Scott D. Noble

Site-specific weed management: sensing requirements— what do we need to see?

Abstract Automated detection and identification of weeds in crop fields is the greatest obstacle to development of practical site-specific weed management systems. Research progress is summarized for two different approaches to the problem, remote sensing weed mapping and ground-based detection using digital cameras or nonimaging sensors. The general spectral and spatial limitations reported for each type of weed identification system are reviewed. Airborne remote sensing has been successful for detection of distinct weed patches when the patches are dense and uniform and have unique spectral characteristics. Identification of weeds is hampered by spectral mixing in the relatively large pixels (typically larger than 1 by 1 m) and will not be possible from imagery where weed seedlings are sparsely distributed among crop plants. The use of multispectral imaging sensors such as color digital cameras on a ground-based mobile platform shows more promise for weed identification in field crops. Spectral features plus spatial features such as leaf shape and texture and plant organization may be extracted from these images. However, there is a need for research in areas such as artificial lighting, spectral band requirements, image processing, multiple spatial resolution systems, and multiperspective images.

Flexible field goniometer system: the Goniometer for Outdoor Portable Hyperspectral Earth Reflectance

Abstract. This paper describes a portable hyperspectral goniometer system for measurement of hemispherical conical reflectance factor (HCRF) data for terrestrial applications, especially in the coastal zone. This system, the Goniometer for Portable Hyperspectral Earth Reflectance (GOPHER), consists of a computer-controlled Spectra Vista Corporation HR-1024 full-range spectrometer mounted on a rotating arc and track assembly, allowing complete coverage in zenith and azimuth of a full hemisphere for recording HCRF. The control software allows customized scan patterns to be quickly modified in the field, providing for flexibility in recording HCRF and the opposition effect with varying grid sizes and scan ranges in both azimuth and zenith directions. The spectrometer track can be raised and lowered on a mast to accommodate variations in terrain and land cover. To minimize the effect of variations in illumination during GOPHER scan cycles, a dual-spectrometer approach has been adapted to link records of irradiance recorded by a second spectrometer during the GOPHER HCRF scan cycle. Examples of field data illustrate the utility of the instrument for coastal studies.

ANALYSIS OF CROP AND WEED LEAF DIFFUSE REFLECTANCE SPECTRA

Diffuse leaf reflectance at the three- to five-leaf stage was measured for six species common to agriculture in Saskatchewan, Canada. The three crop species were wheat (Triticum aestivum), canola (Brassica napus L.), and field pea (Pisum sativum L.). The three weed species were wild oat (Avena fatua L.), wild buckwheat (Polygonum convolvulus L.) and Canada thistle (Cirsium arvense (L.) Scop.). Plants were grown in regulated environmental chambers. For each plant, the leaf reflectance of two different leaves or leaf segments was measured, with between 40 and 50 plants for each species. Measurements were recorded between 270 and 2500 nm with a maximum data interval of 2 nm. Reflectance measurements were recorded with leaves against black and white backgrounds. These reflectance spectra corresponded to the expected green-plant profile. Distribution of the data was investigated numerically using the Anderson-Darling and Shapiro-Wilks goodness-of-fit statistics, standard deviation, skewness, and kurtosis, and qualitatively with histograms. Large deviations from the generally assumed normal distribution were observed, and these deviations were dependent on both wavelength and species.

Characterization of a low-cost diffuse reflectance coating

A simply formulated, inexpensive, and highly reflective diffuse coating was characterized with respect to its reflectance, bidirectional reflectance factor (BRF), minimum thickness, application method, abrasion, and response to weathering. The average reflectance of the coating between 400 and 1600 nm was 94.9 ± 0.13% (minimum reflectance of 86.4 ± 0.28% at 400 nm, maximum reflectance of 98.3 ± 0.14% at 989 nm) when spray-applied, which was slightly lower than, but still comparable to, that of commercially available coatings. The application method was shown to affect reflectance, BRF, and the effects of abrasion on BRF. Overall, spray application gave the best combination of high reflectance and close-to-ideal diffusion profile across the 400–1600 nm range. The coating became more specular with increased abrasion, regardless of application method, but the effect was most prominent with brush-applied coatings at longer wavelengths. The impact was slight enough, given the stiff brush used in testing, that infrequent light brushing with a soft-bristled brush would not adversely affect the coating properties. Weathering resulted in a decrease in reflectance of approximately 2%, with an increase in reflectance variability among samples and increased brittleness.

Sample holder and methodology for measuring the reflectance and transmittance of narrow-leaf samples.

Measuring the reflectance and transmittance of narrow samples can be difficult, as the width of the illuminating beam may be greater than the width of the sample. The small sample area can also compound the already time-consuming process of reconfiguring the instrument between reflectance and transmittance measurements by introducing additional alignment problems. A method of measuring the reflectance and transmittance properties of narrow-leaf samples using reflectance configurations only is developed and tested. The method uses a mask and mask correction and relationships between reflectance measurements against contrasting backgrounds to determine sample reflectance and transmittance. The design of the accompanying sample-holding apparatus is also described. In testing, the mean error was less than 1% reflectance/transmittance, and standard deviation of the error was approximately 1% reflectance and 2% transmittance as compared to samples measured using conventional measurement configurations.

Plant Discrimination Based on Leaf Reflectance

Reflectance spectra of leaves against white, black and soil backgrounds were collected for six plant species (Triticum aestivum, Brassica napus, Pisum sativum, Avena fatua L., Polygonum convolvulus L., Cirsium arvense (L.) Scop.) at the three to five leaf stage under consistent growing conditions. Diffuse reflectance measurements were taken using an UV-Vis-NIR spectrophotometer between 250 and 2500 nm. Two types of infinite reflectance spectra were calculated using a contrasting backgrounds method. Features for species discrimination were selected for inclusion in subsequent discriminant analysis. Classifiers were developed using three and five features for each of five datasets containing the six species. Each classifier was tested using test data corresponding to its training data. Classifiers were also evaluated, using testing data from other datasets as a measure of classifier robustness. The highest classifier accuracy was 97.92% when tested with data corresponding to its training set and using five features. Infinite reflectance based classifiers appeared to have improved robustness, achieving greater than 90% accuracy when tested with any testing set, while using only three features.

ULGS II: A High-Performance Field and Laboratory Spectrogoniometer for Measuring Hyperspectral Bidirectional Reflectance Characteristics

The derivation of the bidirectional reflectance distribution function (BRDF) from Earth surface features provides vital data for improved remote estimation of Earth surface features and properties. The angular component of a reflectance signature represents a valuable source of information that has largely been ignored due to the lack of empirical data. BRDF data products are now readily available from a variety of remote sensing platforms, and these require extensive field validation to ensure data quality and validity. This paper presents a high-performance, low-cost, and computer-controlled goniometer system capable of sampling surface BRDF and demonstrates several example applications. The system introduces a number of technological advances, including real-time reflectance calibration, programmable sampling schemes, zero target interference structure, and ease of portability. Using this instrument, BRDF estimates were gathered from a variety of agricultural crops in field and laboratory settings, with performance gains of four times the speed of current designs. This system can provide very fast acquisition of a BRDF estimate, depending on the desired angular resolution. Initial experiments demonstrate that this new style of hyperspectral goniometer is a significant advance over previous designs, with respect to scan capabilities, angular resolution, calibration, and portability.

BACKGROUND EFFECTS ON APPARENT LEAF REFLECTANCE

In sparse canopies, light striking the background soil and stubble may pass up through the canopy and affect apparent reflectance. The effect of three different backgrounds on the apparent reflectance of single leaves from six species was investigated. Spectral measurements were taken between 250 and 2500 nm using an UV-Vis-NIR spectrophotometer equipped with an integrating sphere. Results confirmed the findings of other studies, but also measured deeper into the UV range. Very small effects were found in the UV and visible spectra, with the exception of the green reflectance peak, which was affected by background. The greatest effects were observed in the NIR between 750 and 1300 nm, and between water absorption bands at 1450 and 1930 nm. Potential techniques for dealing with and exploiting bands which exhibit high sensitivity to background effects are suggested.

Image-Based Rapid Estimation of Frost Damage in Canola (Brassica napus L.)

Abstract Traditional methods of estimating frost damage to crops are labor-intensive and time-consuming. Remote sensing imagery and vegetation indices can be used for condition assessment, however, the utility of using vegetative indices in assessing frost damage specifically is not known. The objective of this study was to estimate the freezing injury using a vegetative index developed from hyperspectral imagery. Three replicates of six 6-leaf stage canola plants were subjected to a temperature of −10 °C for 6hr. The resulting frozen plants were imaged at 6 different thawing times using an imaging spectrophotometer (400-1000 nm). Normalized difference vegetation index (NDVI) and triangular vegetation index (TVI) were calculated. Contrary to expectations, NDVI values from frozen plants increased in the initial thawing treatments (1, 2 and 4hr) and then decreased at 8hr. Whereas, TVI values decreased gradually with increased duration of thawing. Furthermore, when compared to NDVI, TVI clearly differentiated frozen from control plants and within the freezing treatments. The differential modes of response of the indices to post-freezing reflectance changes is the possible reason. These results suggest that with recent advancements in low altitude remote sensing in the areas of spectral, spatial and temporal resolution, early estimation of frost damage is possible.

Optical flow profiling method for visualization and evaluation of flow disturbances in agricultural pneumatic conveyance systems

A flow-profiling apparatus and method is presented for dilute two-phase flows.The apparatus was applied to a laboratory air seeder simulator conveying wheat.Profiles were generated upstream, downstream, and around an obstruction.Obstruction effects on occurrence probability are presented. Pneumatic conveying is widely used for transporting granular materials and agricultural products. Traditional flow visualization methods are used extensively in experimental fluid dynamics but have not been commonly used with agricultural products as the flow seeding particles. A flow visualization method was developed to aid in understanding physical design changes made to agricultural pneumatic conveying systems. This optical flow profiling method is demonstrated by providing qualitative flow images and quantitative values to describe the behaviour of the particle flow, both upstream and downstream of a 25mm spherical obstruction. The sphere was attached to the bottom of an acrylic conveying line that was conveying wheat particles equivalent diameter: 3.66mm at an air speed of 20m/s and a mass flow rate of approximately 5kg/min. Probability density maps of particle occurrence were developed to describe the chance of a wheat particle being present in a particular location of the conveying line. The data contained in these maps were used to determine the centroid of the distribution and to plot the change in the cross-sections over the test area of the pneumatic conveying system.