Final Report for GNE12-039
Project Information
Whole farm nutrient mass balance evaluations have shown that the most difficult information to collect for farms is crop yields, especially for corn silage and hay (alfalfa/grass) fields. In addition, on-farm research, a great tool for adaptive management, is hindered by lack of a practical way to collect yield data for farm- scale plots. In recent years, forage yield monitors have become available. Both the use of an accurate yield monitoring tool and the application of this tool/technology for on-farm testing of management alternatives aid in obtaining long-term sustainability for farms (adaptive management). Recent discussions with farmers showed their interest in evaluating the performance of yield monitors especially for alfalfa/grass fields. In this study, we worked with farms throughout NY to collect whole-farm alfalfa/grass forage yield data with yield monitors to (1) determine their accuracy; (2) determine within-field and between-field variability of yield, and (3) evaluate potential for their use for on-farm strip trials. The aim was to develop an on-farm research protocol. Additionally, we worked with farms throughout NY to determine the accuracy of John Deere’s forage yield monitor (both flow and moisture measurements) for individual loads. A combination of on-farm demonstrations, factsheets and farmer impact stories, together with a user-guide and decision-aids will allow farms to learn how to use forage monitors and implement adaptive management to achieve optimum nutrient use efficiency on a field-by-field basis. This project is expected to have positive impacts on farm economics, field management, and the environment through reduction of the total amount of nutrients brought onto farms.
Introduction
Accurate assessments of yield and forage inventories of corn silage and grass and alfalfa mixtures are essential to improve crop production and nutrient management on dairy farms. Yield maps can aid in the identification of unproductive fields and/or areas within fields, allowing for enhanced, site-specific management. In addition, yield assessments are needed for nutrient management at the field and within-in field level. For example, yield data are required to determine nutrient removal from fields through harvest. In New York State (NY), this is especially relevant for phosphorus (P) management on regulated farms where addition of P is limited to rates that do not exceed crop removal of P, if the field P index exceeds 100 (Czymmek et al. 2003). Yield assessments are also needed in NY if a regulated farm wishes to override the Land Grant University nitrogen (N) guidelines for corn (Ketterings et al. 2013). The latter adaptive management approach requires farmers to document yield and manage the corn stalk nitrate test (CSNT) to be below 3,000 mg kg-1 for two years if N application rates exceed those recommended by Cornell University (NRCS 2013). In addition to allowing for with-in field nutrient management and trouble-shooting, yield records can also aid in better forage inventory management, and thus in whole farm nutrient management and farm productivity and profitability.
Currently, there are many farms that do not have good forage yield records, primarily driven in the past by lack of equipment that allows for accurate assessment of yield. If determined, yields are typically measured using farm or truck scales. Data from the scales can be combined with estimations of forage moisture content using microwave ovens or Koster testers to determine dry matter (DM) yield (Pitt 1993). Both assessments are time-consuming and not well-suited with the busy schedules of farmers at harvest time. They also have limitations in their accuracy (Pitt 1993).
Measuring forage DM and yield has become more practical in recent years with the commercialization of forage yield monitors on self-propelled forage harvesters (SPFHs). One of the first systems, developed in the early 1990’s (Auernhammer et al. 1995) combined a mass flow sensor based on the radiometric principle to measure volume with a radar sensor to determine flow speed. More recent research investigated the use of capacitance type moisture sensors on SPFHs to measure DM, and load cells to measure weight (Lee et al. 2002). Most recently, Digman and Shinners (2008) developed a near-infrared reflectance (NIR) sensor to measure moisture, and subsequently DM, on a John Deere SPFH that was the precursor to the current sensor on the company’s machines.
Forage yield monitors on SPFHs offer geo-referenced yield measurements with DM data as a harvester moves through the field, allowing for real-time adjustments including cut length of the crop and inoculation rates. Most recently, capability to estimate forage components including protein, starch and fiber, was added. These added capabilities can allow for better bunk management and lead to better forage quality on farms.
The John Deere forage harvesting equipment combines a DM measurement with a mass flow reading to give DM yield per acre. The moisture sensor, called HarvestLab™ (HL), works with the company’s GreenStar™ global positioning system (GPS) to provide producers with on-the-go moisture and DM measurements, yield estimations, and coverage maps. The mass flow sensor is located in the cutterhead and measures feed roll displacement with potentiometers (John Deere 2012). This is combined with speed measurements from the GPS to estimate a wet yield per acre. The HL sensor measures moisture and DM using NIR and is located on the spout of the machine to measure the percent DM of the crop and estimate a dry yield. Although other companies have introduced yield monitoring systems, the John Deere system is currently the most common system used on farms in NY.
The adoption of forage yield monitors on SPFH has been slow due to the cost of equipment and lack of confidence in both the performance of the equipment and economic return to the investment (Digman & Shinners 2012). While NIR is an accepted method of measuring DM in both a laboratory setting and in the field, thus far very little work has been done to evaluate the precision and accuracy of forage yield monitoring systems as used on farms. The German Agricultural Society (DLG) conducted a trial in Germany which evaluated the HL sensor in several varieties of corn silage and concluded that it resulted in yield estimates that were within ±2% DM of the mean (DLG 2010). The study did not estimate the accuracy of the overall forage yield monitoring system, or alfalfa and grass crops.
Auernhammer, H., Demmel M., & Pirro, P.J.M. (1995). Yield measurement on self-propelled forage harvesters. Paper No. 951757, ASAE, St. Joseph, MI, USA.
Czymmek, K.J., Ketterings, Q.M., Geohring, L.D., & Albrecht, G.L. (2003). The New York Phosphorus Index. User’s Guide and Documentation. CSS Extension Bulletin E03-13. Ithaca, NY, USA: Cornell University. http://nmsp.cals.cornell.edu/publications/extension/PI_User_Manual.pdf.
Digman, M. F. & Shinners, K. J. (2008). Real-time moisture measurement on a forage harvester using near-infrared reflectance spectroscopy. Transactions of the ASABE, 51(5):1801-1810.
Digman, M. F. & Shinners, K. J. (2012). Technology Background and Best Practices: Yield Mapping in Hay and Forage. In Proceedings of the 42nd California Alfalfa and Grains Symposium, UC Davis, Davis, CA, pp 138-143.
DLG. (2010). HarvestLab- Feuchtemessung in Mais: im mobile Einsatz auf John Deere Feldhäcksler 7550i (Moisture measurement in corn: mobile use on John Deere forage harvester 7550i). DLG Prüfbericht 5913F. Groß-Umstadt: DLG e.V. – Testzentrum Technik und Betriebsmittel. http://www.dlg-test.de/tests/5913F.pdf.
Ketterings, Q.M., Czymmek, K., Albrecht, G., Gates, D. & Lendrum, J. (2013). Nitrogen for Corn; Management Options. Agronomy Fact Sheet 77. Cornell University Cooperative Extension, Ithaca, NY, USA. http://nmsp.cals.cornell.edu/publications/factsheets/factsheet77.pdf.
Lee, W.S., Burks, T. F., & Schueller, J. K. (2002). Silage Yield Monitoring System. Paper no. 021165, ASAE, St Joseph, MI, USA.
NRCS (2013). Natural Resources Conservation Service. Conservation Practice Standard: Nutrient Management. Code 590. Washington, DC, USA: NRCS.
Pitt, R.E. (1993). Forage moisture determination. NRAES-59. Ithaca, NY, USA: Northeast Regional Agricultural Engineering Service.
Cooperators
Research
Eleven farms located in NY with John Deere SPFHs equipped with the forage yield monitoring system, including GreenStar™ and HarvestLab™, were identified. Each farm was visited at least once throughout the 2013 growing season to collect data on crop DM, mass flow and DM yield. Four sets of assessments were done: (1) evaluation of within load DM variability (manual sampling); (2) evaluation of the accuracy and precision in determining DM using NIR HarvestLab™; (3) evaluation of mass flow (in-field wet weight) assessments in the field, and (4) evaluation of DM yield determination using the forage monitor system of John Deere.
Most hay fields in NY are alfalfa/grass mixtures with compositional changes from field to field based on seeding ratio and age of stand, on-farm assessments were done on alfalfa/grass mixtures. John Deere’s calibration curves are based on 100% alfalfa or 100% grass stands. For studies on farms in NY, assessments of dry matter content were done using alfalfa calibration curves. However, a laboratory evaluation of mature alfalfa was included in the assessment as well.
Within-load dry matter variability (manual method)
This section simulated the manual method of sampling that farms would be able to do without the HL equipment. For alfalfa/grass mixtures, 373 individual truck loads were sampled from eight farms. For corn, eleven farms were visited and data were collected for 350 loads. The HL measures the moisture in the crop, and displays both moisture and DM. For each load, the SPFH operator recorded the %DM displayed by HL followed by sub-sampling of the load at the bunk for DM determination with an oven. Sub-sampling consisted of five handfuls of forage taken while walking around the pile of forage immediately after the truck unloaded at the bunk silo. Sub-sampling was repeated two to eight times, depending on the speed at which loads arrived at the bunk. All sub-samples were dried in a 60°C forced-air oven for a minimum of 72 h to determine DM content, per ASABE Standard S358.3 (ASABE Standards 2012).
For 16 alfalfa/grass loads and 23 loads of corn silage, average DM content was determined for each possible combination of number of subsamples (2-8) per load, using MATLAB Release 2013A (Mathworks, Natick, MA, USA). Variance was determined for each subsample size (2 to 8 subsamples per load) based on 20 randomly selected combinations in each subsample size class. The minimum number of subsamples needed was determined as the number of subsamples beyond which the change in variance was less than 0.1% DM.
Statistical analyses were conducted using JMP Version 10 (SAS Institute Inc., Cary, NC, USA). A mixed model was used to calculate the residual variance (within load variance) for all farms and all cuttings for alfalfa/grass, and farm only for corn. This assessment was done for each of the crops separately. Sample size calculations were used to determine the margin of error (MOE) for all farms and cuttings, using the residual variance of the model. Confidence intervals (CI) were calculated for all loads over the whole season, for each farm individually, and for alfalfa/grass for each cutting within a farm.
Evaluation of NIR HarvestLab™ for accuracy and precision in determining dry matter
Controlled laboratory setting
Four field-fresh sub-samples of mature alfalfa and corn were chopped to 95 mm for alfalfa and 65-190 mm for corn to represent an actual forage harvest cut length. Samples were weighed to determine the initial wet weight of the samples. The HL moisture sensor was used to determine DM content using eight repeated individual DM measurements per sample. Following the HL %DM determination, samples were exposed to 60°C in a forced-air oven interrupted by eight individual measuring events over a period of 5 h to create a range from of DM for both crops. At each DM measurement, weight and HL-determined %DM content were recorded (eight repeated measurements per sample averaged) for a total of 32 measurements over eight sampling periods per sample. Following the final sampling round, samples were dried in a 60°C forced-air oven for 72 h to determine the initial DM content per ASABE Standard S358.2 (ASABE Standards 2012). The preloaded DM curve developed by the Association of German Agricultural Analytic and Research Institutes (VDlufa) was used, according to John Deere’s standards (John Deere 2012). Crop calibration curves were changed between corn and alfalfa evaluations.
Statistical analyses were conducted using JMP Version 10 (SAS Institute Inc., Cary, NC, USA). HarvestLab™ DM values were compared to the oven- and scale-determined DM content to determine accuracy, and evaluated using bias calculated as oven-derived DM minus HL values for DM. Mean absolute bias was calculated. Confidence intervals around the 1:1 line were calculated as measures of accuracy. The standard error (SE) of the mean absolute bias was used to evaluate precision.
In-field evaluation
Truckloads of freshly chopped alfalfa/grass mixtures and corn silage were sampled on eight farms (272 individual loads) and eleven farms (342 loads), respectively, in NY during the 2013 growing season. At the beginning of the sampling period, the SPFH mass flow sensor was calibrated according to John Deere specifications (John Deere 2012). The SPFH operator recorded the DM as determined by HL during harvest, followed by sub-sampling of four to eight one-gallon (3.8 L) bags per load at the bunk as outlined above. Dry matter was determined per ASABE Standard S358.3 (ASABE Standards 2012).
A mixed model was run using JMP Version 10 (SAS Institute Inc., Cary, NC, USA) to compare the HL %DM values to the oven- and scale-determined DM content. This model was run for all eight farms and multiple cuttings combined for alfalfa/grass, and for the eleven farms for corn, to determine the accuracy and precision as described in the controlled laboratory setting section above.
Yield monitors for in-field yield (wet weight) determination
To evaluate the precision and accuracy of the mass flow sensor on the forage harvesters, the weight recorded by the yield monitoring system was compared to the scale-weight per load. The yield monitor-displayed weight was recorded per load by the SPFH operator followed by weighing of the truck using on-farm scales. For alfalfa/grass mixtures, 119 loads were evaluated on two farms. Farm 1 was visited at 2nd (29 loads) and 3rd (25 loads) cuttings while Farm 2 was sampled at 1st and 2nd cuttings (25 loads each) and 4th cutting (15 loads). For corn silage, 80 loads were evaluated on three farms, with 30, 40 and 10 loads per farm. Data collection was limited to these farms due to the lack of truck scales in close proximity to the bunk silo on the other dairy farms in the study.
A mixed model was used to compare the forage monitor-determined wet weight with the scale-determined wet weight for all farms (and cuttings for alfalfa/grass mixtures), to determine the accuracy and precision as described above. All models were run using JMP Version 10 (SAS Institute Inc., Cary, NC, USA).
Yield monitors for field dry matter harvest
Yield determined by the forage yield monitor (HL and flow sensor readings combined), were compared to DM yields obtained with truck scale-determined weight and oven-determined DM content. In total, 94 loads of alfalfa/grass mixtures were evaluated on two farms and 80 loads of corn silage were evaluated on three farms.
A mixed model was used to compare monitor-determined yield with scale- and oven-determined yield for all farms and cuttings for alfalfa/grass, and farm only for corn.
To determine whether an accurate prediction of dry yield can be made using wet weight only, with an estimation of average DM (i.e. without the NIR sensor), two models were generated, using as inputs: (1) machine-calculated dry yield only and (2) mass flow only. Model fit was determined using Akaike information criterion (AIC), -2 log likelihood, root mean square error (RMSE) and R² values.
Documenting methods for obtaining and combining yield data with crop records
The yield evaluations were done using data from a 1000-cow dairy farm in Wyoming County, NY, that grows both alfalfa/grass hay and corn silage. The farm’s typical crop rotation is three years of corn silage followed by three years of an alfalfa-grass mixture.
Yield was measured from 2000 through 2013 and recorded for a total of 107 fields. The records included harvested area, crop grown, and dry matter (DM) yield for each field. Yield was calculated using the sum of the weight of all loads for each field determined with a farm scale that was located near the bunk silo (Fig. 1). For each year, area-weighted mean DM yield of each crop was calculated to determine whole-farm (corn silage and alfalfa/grass) yield.
Soil physical properties for each field included soil series, hydrologic group (Ketterings et al. 2003a), drainage class (Soil Survey Division Staff 1993), and soil management group (Cornell Cooperative Extension 2013). Soil chemical properties measured included soil organic matter (OM), pH, phosphorus (P), potassium (K), calcium (Ca), and magnesium (Mg). Analyses were conducted by Spectrum Analytic Inc. (Washington Court House, OH). Organic matter and pH (1:1 (w:v) water extract) were analyzed using methods as described by Storer (1984), with OM method adapted to 360°C for two h as described in Schulte and Hoskins (1995). Phosphorus, K, Ca and Mg were analyzed by Spectrum Analytic (Washington Court House, OH) using the Mehlich-3 extraction as outlined in Wolf and Beegle (1995). Mehlich-3 P values were converted to Cornell University Morgan-P equivalents based on Ketterings et al. (2002) and soil P levels were classified as low, medium, high or very high according to Cornell University guidelines for field crops as documented in Ketterings et al. (2003b).
Trends in annual weighted mean DM yields (corn silage, alfalfa/grass, and total yearly production) were analyzed using simple linear regression. Annual climate data included rainfall and growing degree days obtained from the Climate Information for Management and Operational Decisions (CLIMOD 2014). These data were used to evaluate the impact of weather on trends in yield over time; analyses were done for March to October (full growing season), March to April (corn planting season), and July to August (corn tasseling window). For alfalfa/grass cuttings monthly weather data were analyzed for their impact on yield. Simple linear regression was used to compare the amount of rainfall during each of the periods to the mean yield, and R2 and p-value were used to evaluate the trends.
Spatial variability was determined using 107 fields with two or more rotations of data. Of those fields, 61 fields had six corn years each and 71 fields had five full production years for alfalfa/grass. The mean yield and coefficient of variation (CV) were calculated for each field. The fields were divided into four quadrants (Q1-Q4), using the overall weighted mean yield and mean CV as cutoffs for the quadrants: (1) above mean yield, below mean CV (Q1); (2) above mean yield, above mean CV (Q2); (3) below mean yield, above mean CV (Q3); and (4) below mean yield, below mean CV (Q4). Mean yield and CV were calculated for each quadrant, and significant differences among quadrants were determined using Tukey’s least significant difference (p≤ 0.05) in JMP Version 10 (SAS Institute Inc., 2012). Significant differences among quadrants were determined for physical (hydrologic group, drainage class, and soil management group) and chemical (OM, pH, available P, K, Ca, Mg) soil properties.
Evaluation of Forage Within-Field Variability Using GIS
Within field yield data for corn silage was collected from three farms in New York from 2011-2013. Initial data processing was conducted by Agrinetix, LLC to standardize the data, check for measurement errors, and delete/correct erroneous data. The average yield and CV were calculated for each field with three years of data on each farm and graphed using the average and CV to separate into quadrants, similar to the previous section. For each of the three farms, two fields were selected; one field each with (1) above average yield, below average CV and (2) below average yield and above average CV. The field yield data was imported into ArcGIS 10.0 and processed using the kriging interpolation method to determine individual areas of the field that have high/low yield and high/low CV.
ASABE Standards (2012). S358.3: Moisture measurement- Forages. St. Joseph, MI, USA: American Society of Agricultural and Biological Engineers.
CLIMOD (2014) Ithaca, NY: Northeast Regional Climate Center.
Cornell Cooperative Extension (2013) 2014 Cornell Guide for Integrated Field Crop Management. Ithaca, NY: Cornell University.
Ketterings QM, Klausner SD, Czymmek KJ (2003a) Nitrogen guidelines for field crops in New York. CSS Extension Series E03-16. Cornell University, Department of Crop and Soil Sciences, Ithaca NY.
Ketterings QM, Klausner SD, Czymmek KJ (2003b) Phosphorus guidelines for field crops in New York. CSS Extension Series E03-15. Cornell University, Department of Crop and Soil Sciences, Ithaca NY.
Schulte EE, Hoskins B (1995) Recommended soil organic matter tests. In: Sims JT, Wolf A (eds) Recommended soil testing procedures for the northeastern United States. Bulletin 493, NEC-67, 3rd edn. University of Delaware, Newark, pp 63-74.
Soil Survey Division Staff. (1993) Soil survey manual. Soil Conservation Service. U.S. Department of Agriculture Handbook 18. Washington, DC.
Storer, DA (1984) A simple high sample volume ashing procedure for determination of soil organic matter. Commun Soil Sci Plant Anal. doi:10.1080/00103628409367515.
Wolf A, Beegle DM (1995) Recommended soil tests for macronutrients: phosphorus potassium, calcium and magnesium. In: Sims JT, Wolf A (eds) Recommended soil testing procedures for the northeastern United States. Bulletin 493, NEC-67, 3rd edn. University of Delaware, Newark, pp 30-38.
Yield Monitor Accuracy Evaluation:
In the field study that involved multiple farms, the HL sensor underestimated %DM of alfalfa/grass mixture by 3.7% DM, which is less accurate than what can be accomplished in manual sampling of the harvest. The HL sensor was more accurate in the corn silage crop, where %DM was estimated within 3.0%, but still less accurate than manual sampling. In the alfalfa crop, much of the variability was due to field-to-field variability, which could be due to differences in alfalfa/grass mixtures among fields. For corn silage, the variability in DM content was explained mostly by farm-to-farm (equipment) differences. To make the DM prediction more accurate, the alfalfa/grass mixtures present in NY fields needs to be addressed using new calibrations.
The mass flow sensor was more accurate for corn silage than alfalfa/grass mixtures. The mass flow sensor was able to predict the field wet weight of alfalfa/grass mixture within 0.5 Mg per load and 0.2 Mg per load for corn silage, which represented 15% and 7% errors, respectively. Due to high field-to-field variability, it is recommended that calibrations take place frequently.
Overall, the yield monitor was able to predict the DM yield of alfalfa/grass mixtures within 0.5 Mg DM ha-1 and 1.1 Mg DM ha-1 for corn silage. The sensor was more accurate for corn silage than alfalfa/grass mixtures, based on percent error of total yield. The mean bias of 0 Mg DM ha-1 in alfalfa/grass and -0.2 Mg DM ha-1 in corn silage indicate that for many loads, the over- and under-estimation errors balance each other, resulting in accurate whole farm yield estimations. Model comparisons showed that machine-determined DM yield using both mass flow and DM content is needed to accurately estimate yield of alfalfa/grass mixtures. For corn silage, the model suggested that HL %DM measurements may not be needed to accurately predict yield, and accurate yield estimations can be made using mass flow measurements and traditional DM estimations (i.e. oven-drying or on-farm Koster tester).
Forage yield monitors with NIR estimation of DM and mass flow estimation for volume can provide precise and accurate measures of DM yield assuming calibrations are conducted regularly and best management practices are followed. Yield monitors as evaluated in this study may not be accurate enough for determination of yield of small-plots in on-farm research trials or for research where replications are spread over multiple fields. They can, however, be used for implementing adaptive management and large-scale on-farm trials (at least one truck load per plot), where treatment difference greater than 0.5 Mg DM ha-1 (alfalfa/grass mixtures) or 1.1 Mg DM ha-1 (corn) are expected.
This worked showed that the yield monitoring equipment is not yet accurate enough to be used for plot research studies in corn silage and alfalfa, but is accurate over a cutting of hay or whole growing season of hay and corn silage, which can still be utilized for adaptive management purposes.
Additionally, protocol for calibration of the forage yield monitors was developed fall 2012 using four farms. Protocol was finalized during winter 2013, and distributed to 11 farms in the 2013 growing season. A “Season Checklist and Calibration Procedure” document was created in collaboration with Agrinetix, LLC and distributed for farms in the 2013 growing season. Both the calibration protocol and season checklist are available on the NMSP website.
Yield Record Keeping Case Study:
On this case study farm, corn silage yields increased from 2002-2013, while alfalfa/grass yields remained constant. Yield varied both temporally with rainfall throughout the growing season and spatially among fields. The consistently high yielding corn fields exceeded 15.6 Mg ha-1 with a CV less than 16.4%. Alfalfa/grass fields that were consistently high yielding exceeded 9.9 Mg ha-1 with a CV less than 21.6%. Results showed that the higher yielding fields tended to be more consistent in yield over time than below average yielding fields. The highest and most consistently yielding fields for both crops had better-drained soils, and were classified as optimum or higher in soil test P, and higher in OM than the lower yielding and more variable fields. High yielding corn silage fields also had higher extractable K, Ca and more OM, which could reflect a longer manure history of these fields. Across all fields for both crops, yield increased as soil test P increased up to 16.1 mg kg-1 for corn silage, and 14.6 mg kg-1 for alfalfa/grass. There was no relation between yield and soil test P at higher soil test levels. These results support previous findings in NY which showed when a field has a soil test P greater than 10 mg kg-1, P fertilizer addition did not increase yield (Ketterings et al. 2005). These results could suggest that farmer practices that improve soil drainage (tile drainage), conserve or even increase organic matter (reduced tillage and cover crops), and enhance soil test P (manure application) to optimal levels, might be effective in increasing the overall corn silage yield.
Within field yield variability assessment:
For the fields in the study, the within field variability was correlated to the overall field performance. Fields with high yield and low CV had a larger area of the field with both high yield and low CV, and likewise with field with low yield and high CV. These results indicate that reducing within field variability through improved drainage and variable rate fertilizer application may increase overall field performance over time.
Ketterings QM, Swink SM, Godwin G, Czymmek KJ, Albrecht GL (2005) Maize silage yield and quality response to starter phosphorus fertilizer in high phosphorus soils in New York. J Food Agric Environ 3(2): 237-242
This project has increased awareness of forage yield monitor use throughout NY through on-farm meetings with the farm operators, meetings with equipment dealerships throughout the state, presentations and field days. Farms who grow forage crops are now aware of the importance of machine calibration, which will lead to increased accuracy and ability to make decisions on nutrient management, manure application, and harvest. Several farms in the study expressed being able to make better decisions when pricing their custom harvested forage, make more informed harvesting decisions that will impact forage quality, and are now able to use the data from the machines to tighten crop rotations. Equipment dealers are better informed about the accuracy of the machines they are selling, which will help farms be better educated when making purchasing decisions. With an updated calibration protocol, we were able to implement proper calibration procedures on farms in the study, as well as demonstrate to John Deere equipment dealers to ensure more accurate data collection, and thus maps for management decisions.
Additionally, with protocol for field data collections and map creation, the farmers will be able to document yield levels over time, determine nutrient use efficiency, and make more informed nutrient management decisions which will enhance nutrient management on their farms. These maps also demonstrated the presence of within field variability and showed the importance of managing nutrients based on that inherent variability, as many farms still do not adjust nutrients on a field-by-field basis, let alone variable rate within a field. While this equipment is not accurate enough for plot-by-plot research, it has demonstrated that it can be a useful tool for adaptive management, which includes assessment of crop yield response to management alternatives (NRCS 2013). In NY, the standard refers to Land-grant University guidelines which, for nitrogen (N) management, now state that farmers can determine N application practices for corn based on: (1) soil type specific corn yield potentials as documented in the Cornell University yield and soil database ( Ketterings et al. 2003); (2) three years of actual corn yield records; (3) findings of two years of on-farm replicated trials with a minimum of four replications and five N rates including a zero-N control treatment; or (4) yield measurements and corn stalk nitrate test (CSNT) results (Ketterings et al. 2013). The latter is a recent adaptive management strategy that allows farmers to override the Cornell University yield database without evidence of higher yields, as long as yields are documented and CSNTs are managed below 3000 mg kg-1 for each year in which this strategy is used (Ketterings et al. 2013). This adaptive management approach allows for continued adjustments to field management practices to achieve better nutrient use efficiency and yields over time. This project has shown that yield data from forage harvesters can be used as a tool for adaptive management implementation.
NRCS (2013). Natural Resources Conservation Service. Conservation Practice Standard: Nutrient Management. Code 590. Washington, DC, USA: NRCS.
Ketterings, Q.M., Czymmek, K., Albrecht, G., Gates, D. & Lendrum, J. (2013). Nitrogen for Corn; Management Options. Agronomy Fact Sheet 77. Cornell University Cooperative Extension, Ithaca, NY, USA. http://nmsp.cals.cornell.edu/publications/factsheets/factsheet77.pdf.
Education & Outreach Activities and Participation Summary
Participation Summary:
On-Farm Demonstrations:
Two field days were held in 2013 to demonstrate the John Deere self-propelled harvester yield monitors. Each focused on one crop and talks included the importance of yield monitoring, an overview of harvester components involved in yield monitoring, information on calibration, and farmer experiences.
- Alfalfa Field Day: Gary Swede Farms, Pavilion, NY. Monday, July 15th. ~25 people.
- Corn Silage Field Day: O’Hara Machinery, Auburn, NY. Thursday, August 22nd. ~20 people. News article- Sergeant, D.J. (2013). Researchers focus on yield monitor accuracy. Lancaster Farming. 8/31/2013. Available online at: http://www.lancasterfarming.com/Researchers-Focus-on-Yield-Monitor-Accuracy-#.U2FNs_ldWSo
Talks:
- Long E. (2013). Assessing accuracy of forage yield Monitors for use in alfalfa. NEBSCA Annual Meeting 2013. Newark, DE, June 23-26, 2013. ~30 people.
- Long E. (2013). Assessing accuracy of forage yield Monitors for use in alfalfa. ASA-CSA-SSSA International Annual Meetings 2013. Tampa Bay, FL, November 2-5, 2013. ~40 people.
- Long E. (2013). Assessing accuracy of forage yield Monitors for use in alfalfa and corn silage. 2013 Cornell Cooperative Extension Agriculture and Food Systems In-Service. Ithaca, NY, November 19-21. ~25 people.
- Long E., D. Russell, B. Stoll (2013). Yield monitors for forages: Experiences in NY with corn and alfalfa. Northeast Region Certified Crop Advisor Annual Training. Advanced Training. December 5, 2013. Syracuse, NY. ~30 people.
- Long E. (2014). Yield Monitors for Forage Production: Experiences in NY with corn silage and alfalfa/grass. Oneida County Cornell Cooperative Extension Crop Congress. January 7th, 2014. Waterville, NY. ~60 people.
- Long E. (2014). Whole Farm and field Corn and Alfalfa Yield Variability on a CAFO Dairy Farm. ASA-CCA-SSSA International Annual Meetings 2014. Long Beach, CA, Nov 2-5, 2014. (poster)
- Long E. (2014). Assessment of Yield Monitoring Equipment in Determining Moisture and Yield of Forage Crops. ASA-CSA-SSSA International Annual Meetings 2014. Long Beach, CA, Nov 2-5, 2014. (poster)
Publications:
Long, E., Ketterings, Q.M. (2015) Whole farm corn and hay yield variability; a dairy farm case study. Agronomy for Sustainable Development. Under review.
Long, E., Ketterings, Q.M., Russell, D., Vermeylen, F., and DeGloria, S. (2015) Assessment of yield monitoring equipment for dry matter and yield of corn silage and alfalfa/grass. Journal of Precision Agriculture. Under review.
Project Outcomes
Economic Analysis
While this project does not quantify any direct changes in farm income or economic status of farms, the HarvestLab™ system for a self-propelled forage harvester is very expensive and this project showed farms how to better make use of their investment. One of the ongoing concerns that farmers have when implementing precision agriculture practices is the cost. This project demonstrated how to increase precision and accuracy of the forage yield monitoring technology. Better data during harvest will increase forage quality, thus increasing efficiency and milk production and can increase income for the farm. Most of the participating farms were only using the on-the-go data component to make decisions during harvest, and not utilizing the yield maps post-harvest. This project demonstrated the ability to make useful maps using forage yield data which can increase nutrient efficiency, increase yields over time, and increase economic efficiency of forage production.
Farmer Adoption
All participating farms were very appreciative of the work conducted in this study. When the study began, most farms did not understand the calibration process, and thus were complaining of what they felt were inaccurate data. Through the research project, they gained understanding and appreciation for the calibration process and how attention to details during harvest can lead to better results.
One farm in the study does a large amount of custom harvesting. Through this project, they gained enough confidence in their equipment that they decided to no longer weigh all loads to estimate total amount of forage harvested. This will cut down on time during harvest and increase the farm’s efficiency.
This project not only impacted farmers directly, but also consultants, Cooperative Extension personnel and equipment dealers across NY. After the field day, an independent consultant commented that he was much more comfortable discussing forage yield monitoring equipment with his clients. Also, at an unrelated forage field day this past summer, an equipment dealer referenced this study when describing to attendees how accurate the HarvestLab™ system is.
At the time this project was planned, John Deere was the only equipment company with a system common enough to evaluate. Since then, other equipment companies have come out with their own forage yield monitoring system. It would be useful to conduct a similar study on those other systems to compare the accuracy and precision, and calibration procedures across companies.
As mentioned, the most common alfalfa system in NY is an alfalfa/grass mixture, rather than a pure stand of alfalfa. This system is calibrated for either 100% grass or 100% alfalfa. It would be useful to do work related to accuracy of each of those calibrations and the percent alfalfa in the mixtures to assist farms in our cropping system when making decisions on which calibration curve to use.
Many other questions came up during this study related to impact of accuracy on different corn hybrids (ie. BMR vs. conventional) and how useful they system is in alternative crops like triticale, rye or sorghum. All of these questions could not be answered through this one project, and would be useful to research to continue to improve how farmers use this precision agriculture tool.