Estimating soil carbon change using the web-based Nutrient Tracking Tool (NTT) with APEX

Abstract

Understanding soil carbon (C) balance within agroecosystems is an important piece of reducing agriculture-related climate impacts and improving soil quality. The web-based Nutrient Tracking Tool (NTT) has been widely applied for estimating nutrient fate and transport, erosion potential, and crop yield using the Agricultural Policy Environmental eXtender (APEX) model. NTT simulates a variety of agricultural systems and is in the process of being improved to provide a more holistic understanding of the impact of management practices on agricultural sustainability as it is adopted in various parts of the United States. One improvement in NTT is the incorporation of APEX’s soil organic C (SOC) estimation into NTT to allow decision-makers the ability to estimate how management practices impact C balance on a free and user-friendly platform. In order to test this additional outcome, NTT was used to estimate SOC in a series of simulations using recorded SOC change from a literature review. Nine studies with SOC measurements at least five years apart that took place in the contiguous United States and had sufficient management data to reliably run NTT were selected. The selected studies consisted of 131 paired SOC measurements (initial and final) across a wide range of cropping systems, including no-till, conventional tillage, cover crops, nutrient management systems, and crop rotations. Details from each study location were input into NTT (location, slope, planting date, tillage practice, fertilization rate, and soil properties—texture and initial SOC) and run using modified NTT/APEX 806. Measured SOC and SOC change were then compared with those of predicted values. Overall, the correlation between measured and predicted final SOC was r 2 = 0.57. The average deviation between simulated and measured soil C change was −0.39 ± 0.03 Mg ha−1 (12.5% difference). This corresponds to an average percentage change of 0.27% with the simulation and −0.68% with measured values across all sites; the percentage change is relatively low because of averaging sites with opposing change directions. Sites were also grouped by management practice to determine how NTT functions in varying management practices; the practices with the lowest deviation were continuous corn (0.12 Mg ha−1 error; 39.55% difference) and intensive tillage (−0.16 Mg ha−1 error; −35.33% difference) and the practices with the highest deviation were zero fertilizer systems (3.75 Mg ha−1 error; 146.27% difference). Considering the fact all weather information was obtained from NTT databases (PRISM database) and few parameters were modified in APEX, the results obtained from this comparison study are promising. One major limitation with this study is that most of the measured values for verification came from the Midwest and north central United States with few from the southern or western states. Nevertheless, this initial look is a good first step toward a robust C decision-making tool. In future work we plan to verify C results from a wider variety of locations in the United States and a wider variety of agricultural land uses.