RICE CULTURE An Analysis of Aerial Imagery and Yield Data Collection as Management Tools in Rice Production C.W. Jayroe, W.H. Baker, and W.H. Robertson ABSTRACT Early estimates of yield and correcting yield-limiting factors are now obtainable goals with the use of multispectral aerial imagery. Through the use of various image enhancements in a Geographical Information System (GIS), plant canopy can be analyzed for determining plant health. These multispectral images tend to positively correlate with data gathered by yield-monitoring systems. In this study, two rice production fields were observed. The objective was to relate multispectral aerial imagery to final yield. The images of the two fields did positively correlate with the application of various image enhancement techniques. The results of this study reveal the possibility of correcting yield-limiting factors during the season with the establishment of management zones. INTRODUCTION The use and implementation of advanced remote-sensing techniques, the science and art of obtaining information about an object or area through analysis of data acquired by a device that is not in contact with the object or area (Lillesand, 1994), such as multispectral aerial imagery and data gathered by yield monitors are proving to be an asset in many production management decisions. The application of these methods in GIS extends the capabilities of these innovative techniques by allowing users to overlay separate map layers. In agriculture, monitoring of crop growth and development and early estimates of final yield are of particular interest (Senay, 1998). With the use of infrared aerial 270
B.R. Wells Rice Research Studies 2002 photography, plant physiological and morphological differences can be distinguished in order to determine plant disease or to access plant health in relation to other areas of the field (Johannsen and Sanders, 1982). Also of great importance is the establishment of management zones, a portion of a field that expresses a homogeneous combination of yield-limiting factors for which a single rate of a specific crop input is appropriate (Doerge, 2002). Through analysis of remotely sensed and ground-truthed data in a GIS, much can be learned about a field. Previous studies have found a strong correlation between yield and data obtained by remotely sensed methods. The ability to accurately predict yield of field crops allows producers to make timely decisions with respect to crop management (Ma et al., 2001). The objective of this study was to determine if a correlation between multispectral aerial imagery and data gathered by a yield monitor exists. This would be a useful discovery because remote sensing has the potential to allow producers the opportunity to see yield variations in their production fields (Vellidis et al., 1999). These variations can be mapped, using the multispectral image, in a more timely manner than producers and consultants have ever been able to utilize. PROCEDURES On 28 July 2002 multispectral aerial images were obtained with a Duncan Tech camera, during the growing season, at a resolution of approximately one meter. The camera has three bands with one of those acquiring images at a near-infrared frequency. Since plant canopy reflects at a very high rate of near-infrared frequency, this band was believed to be the most useful in predicting yield variances. A winter image was also analyzed in order to capture a mostly bare soil view. This image was a color-infrared digital otrthoquarterquad (DOQ) photograph obtained from the University of Arkansas s Center for Advanced Spatial Technologies (CAST) website. Color-infrared imagery is useful in identifying those areas of a field that are more lush and healthy (Sfiligoj, 2002). The DOQ image was not only used for visual observation but was also used in rectifying and georeferencing the multispectral image in ESRI s ArcView 3.2a geographical information software. The image was then enhanced by performing a normalized difference vegetation index (NDVI) and later classified into acres in order to gain a better understanding of the variability across the entire field also using ArcView 3.2a. These classifications were later compared to yield data. Yield-monitor data were also gathered when the rice was harvested in 2002, with an Ag Leader system. A GPS was used on the yield-monitoring system to record latitude and longitude coordinates. The yield information was exported after harvest with SMS Basic, Ag Leader s software, as a shape file. The shape file was then added into ArcView 3.2a and interpolated so that it could be overlaid along with other data layers. 271
AAES Research Series 504 RESULTS AND DISCUSSION A strong correlation was found between the data gathered by the yield monitor and the multispectral image. The two data layers compared closely when the image was inversly stretched and a histogram equalize was performed in band 2. The classification of the multispectral image also proved to correlate strongly with the yield data. Further investigation into this method will be needed in order to determine if it can be used to accurately predict yield. It does however show potential in distinguishing and accessing variation by grouping similar areas together in acres. In field one, the strongest correlation was found between the multispectral image in band 2 (Fig. 1) and yield data (Fig. 2). The image identified key patterns that the yield monitor data also displayed. This image was relatively cloud and shadow free so the classification was able to accurately represent the vegetative canopy (Fig. 3). The second field however, was hindered by clouds and shadows. At least two portions of the field were not visible due to the interference. However the areas that were cloud free also correlated well with yield (Fig. 5) in band 2 with an inverse stretch and a histogram equalize (Fig. 4). The classified image (Fig. 6) also correlated strongly, aside from the areas covered by clouds. In general, remotely sensed data has the capability to offer timely information that can be used as a tool in many management decisions. The opportunity to view a field in production at various stages of development is one of the most advantageous aspects of techniques such as aerial- and satellite-based sensing. These methods offer the chance to view lower production areas, irrigation efficiency, diseased areas, and insect infestation damage during any phase of development needed. Most importantly, remote sensing has the potential to provide vital information in a timely manner in order to correct any problems on time. In short, the scope of digital-image processing and its application in spatial analysis is virtually unlimited (Lillesand, 1994). SIGNIFICANCE OF FINDINGS Scheduling flights strategically was the most important lesson learned. Flights should be carefully planned with time of day and cloud condition in mind. The time should be somewhere between 10 am and 2 pm in order to benefit from the most intense energy of the sun. Also, scheduling more than one flight during the season would be advantageous. That would allow a better monitoring of rice throughout various stages of growth. Lastly, all images should be acquired before the rice seed head has emerged since at this stage very little can be done to correct problems in the field. It is also difficult to obtain a good representation of plant health due to the reflectance interference from the seed head. ACKNOWLEDGMENTS The authors wish to thank our cooperators; John Andrews, Theodore Andrews, and Kevin Hoke for all of their help and patience. We would also like to thank Johnny 272
B.R. Wells Rice Research Studies 2002 Williams, Global Positioning Solutions Inc., for his cooperation with all of the aerial imagery. Thirdly, we appreciate Melvin Wamock, Melton Brothers Inc., and John Deer for their generous assistance in supplying the GPS during harvest. LITERATURE CITED Doerge, T. 2002. Defining management zones for precision farming, Part 7: Management zone delineation in the future. http://www.pioneer.com/usa/technology/ management%5fzone%5ffuture.htm (Verified April, 2002). Johannsen, C.J. and J.L. Sanders. 1982. Remote Sensing for Resource Management. Soil Conservation Society of America, Ankeny, Iowa. Lillesand, T.M. and R.W., Kiefer. 1994. Remote Sensing and Image Interpretation, 3 rd edition. John Wiley & Sons, Inc., New York. Ma, B., L. Dwyer, C. Costa, E. Cober, and M. Morrison. 2001. Early prediction of soybean yield from canopy reflectance measurements. Agronomy J. 93:1227-1234. Senay, G.B., A.D. Ward, J.G. Lyon, N.R. Fausey, and S.E. Nokes. 1998. Manipulation of high spatial resolution aircraft remote sensing data for use in site-specific farming. Transactions of the ASAE 41(2):489-495. Sfiligoj, E. 2002. Not so remote. Crop Life, May 2002. pp. 38-41. Vellidis, G., D. Thomas, T. Wells, and C. Kvien. 1999. Cotton yield maps created from aerial yield maps. 1999 ASAE Annual International Meeting Toronto, Ontario, Canada. ASAE, St. Joseph, MI. 273
AAES Research Series 504 Fig. 1. Multispectral image of field 1, enhanced in band 2 with a histogram equalized and an inverse stretch. 274
B.R. Wells Rice Research Studies 2002 Fig. 2. Yield monitor data from field 1, mapped and grouped into yield classes. 275
AAES Research Series 504 Fig. 3. Classified multispectral image of field 1 that displays vegetative difference. 276
B.R. Wells Rice Research Studies 2002 Fig. 4. Multispectral image of field 2, enhanced in band 2 with a histogram equalized and an inverse stretch. 277
AAES Research Series 504 Fig. 5. Yield monitor data of field 2, mapped and grouped into yield classes. 278
B.R. Wells Rice Research Studies 2002 Fig. 6. Classified multispectral image of field 2 that displays vegetative difference. 279