiliser, animal manure, animal manure + inorganic fertiliser and control,) replicated thrice for 1 year in Central Highlands of Kenya. We sampled soil N2O fluxes using static chambers installed in each plot. We analysed soil N2O using gas chromatography and calculated cumulative fluxes using trapezoidal rule linear interpolation between sampling dates. DNDC showed poor results in simulating daily N2O fluxes (157.16% ≤ normalised root mean square error (nRMSE) ≤ 324.01% and 0.90 ≤ modelling efficiency (NSE) ≤ 0.96), good to excellent performance in simulating cumulative annual soil N2O fluxes (6.16 ≤ nRMSE ≤12.86 and 0.63 ≤ NSE ≤ 0.86) and good to excellent performance in simulating maize yields (1.15% ≤ nRMSE ≤13.86% and 0.51 ≤ NSE ≤ 0.88) across all soil fertility treatments. The DNDC model had good to excellent performance in simulating cumulative soil N2O fluxes and crop yields across treatments. Though the model captured yield-scaled N2O fluxes and N2O emission factors across treatments, they were underestimated under manure treatment. There is a need to continue calibrating the DNDC model for improved capture of daily N2O fluxes and uptakes on Kenyan soils.