Manure is rich in essential elements, including nitrogen (N), phosphorus (P), and potassium (K), for plant growth. Although applying manure as a fertilizer at agronomic rates can restore organic matter and nutrients to the soil, over-application of manure may contribute to environmental issues such as eutrophication and water contamination. Manure nutrient prediction and variable rate application are promising new technologies to reduce the risk of over-application, however, the variability in manure nutrient concentrations and the time-lag caused by traditional chemical analysis of manure composition make precise nutrient application difficult to achieve.
Near-infrared (NIR) spectroscopy is a high-energy vibrational spectroscopy performed in the wavelength range between 750 to 2500 nm and has been proven to accurately determine total solid (TS), organic matter (OM), total nitrogen (TN), and ammoniacal nitrogen (NH4-N) of animal manure in several previous studies. A low-field nuclear magnetic resonance (NMR) device that is designed based on the absorption and emissions of energy in the radio frequency range of the electromagnetic spectrum is another potential method for predicting manure nutrients accurately. The main purpose of this manure sensing project was to determine if the NIR and NMR sensing techniques can provide robust prediction of manure nutrients and, therefore, improve the precision of field application.
What Did We D
We investigated NIR spectroscopy with reflectance and transflectance modes to predict micronutrients in dairy manure. In this study, 20 dairy manure samples were collected and spiked by dissolving a specific amount of ammonium chloride (NH4Cl) or Arginine to achieve incremental NH4-N and organic nitrogen (Org-N) concentrations, respectively. Each raw sample was spiked at four levels which were 1.25, 1.5, 2, and 4 times the NH4-N or Org-N concentrations of the raw manure as analyzed by a certified lab. All samples were scanned and analyzed using a NIR with a reflectance head sensor and a transflectance probe of three different optical path lengths. NIR calibration models were developed using partial least square regression analysis and the coefficient of determination (R2) and root mean square error (RMSE) were calculated to evaluate the models.
The accuracy and precision of a low-field NMR designated for manure nutrient prediction was assessed. Twenty dairy manure samples were collected and analyzed for TS, TN, NH4-N, and total phosphorus (TP) in a certified laboratory and using the NMR analyzer. Runtimes of 15 min to 90 min were tested to investigate their effects on accuracy and precision of NMR.
What Have We Learned
For the NIR study, the transflectance probe yielded calibrations that had higher R2 and RMSE for TS, ash, and particle size (PS), and reflectance sensor improved the accuracy of NH4-N and Org-N predictions. NIR sensors have the potential to predict N concentrations without being affected by the TS, ash content, and PS of the dairy manure.
The NMR predictions of TS, NH4-N, and TN were accurate for samples with relativley low TS, but not well correlated to the lab measurements for high TS samples. TP predicted by NMR was not affected by TS levels and the TP prediction was not precise and robust. The effects of runtime on the accuracy and precision of NMR prediction were not consistent.
Additional work is needed to improve the accuracy and precision of NIR calibration models. The procedure of spiking method in manure analysis using NIR techniques needs to be enhanced in order to be widely applied for preparing manure samples for NIR calibrations. Finally, further investigation of the methodology with other manure constituents such as P and K and conducting online variable rate application of organic fertilizer using NIR sensing system are needed to evaluate the potential effects of reducing the overall system variability.
Additional work to improve NMR prediction includes recalibrating the system based on specific manure samples and improving the accuracy and precision of TP prediction.
Xiaoyu Feng, Research Associate, University of Wisconsin-Madison
-Rebecca Larson, Associate Professor and Extension Specialist, University of Wisconsin-Madison; Matthew Digman, Assistant Professor, University of Wisconsin-Madison;
-Joseph Sanford, Assistant Professor, University of Wisconsin- Platteville
Feng, X.Y., R.A. Larson, and M. Digman. 2022. Evaluating the Feasibility of a Low-Field Nuclear Magnetic Resonance (NMR) Sensor for Manure Nutrient Prediction. Sensors 22(7):2438. https://doi.org/10.3390/s22072438
Feng, X.Y., R.A. Larson, and M. Digman. 2022. Evaluation of Near-Infrared Reflectance and Transflectance Sensing System for Predicting Manure Nutrients. Remote Sensing 14(4): 963. https://doi.org/10.3390/rs14040963
Support for this project was provided by the Wisconsin Dairy Innovation Hub.
The authors are solely responsible for the content of these proceedings. The technical information does not necessarily reflect the official position of the sponsoring agencies or institutions represented by planning committee members, and inclusion and distribution herein does not constitute an endorsement of views expressed by the same. Printed materials included herein are not refereed publications. Citations should appear as follows. EXAMPLE: Authors. 2022. Title of presentation. Waste to Worth. Oregon, OH. April 18-22, 2022. URL of this page. Accessed on: today’s date.