Yield Monitor Data Cleaning is Essential for Accurate Corn Grain and Silage Yield Determination

Abstract

Core Ideas Corn silage and grain yield monitors collect yield data of relevance to farmers. Evaluation of quality of yield monitor data is essential, especially for silage. A data cleaning protocol, consistent across fields, farms, and years, is needed. Semi‐automation is needed for quick and consistent processing of whole‐farm data. Yield monitor data are being used for a variety of purposes including conducting on‐farm studies, assessing nutrient balances, determining yield potential, and creating management zones. However, standardization of raw data processing is needed to obtain comparable data across fields, farms, and years. Our objective was to evaluate the impact of data cleaning protocols on corn (Zea mays L.) grain and silage yield data at the whole field (with and without headlands) and within field (soil map unit) scales. Corn silage data from 145 fields (three farms) and grain data from 88 fields (three farms) were processed. Comparisons were made to evaluate yields among three levels of cleaning: (i) none; (ii) automated cleaning (“Auto”) with filter settings derived for 10 fields per farm; and (iii) automated cleaning with manual inspection for unrepresentative patterns, after the automated cleaning step was completed (“Auto+”). The Auto+ cleaning process was conducted separately by three individuals to evaluate person‐to‐person differences. Spatial Management System software was used to read raw data and transfer to Ag Leader format. Yield Editor software was used to clean data (Auto and Auto+). Results showed the necessity of data cleaning, especially for corn silage. However, considering less than 5% deviation between methods at three spatial scales, the Auto and Auto+ cleaning resulted in similar output, as long as (i) each field or subfield included at least 100 harvester measurement points, and (ii) a moisture filter was applied for corn silage data.

Combining Spatial and Temporal Corn Silage Yield Variability for Management Zone Development

“…P recision agriculture has the potential to increase crop yield and reduce the agricultural environmental footprint by applying precise inputs that meet the needs of the crops in each subsection of a field. Technology is advancing with improvements in the collection and processing of yield monitor data (Kharel et al, 2019; Khosla and Flynn, 2008), imagery from active crop and soil sensors (Li et al, 2014; Solari et al, 2008; Tagarakis and Ketterings, 2017), data from sensors mounted on planes (Cilia et al, 2014; Maresma et al, 2018; Scharf and Lory, 2002; Sripada et al, 2005), and data from unmanned aerial vehicles or satellites (Bausch et al, 2008; Bausch and Khosla, 2010; Maresma et al, 2016; Sakamoto et al, 2013), as well as advances in capturing and prediction of weather over time. The various layers of information each have their own spatial and temporal resolution, requiring careful evaluation of information in addition to development of better strategies to integrate and use the information for decision‐making at the farm, field and within‐field scales.…”

In‐Field Spatial Variability of Corn Stalk Nitrate Test Results

“…Yield map interpretation has often focused on identifying generalized zones of low, medium, or high, although more zones can be generated (Schepers et al, 2004). Once yield monitor data are properly cleaned of errors (Kharel et al, 2019b), single‐year yield maps can be created for each field on a farm. Yield maps can give guidance for CSNT sampling in future years, highlighting areas of similar yield, especially when multiple years of yield data can be used to derive yield stability zones (Long and Ketterings, 2016; Kharel et al, 2019a).…”

Impact of headland area on whole field and farm corn silage and grain yield

“…(2018), based on research in New York by Kharel et al. (2019b). Briefly, appropriate filter settings (flow delay, moisture delay, start, and end pass delay) for the farm, year, and equipment/operator were determined based on settings from 10 large fields with distinct features.…”