Technical Summary: 

Mississippi currently ranks third in the United States for sweetpotato production.  In 2001, production increased from 14,500 to 16,500 acres.  US No.1 grade sweetpotatoes are at least eight times more profitable than other potato grades (Anonymous 1998).  Therefore, research is currently focused on identifying factors that affect the spatial variability of sweetpotato yield grades.  Directed sampling points will be selected based on US No. 1 sweetpotato yield or bare soil images from fields located in North Mississippi (hill country) and in the delta of Mississippi.  Multispectral data will be collected from an aerial platform of conventional and unmanned aircraft.  Hyperspectral data will also be collected using a hand-held spectroradiometer.  Ground-truthed data collection throughout the growing season will include soil properties (chemical properties, texture, compaction, and moisture), tissue samples, estimated leaf chlorophyll level, percent ground cover (crop canopy, bare ground, weed species, and residue), insect identification and population, and sweetpotato grade yield.  Results from this research will indicate relations of ground-truthed parameters with spectral measurements to detect factors that influence the spatial variability sweetpotato yield.  Once these factors have been determined, site-specific strategies can be managed to maximize economic benefits.  An optical yield monitor developed at Mississippi State University will be modified and field-tested.  Hand graded samples will be used to validate the monitors accuracy in determining total yield and grade proportions.  

Objectives: 

1. Use directed grid sampling and statistical analyses to map soil characteristics, plant characteristics, pests, and sweetpotato yield to determine potential parameters for site-specific management.
2. Identify localized areas (management zones) in seemingly homogeneous production fields reported to produce consistently high yields of high quality sweetpotato relative to adjacent less productive areas.
3. Conduct research trials to investigate factors that had an effect on sweetpotato yield in past research.
4.  Evaluate an optical yield monitor to determine sweetpotato yield grade.  

Additional Objectives:

(contingent on leveraged funding by Mississippi Space Commerce Initiative)

1. Evaluate the use of remotely sensed data collected by Air-O-Space International L.L.C. with unmanned airborne vehicle (UAV) technology to maximize profits in sweetpotato production.
2. Evaluate the use of remotely sensed data collected by GPS L.L.C. (subsidiary of EMC Inc.) with conventional aircraft to maximize profits in sweetpotato production.
3. Assess appropriate wavelengths for evaluation of sweetpotato biophysical characteristics using a spectroradiometer.

Procedures: 

Field research will be conducted at producer locations in North Mississippi (hill country) and in the delta of Mississippi.  A 15 acre field at the Danny Clark Farm in Northeast Chickasaw County (near Vardaman, MS) will be used to evaluate spatial variability in a field influenced by established grower production practices.  Soil at this location is a Falkner silt loam (fine-silty, siliceous, thermic Aquic Paleudalf), which is commonly used to produce sweetpotato. Research will also be conducted at Mr. Wardel Sanders farm (limited resource cooperative producer) in the newly developing sweetpotato production area in Bolivar County (near Mound Bayou, MS).  The delta location is approximately 10 ac in size and represents common cultivation practices in delta sweetpotato production.  Prior to transplanting, soil samples at six inch increments (0-18 in depth) will be collected based on previous years yield maps and bare soil images.  Sample locations will be georeferenced using differentially-corrected GPS.  Samples will be analyzed for extractable N, P, K, Ca, Mg, S, Fe, Zn, Al, Na, B, Mn and Cu; and cation exchange capacity, organic matter, pH, lime requirement, and particle size distribution.  After transplanting, ground-truted data will be collected at 30 to 40 days after transplant (DAT)(root initiation), 50 to 60 DAT (root expansion), and 90 to 100 DAT (full root size).  Soil compaction will be measured at soil sampling points at two inch depth increments from 0 to 18 inches using a digital cone penetrometer, leaf chlorophyll levels will be measured using a chlorophyll meter, and plant tissue will be analyzed to determine N, P, K, Ca, Mg, S, Fe, Zn, Al, Na, B, Mn and Cu.  Insects will be collected using a vacumn apparatus.  Percent ground cover will be visually assessed relative to crop cover, bare soil, weed identification, and residue.  Hyperspectral data will be collected using a hand-held spectroradiometer measurements will be collected to identify sensor system wavelengths that best describe sweetpotato biophysical characteristics.  The hyperspectral data will be visually examined using PV-WAVE software.  This specialized visualization technique will provide insight to potential wavelengths or range of wavelengths that may indicate sweetpotato plant healthiness. Since reflectance data will be collected throughout the growing season, we will determine if the range of wavelengths change relative to crop development stage.  Multispectral data will be acquisitioned with an unmanned aerial vehicle and conventional aircraft.  Sweetpotato will be harvested from 10 x 10 ft yield squares at each sample location, weighed, and graded by MAFES personnel.  The spatial variability of yield and quality will be compared with that of soil characteristics (physical and chemical), plant characteristics (nutrients), pests (weeds, diseases, insects), spectral data (aerial and hand-held collection), and elevation (collected using an RTK) using correlation, multiple regression, and comparisons of spatial structure to determine factors likely to respond to site-specific management.   

An optical yield monitor system will be evaluated for use in sweetpotato.  This system is uniquely different than optical systems developed for use in other agronomic crops such as cotton.  This system is currently being developed by the Agricultural and Biological Engineering Department at Mississippi State University.  Several modifications that are necessary to strengthen commercialization include; developing the ability to change the frequency of image capture with variations in conveyor belt speed, improving the lighting system to eliminate uneven lighting at image edges, and improving the algorithm so that it can distinguish sweetpotatoes that are very close to each other.  An extensive field test of the system is necessary.  This will involve the harvest and classification of many sweetpotatoes on a site-specific basis, and subsequent comparisons with yield monitor output.  In addition, a commercially available load cell yield monitor will be evaluated for compatibility with the experimental optical system. 

Mississippi State University Extension will be involved with coordinating research efforts with sweetpotato producers.  The principal investigator will seek advice and provide updates to the Geospatial Extention Specialist throughout the course of the research project.  In addition, research project activities will be coordinated through the Intellectual Property of Technology Licensing Office at the Mississippi State University.  

Justification: 

Because of its high value, sweetpotato has a greater potential to reap rewards from costly investments necessary for site-specific management than do crops of lesser value.  Sweetpotato variability both in total yield and proportion of the most valuable US No.1 grade roots is widespread.  Research to date has indicated that some of this variability may be due to manageable soil factors such as soil pH, P, K, Zn, Cu, and Mn, organic matter content, and compaction.  These and other important factors affecting sweetpotato production have demonstrated substantial spatial variability in soils representative of those used for the majority of sweetpotato production in the state. A yield and grade monitor is very important for the future of precision agriculture in sweetpotatoes.  Existing yield monitors (e.g., “HarvestMaster” HM 500 Yield Mapping System) for sweetpotato and other conveyor-harvested crops (e.g., “Irish” potato, sugar beet, onion) utilize load cells on the conveyor chain to measure the weight of the crop as it is carried from the soil to bins or a picking line.  This system is accurate to within two to five percent on light-textured soils under low-trash conditions (Anonymous, 1998b; and Campbell, 1998).  However, sweetpotato is grown on heavier soils (silt loams, silty clay loams) in Mississippi, and significant quantities of soil and vine trash are carried on the conveyor chain before roots are hand picked.  Thus, conveyor chain load cells as currently configured are virtually useless for purposes of monitoring yield, and they make no distinctions among grades. 

Literature Review: 

The variable responses of sweetpotato to an array of soil management parameters including macronutrients (N, P, K) and liming are the subject of vast literature. As with most crops, sweetpotato response to P and K fertilization usually depends on their soil availability (Jones et al., 1979), but in many cases, responses probably have gone undetected because of high variability (Bouwkamp, 1985).  In recent research at the Pontotoc Ridge-Flatwoods Branch Experiment Station, yield response to P on low-P soils was observed, as was response to K (Rodriguez, 1997).  On a variety of soils in Georgia, sweetpotato responded positively to Mn applications, while Mn toxicity was implicated in experiments in Louisiana using lime and sulfur to modify soil pH (Jones et al., 1977).  A positive response to Cu has been observed on some sandy soils (Bouwkamp, 1985).  There has been considerable research on soil spatial variability (Warrick et al., 1986; White and Zasoski, 1999) and on site-specific management of agronomic crops (Robert et al. 1996), but relatively little on horticultural crops.   Little or no research has targeted the factors affecting the proportion of US No. 1 yield to total yield, a crucial element of profitability.  Recent research in Mississippi on soils representative of those used for the majority of sweetpotato production indicated substantial spatial variability for P and K spanning all MSUES soil test classes, indicating a potential for site-specific variable rate fertilization.  In addition, the micronutrients Zn, Cu, and Mn were correlated with spatial variations in sweetpotato yield or grade proportion (White et al. 1998a,b). 

Losses to insects in sweetpotato fields are almost always recognized after it’s too late to do anything and pest identification is often circumstantial (Cuthbert, 1967).  Pests of this crop include – the grub complex  (white grubs, whitefringed beetles, cucumber beetles, and others), the black flea beetle, and wireworms below ground, and the Lepidoptera – mostly armyworm complex above ground.  In recent years cutworms have become a major problem both in the early season and on the developing roots later in the growing season.   

A large number of research projects have attempted to use visible color, shape, and or texture to classify various agricultural and food commodities with optical sensors and image-processing techniques.  For example, image-processing techniques have been used to improve cotton color measurement (Thomasson et al., 1997), to grade roses (Steinmetz, 1994), to measure the shape of corn kernels (Ni et al., 1997), and to detect nutritional disorders in lettuce (Hetzroni and Miles, 1994). LaBonte and Wright (1993) reported the only example of image analysis used with sweetpotato that was found in the literature.  It concerned detection of skinning injury.  An optical yield and grade monitoring system developed in this project has been reported (Wooten et al., 1999; Wooten et al., 2000; and Thomasson et al., 2000).  It was constructed to enable imaging of the conveyor on a sweetpotato harvester during harvest.  Computer algorithms were developed for detection of sweetpotatoes.  The reports further stated that manual image measurements of sweetpotatoes related well to mass and class grade. 

Current Research:   

In 2001, a 15 acre sweetpotato production field was mapped, and directed sampling locations determined from 2000 yield and nutrient maps.  Soil chemical and physical analyses were conducted on samples taken at each location.  Sweetpotato yield was determined by harvesting yield squares at soil sampling sites, and roots were graded.  Data were analyzed, maps developed, and research results reported in NMREC annual reports and at the Southern Association of Agricultural Scientists annual meeting.  Sweetpotato producers were informed of project activities at a field day and a survey is being conducted to identify cooperator fields with potential for site-specific management.  Refer to 2001 progress report attached to this proposal.  

Literature Cited: 

Anonymous.  1998a.  Terminal market prices: sweetpotato.  Internet address: http://mis.ifas.ufl.edu/%7emarket/fvprod/tsweetpotatoes.html.  Gainesville, FL: University of Florida. 

Anonymous.  1998b.  HM-500 specifications and HM-500 yield mapping system information packet.  Internet address: http://www.harvestmaster.com/hm5info.html.  Logan, UT: HarvestMaster Inc. 

Bouwkamp, J.C. 1985. Production requirements. p. 9-33. In J.C. Bouwkamp (ed.) Sweet potato products: a natural resource for the tropics.  CRC Press, Boca Raton, FL. 

Campbell, R.H.  1998.  Personal communication.  Marketing Communications Specialist.  Logan, UT: HarvestMaster Inc. 

Cuthbert, F. P., Jr., 1967.  Insects affecting Sweetpotato.  USDA  Agric. Handbook 329. 28p. 

Jones, L.G., R.J. Constantin, J.M. Cannon, W.J. Martin, and T.P. Hernandez.  1977. Effects of soil amendment and fertilizer applications on sweet potato growth, production, and quality.  Louisiana Agric. Exp. Stn. Bull. 704. 

Jones, L.G. R.J. Constantin, and T.P. Hernandez. 1979. The response of sweetpotatoes to fertilizer phosphorus and potassium as related to levels of these elements available in the soil.  Louisiana Agric. Exp. Stn. Bull. 722. 

LaBonte, D. R. and M. E. Wright.  1993.  Image analysis quantifies reduction in sweetpotato skinning injury by preharvest canopy removal.  Hort. Sci. 28(12). 

Ni, B., M. R. Paulsen, and J. F. Reid.  1997.  Corn kernel crown shape identification using image processing.  Trans. ASAE 40(3):833-838. 

Robert, P.C., R.H. Rust, and W.E. Larson (ed.) 1996. Proceedings of the Third International Conference on Precision Agriculture, June 23-26, Minneapolis, MN.  ASA, CSSA, SSSA, Madison, WI. 

Rodriguez, G.A. 1997. Response of >Beauregard= sweetpotato to rates and time of application of fertilizers (root yield).  Ph.D. diss. Mississippi State University, Mississippi State, MS. Diss. Abst. AAG9801499; DAI 58‑07B:3377. 

Steinmetz, V., M.J. Delwiche, D.K. Giles and R. Evans.  1994.  Sorting cut roses with machine vision. Trans. ASAE. 37(4):1347-1353. 

Thomasson, J. A., S. A. Shearer, and R. K. Byler.  1997.  Cotton color measurement improvement with image processing. ASAE paper No. 971016.  St. Joseph, MI: ASAE. 

Thomasson, J. A., J. R. Wooten, S. Gogineni, and P. R. Thompson.  2000.  Sweetpotato yield monitor based on optical imaging techniques.  North Miss. Res. & Ext. Center, Ann. Report 2000: Miss. State Univ. Inf. Bull. Verona, Miss.: North Miss. Res. & Ext. Center. 

Warrick, A.W., D.E. Meyers, and D.R. Nielsen. 1986. Geostatistical methods applied to soil science. p. 53-82. In A. Klute (ed.) Methods of soil analysis, part 1. ASA and SSSA, Madison, WI. 

White, J.G., W.B. Burdine, Jr., P.G. Thompson, and J.L. Main. 1998. Effect of soil spatial variability on sweetpotato yield and quality.  p. 284-285. In C.P. Bagley (ed.) 1997 Annual Research Report, North Mississippi Research & Extension Center.  MAFES Information Bulletin 336. 

White, J.G., P.G. Thompson, and W.B. Burdine, Jr. 1998. Effects of soil spatial variability on sweetpotato. Agronomy Abst., p. 315. 

White, J.G. and R.J. Zasoski. 1999. Mapping soil micronutrients.  Field Crops Research (in press). 

Wooten, J.R., J. A. Thomasson, J. G. White, and P. G. Thompson.  1999.  Sweetpotato yield monitor based on optical imaging techniques.  North Miss. Res. & Ext. Center, Ann. Report 1999: Miss. State Univ. Inf. Bull. 365.  Verona, Miss.: North Miss. Res. & Ext. Center. 

Wooten, J. R., J. A. Thomasson, J. G. White, and P. R. Thompson.  2000.  Yield and quality monitor for sweetpotato with machine vision.  ASAE Paper No. 001123.  St. Joseph, Mich.: ASAE.

Back to Top

Back to Menu