Abstract Data‐driven models used for predicting soil temperature usually have increasing errors with increasing depth. By exploring the integration of knowledge‐based and machine learning approaches, this study used a novel transformation of meteorological variables to increase prediction accuracy of soil temperature with increasing soil depth. Using datasets for two soil textures (silty clay loam and loamy coarse sand) at two locations with different climates, predictive models were developed for five depths (5, 10, 20, 50, and 100 cm) as a function of meteorological features using an adaptive neuro‐fuzzy inference system (ANFIS). For each depth, soil temperature was predicted with nontransformation (NT), autocorrelation (AC), moving average (MA), and a combination of transformations (NT‐ACMA) of meteorological features. Across all depths, the predictive accuracy of NT‐ACMA models was significantly higher than that of NT, AC, and MA models for both soil textures (R2 = .99, RMSE = 1 °C for silty clay loam; R2 = .99, RMSE = 1.2 °C for loamy coarse sand), with increasing prediction accuracy as soil depth increases. Results for different soil textures and climates in 11 locations across the contiguous United States show that, except for 100‐cm depth, there seems to be significant positive linear relationships between the best moving average and the solar inclination. This makes our NT‐ACMA technique transferable to any location in the contiguous United States irrespective of the location, climate, and soil texture. We conclude that integrating lag times and moving averages of meteorological features can lead to better prediction of soil temperature at different soil depths.

Strawberries are only seasonally available in Nebraska (NE). If an affordable heated structure can be designed, then opportunity exists to increase on-farm income by producing strawberries off-season during high-value market periods. A series of university greenhouse trials were conducted from 2010 to 2012. The varieties Evie-2 and Seascape were identified as being most productive under a low technology growing scheme. A new research project, which ran from fall to late spring (2013-2014), was designed to determine if this production scheme would translate to a commercial grower. Varieties Seascape and Evie-2 (each at two grades indicated by +) and San Andreas plants were grown simultaneously at the university research greenhouse and a cooperative specialty crop grower greenhouse, similar structures and production timelines. Top performers for both experiment locations were 'Evie-2MODIFIER LETTER PRIME, and 'Evie-2+' plants, with average harvest berry weights (marketable) per plant (pp) of over 0.454 kg, and 'Seascape' with 0.390 kg. 'Seascape+' plants performed well at the cooperator location (0.399 kg pp), but not at the university location. 'San Andreas' plants performed well at the university location with 0.454 kg pp but not at the cooperator location. Productivity was greatest during the winter-spring season, which accounted for more than 82% of the total berry mass harvested at the cooperator site and 88% at the university site. The results support the concept that the growing system used within a controlled university research setting is representative of the crop productivity a specialty crop grower might expect, particularly when strawberries are grown in late winter and early spring. However, securing low fuel costs, eliminating delivery and having a secondary market for culls is key for profitability.

Environmental and soil conditions affect plant productivity. Determining and understanding how plants respond to moisture stress may be considered in different ways, but often not easily done. This study focused on measuring canopy and individual leaf temperature response times of selected horticultural and agricultural crops, treated to very dry (moisture stress) and well-watered conditions, and subjected to an incident step radiation change at the top of the canopy, in a controlled environment at a constant air temperature. A light-emitting-diode (LED), grow light system, delivering approximately 400 µEinstein s-1 m2 of photosynthetically-active-radiation (PAR) was turned on and off to produce light and dark radiation step functions for the tests. Canopy and leaf temperature time series were measured using both an infrared thermometer and a FLIR thermal imaging camera, over a 10-minute elapsed time, respectively. Temperatures of a selected single, unobstructed leaf at the top of the canopy were analyzed with visual and thermal imaging processing. Response data were modeled using mathematical transfer functions and model discovery methods. Both canopy and leaf temperature-time series for dry plants were found to be a first-order response. Well-watered plants were first order followed by second order response phase, possibly indicating stomatal activity. Thermal response times to a step increase of PAR (lights-on), were higher for well-watered plants than for water-deprived plants. (It took longer for the leaves of well-watered plants to heat up). The thermal response times, to a step decrease of PAR (lights off), were lower for well-watered plants than for water -deprived plants. (It took less time for the leaves of well-watered plants to cool off). These results show that leaf thermal response may be used as an indicator of plant water status, and as a potential instrumentation approach.

In agricultural areas where groundwater is the main water supply for irrigation, forecasts of the water table are key to optimal water management. However, water management can be constrained by semi-seasonal to seasonal forecast. The objective is to create an ensemble of water table one- to five-month lead-time forecasts based on multiple data-driven models (DDMs). We hypothesize that data-driven modeling capabilities (e.g., random forests, support vector machines, artificial neural networks (ANNs), extreme learning machines, and genetic programming) will improve naive and autoregressive simulations of groundwater tables. An input variable selection method was used to carry the maximum information in the input (precipitation, crop water demand, changes in groundwater table, snowmelt, and evapotranspiration) and output relationship for the DDMs. Five DDMs were implemented to forecast groundwater tables in an unconfined aquifer in the Northern High Plains (Nebraska, USA). Root mean squared error (RMSE) and Nash-Sutcliffe efficiency index were used to evaluate the skill of the model and three hydrologic regimes were determined statistically as high, mid, and low groundwater table levels. Additionally, varying storage regimes were used to construct rising and falling limbs in the tested well. Results show that all models outperform the baseline for all the lead times, ANNs being the best of all.

Introduction Costs of transportation and food safety concerns have spawned an increase in public support of locally grown vegetables and fruit. With this in mind, a two-phase investigation was planned with the goal of combining low start up costs for sustainable greenhouse production with selection of strawberry cultivars that would provide the greatest number and largest size of berries. Additionally, berries from each cultivar will be analyzed for their beneficial nutraceutical properties to determine if there is a difference among cultivars.

This paper presents a discussion and comparison of the methods used to process images retrieved from a dual sensor system deployed on an unmanned aircraft for agricultural use. A challenge to the broader use of unmanned aircraft systems stems from the FAA imposed maximum flying altitude threshold of 400 feet (120 meters) above ground level. However, additional FAA permission is needed for flights conducted at altitudes greater than 400 feet above ground level. While this low altitude threshold provides an opportunity to collect high-resolution spatial information, it also requires that remotely sensed information be collected as discrete individual sub images. These individual images are then stitched or mosaicked together as a larger image to depict an agricultural crop field. This process requires the use of sensor characteristics along with the amount of sensor overlap during the flight phase. Image processing algorithms were tested to assemble a multitude of separate images into a single image, depicting the field area or target interest. Sensor configuration, geotagged file structure, flight planning, and approaches to processing the remotely sensed information collected during a typical flight operation are presented.

Basil (Ocimum basilicum) is a high value crop, currently grown in the field and greenhouses in Nebraska. Winter-time, greenhouse studies were conducted during 2015 and 2016, focusing on cultivars of basil grown on a Cap MAT II® system with various levels of fertilizer application. The goal was to select high value cultivars that could be grown in Nebraska greenhouses. The studies used water content, electrical conductivity, photosynthetically active radiation (PAR), and relative humidity, air and soil media temperature sensors. Greenhouse systems can be very complex, even though controlled by mechanical heating and cooling. Uncertain or ambiguous environmental and plant growth factors can occur, where growers need to plan, adapt, and react appropriately. Plant harvest weights and electronic sensor data was recorded over time and used for training and internally validating fuzzy logic inference and classification models. Studies showed that GENFIS2 ‘subtractive clustering’ of data, prior to ANFIS training, resulted in good correlations for predicted growth (R2 > 0.85), with small numbers of effective rules and membership functions. Cross-validation and internal validation studies also showed good correlations (R2 > 0.85). Decisions on basil cultivar selection and forecasting as to how quickly a basil crop will reach marketable size will help growers to know when to harvest, for optimal yield and predictable quantity of essential oils. If one can predict reliably how much essential oil will be produced, then the methods and resultant products can be proposed for USP or FDA approval. Currently, most plant medicinal and herbal oils and other supplements vary too widely in composition for approval. The use of fuzzy set theory could be a useful mathematical tool for plant and horticultural production studies.

A fuzzy logic feedback control system was developed for process monitoring and feeding control in fed-batch enzymatic hydrolysis of a lignocellulosic biomass, dilute acid-pretreated corn stover. Digested glucose from hydrolysis reaction was assigned as input while doser feeding time and speed of pretreated biomass were responses from fuzzy logic control system. Membership functions for these three variables and rule-base were created based on batch hydrolysis data. The system response was first tested in LabVIEW environment then the performance was evaluated through real-time hydrolysis reaction. The feeding operations were determined timely by fuzzy logic control system and efficient responses were shown to plateau phases during hydrolysis. Feeding of proper amount of cellulose and maintaining solids content was well balanced. Fuzzy logic proved to be a robust and effective online feeding control tool for fed-batch enzymatic hydrolysis.

The area cultivated under conservation tillage practices such as no-till and minimal tillage has recently increased in Midwestern states, including Nebraska. This increase, consequently, resulted in changes in some of the impacts of cropping systems on soil, such as enhancing soil and water quality, improving soil structure and infiltration, increasing water use efficiency, and promoting carbon sequestration. However, there are no methods currently available to quantify the percent crop residue cover (CRC) and the area under conservation tillage for maize and soybean at large scales on a continuous basis. This research used Landsat-7 (ETM+) and Landsat-8 (OLI) satellite data to evaluate six tillage indices [normalized difference tillage index (NDTI), normalized difference index 7 (NDI7), normalized difference index 5 (NDI5), normalized difference senescent vegetative index (NDSVI), modified CRC (ModCRC), and simple tillage index (STI)] to map CRC in eight counties in south central Nebraska. A linear regression CRC model showed that NDTI performed well in differentiating the CRC for different tillage practices at large scales, with a coefficient of determination (R-2) of 0.62, 0.68, 0.78, and 0.07 for 25 March, 18 April, 28 May, and 6 June 2013 Landsat images, respectively. A minimum NDTI method was then used to spatially map the CRC on a regional scale by considering the timing of planting and tillage implementation. The measured CRC data were divided into training (calibration) and testing (validation) datasets. A CRC model was developed using the training dataset between minimum NDTI and measured CRC with an R-2 of 0.89 (RMSD = 10.63%). A 3 x 3 matrix showed an overall accuracy of 0.90 with a kappa coefficient of 0.89. About 26% of the maize area and 15% of the soybean area had more than 70% CRC in south central Nebraska. This research and the procedures presented illustrate that multi-spectral Landsat images can be used to estimate and map CRC (error within 10.6%) on a regional scale and continuous basis using locally developed tillage practice versus crop residue algorithms. Further research is needed to incorporate soil and residue moisture content into the CRC versus tillage index to enhance the accuracy of the models for estimating CRC.

Modeling the extrusion process is complex because of the many confounding variables and the dynamic properties of the materials subjected to heat, shear and pressure. Efforts have been made to predict various system and product properties using a flow-modeling phenomenological approach. However, because of the many assumptions and the complexities involved, these models become often impractical for application. In this study, dimensional analysis was used to determine significant dimensionless parameters for the inputs and outputs of a model constrained only by the available experimental data. Thereafter, a rule-based FIS was used, instead of a conventional exponential model, for the prediction of output dimensionless parameters. Optimization or selection of subtractive cluster radii for FIS was achieved using a genetic algorithm. Data were obtained from 4x3x3x2 experimental design (16, 20, 24 and 28% moisture contents; 80, 120 and 160rpm screw speeds; 3, 4 and 5mm nozzle diameters and 120 and 140C barrel temperatures) and 3x2 (80, 120 and 160rpm screw speeds; 120 and 140C barrel temperatures for a 4mm nozzle diameter and 26% moisture content). These factorial design experiments were conducted using a laboratory twin-screw extruder. After training, the FIS captured the process trend based on the experimental data. Correlation coefficient (r(2)) values were found higher than those obtained from a linear regression model.Practical ApplicationsExtrusion process is widely used to produce human foods, dog foods and polymers. Prediction of extrusion performance and scale-up of the process are very challenging because of complex and dynamic characteristics of biological materials when shear and heat are applied simultaneously. This study presents a novel modeling method to predict dimensionless parameters involving MRT and torque experienced by the extruder. The model was validated with data collected from extrusion of corn starch. This model can be used by producers to predict change in energy consumption and MRT and facilitate scale-up when feed moisture content, die diameter, screw speed and barrel temperature are changed. The modeling methodology can be further expanded to predict other extrusion performance parameters and for other materials by training and validating the model with new data set.

A prototype on-line hyperspectral imaging system (lambda = 400-1000 nm) was developed and used to acquire images of exposed ribeye muscle on hanging beef carcasses (n = 274) at 2-day postmortem in a commercial beef packing plant. After image acquisition, a strip steak was cut from each carcass and vacuum packaged. After aging for 14 days, the steaks were cooked and Warner-Bratzler shear force values were collected as a measure of tenderness. Four different principal component analysis-based dimensionality reduction methods were implemented to reduce information redundancy in beef hyperspectral images. Textural features extracted from the 2-day hyperspectral images were modeled using Fisher's linear discriminant (FLD), support vector machines (SVM), and decision tree (DT) models to predict 14-day aged, cooked beef tenderness. Based on a true validation procedure using 101 samples, the FLD model yielded a tender certification accuracy of 86.7%. In addition, wavelengths corresponding to myoglobin and its derivatives (541, 577, and 635 nm), beef aging (541, 577, 635, 756, and 980 nm), protein (910 nm), fat (928 nm), and water (739, 756, and 988 rim) were identified. (C) 2015 Elsevier Ltd. All rights reserved.

This paper presents measures of performance for a fixed wing unmanned aircraft system (UAS) serving as a sensor platform for production agriculture mission profiles. UAS deployment in production agriculture require that onboard navigation systems guide the aircraft according to predefined flight plans for accurate sensor placement. The objective of this research was to evaluate fixed wing UAS performance measures for predefined flight paths, indicated air speed (IAS), and altitude (MSL). A transit pattern was selected to test the autonomous navigation system, featuring GPS waypoints and rectangular loops, with each complete loop shifting incrementally across a research field. Fourteen flights covering areas of 16.2, 32.4, or 64.7 ha were conducted over a two-month period with variable environmental conditions. Two 16.2 ha and two 32.4 ha flights were selected for analysis. Root mean squared error (RMSE) results for the flight path data were between 3.88 m and 14.05 m, with a maximum difference of 52.8 m. IAS analysis found that the average was slightly higher than expected, with an RMSE between 0.8 m s-1 and 1.74 m s-1. The altitude hold capability (MSL) was close to its predefined value; however, an RMSE range between 2.04 m and 3.02 m suggested deviations from the target MSL for different flights. A tolerance limit analysis confirmed these findings. Results suggested that measures of different flight characteristics can be used to define UAS performance as a sensor platform. Research results demonstrated both great promise, and an associated potential for improvement, in currently available unmanned aircraft with autonomous navigation systems deployed for sensor placement over large parcels of agricultural land.

Adaptive neuro-fuzzy inference system (ANFIS) models were developed to predict and evaluate the impact of temporal and spatial precipitation amounts on rainfed maize and soybean yields in Nebraska for individual years. Precipitation data from 1996 to 2012 were divided into three datasets: training, checking, and testing. The testing dataset comprised all counties with reported grain yields for the years 2000, 2005, and 2010. The statewide average seasonal (May 1 to Sept. 30) precipitation for 2000, 2005, and 2010 was 270, 358, and 482 mm, respectively, and the long-term (1996 to 2012) average was 361 mm. The ANFIS models were tested by adding individual cumulative monthly precipitation amounts as model inputs. ANFIS was unable to accurately predict maize and soybean yields with only May precipitation, with R-2 and root mean squared difference (RMSD) values, respectively, of 0.01 and 2.48 Mg ha(-1) for maize and 0.00 and 0.75 Mg ha(-1) for soybean. Model performance improved as monthly precipitation amounts were added to the maize and soybean ANFIS models, with the ANFIS model Y = f(May, June, July, Aug., Sept.) performing best for the pooled testing years (2000, 2005, and 2010), with R-2 and RMSD values, respectively, of 0.57 and 1.64 Mg ha(-1) for maize and 0.47 and 0.56 Mg ha(-1) for soybean. Longitude was added as a model input to implicitly account for differences in climatic variables (e.g., solar radiation) that impact grain yield across Nebraska. The inclusion of longitude into the time-step ANFIS models improved the performance for both individual and pooled testing years, with greater improvement for maize than soybean. The potential effects of dormant (Oct. 1 to April 30), seasonal, and annual (Oct. 1 to Sept. 30) precipitation amounts on grain yields were also investigated The best performing model of the pooled testing data for both maize and soybean was Y = f(Lng, Annual), with R-2 and RMSD values, respectively, of 0.75 and 1.22 Mg ha(-1) for maize and 0.67 and 0.44 Mg ha(-1) for soybean. Furthermore, prior to the growing season, the Y = f(Lng, Dormant) model explained 65% and 61% of the variability in maize and soybean grain yield, respectively, for the pooled testing data. This study demonstrated that ANFIS can be used to predict maize and soybean grain yields in Nebraska with reasonable accuracy from precipitation data.

This research documents performance measures for an autonomous navigation system serving as the controller for an unmanned aircraft system (UAS) that is flying a mission profile suited for production agriculture. The aircraft is a single engine, fixed wing type with high efficiency characteristics allowing long flight duration with minimal energy consumption. The combination of autonomous navigation and efficient aircraft will allow the placement of aerial sensor systems over considerable expanses of land and large agricultural production operations, such as those found in the Great Plains and Midwest. Deployment of UAS in production agriculture will require that onboard navigation systems guide the aircraft according to predefined flight plans, so that sensors can be placed at the appropriate location in both space and time. Research under this project involved defining a flight plan that would allow for sequential transit of the fixed wing UAS across a rectangular field, such that optical data could be collected for subsequent analysis. A transit pattern involving rectangular loops, with each complete loop shifting incrementally across the field, was selected to test the navigation system. This paper presents results of test flights to explore potential measures of performance for path-tracking capability based on analysis of the flight computer data. In summary, opening of the National Air Space to unmanned aircraft systems will be a “game changer” for the agricultural industry. UAS will offer an unparalleled opportunity to place crop and soil sensors, robotics, and advanced information systems at more timely and desired field locations for increasing production and improving efficiency of agricultural operations. The results of this research show great promise for currently available autonomous navigation systems guiding single engine, fixed wing unmanned aircraft for placement of sensors at specific locations over large parcels of agricultural land.

It is important for a greenhouse strawberry grower to know that their capillary mat (CapMat™) fertigation system is working correctly and that plants are receiving the correct amounts of water and fertilizer. Pots with soilless mix are not expected to hold more than 70% water on a volumetric basis. Pots with less than 40% water content continuously are not supplied enough water and nutrients to the plants. Typically, pots located near the manifold distribution system get a little more water than those at the other locations, but water use will really vary according to the factors listed above as well as environmental parameters, but should not vary more than 20%. Fertigation is based on the drip tape distribution system, the media density of individual pots, and the spatial energy distribution within the greenhouse. To understand how these factors interact, pot moisture, media temperature, and electrical conductivity were spot checked with a relatively new commercial sensor and also monitored continuously along with greenhouse temperature, humidity, and photosynthetically active radiation (PAR) using a data logger system. We found that the variance in pot medium moisture and fertilizer was as expected as were fluctuations in air and mix temperatures. Calibrated commercial electrical conductivity and soil moisture sensors for measuring pot moisture and/or electric conductivity were reliable. Having this data may be a key to determining why plants in the UNL greenhouse produced more marketable fruit than plants in the commercial house.

Materials and Methods A heated high tunnel 25’ by 75’ was constructed and equipped with an overhead furnace, three low cost benches and a sustainable capillary mat system (Figure 1). Holes were cut in the plastic to allow the pots to sit directly on the mat while minimizing algae growth (Figure 2). The double polyethylene greenhouse at UNL was about the same size with similar features except the heat was funneled through poly tubes underneath the benches and only 1/3 of the two benches were used. The experimental design for the university study was a RCBD with 5 cultivars (Seascape, Seascape+, Evie-2, Evie-2+, and San Andreas) and twelve replications across two benches. Data included: number and pounds of berries per cultivar per week, total water and natural gas usage, monitoring and recording of growing conditions to include incident light level, soil moisture capacity, light reflectance from leaves and relative humidity of the greenhouse. The experimental design for the grower production experiment was also a RCBD with 3 benches (replications) and the same 5 cultivars (treatments). Data included all of the above as well as associated building, growing and marketing costs including sale price.

Undesirable plants are negatively impacting many ecosystems and causing significant losses both economically and environmentally. Examples of losses include altered stream flow in riparian areas, increased frequency of fires in rangelands and forests, fewer habitats with high species diversity and increased management for aesthetics and yields in natural areas and production systems, respectively. The services provided by the terrestrial systems in which undesirable plants have invaded or are established have yet to be completely quantified, but estimates are in the billions of dollars. An improved plant species identification program is being developed and is currently being tested as potential "ap" for identifying selected plant species found in Nebraska and fields of the Great Plains. The identification process takes place through an acquired digital leaf image. The identification algorithm performs a classical analysis using a Fast Fourier Transform but also concentrates on detailed leaf venation and surface texture. Leaf species samples were initially supplied from plants grown in an environmental chamber, but will extend to spring/summer/fall field specimens. Plants will eventually include various herbaceous species that are native and non-native (e.g., velvet leaf, pigweed, downy brome, phragmites, common reed, leafy spurge, cheat grass). Specimens were imaged using a high-resolution digital camera and a special portable lighting system. The impact from the development of a real-time plant species identification system (APSIS) to identify desired and undesirable plants would revolutionize the way plant populations are studied by researchers, ecologists, and monitored by crop and land managers.

In cropping systems, integrated weed management is based on diversification. Rather than relying solely on one or two herbicides, a multiplicity of weed control strategies is employed. Yet, integrated weed management as currently practiced is far from integrated; every weed is still managed the same regardless of location or season. The recent development of precision application technology is now allowing for smaller treatment units by making applications according to site-specific demands. The automated systems of the future will have sensor and computer technologies that first categorize each and every plant in the field as either weed or crop and then identify the species of weed. Following identification, multiple weed control tools located on a single platform are applied at micro-rates to individual plants based on their biology. For example, if the system identified a weed resistant to Roundup, it could be spritzed with a different herbicide or nipped with an onboard cutter or singed with a burst of flame. This system and others like it will be capable of targeting different weed-killing tools to specific weeds. This chapter will discuss the challenges and tools of the future.