Browsing by Author "Grafton M"

Now showing 1 - 20 of 21
  • Thumbnail Image
    Item
    A low-cost simple lysimeter soil retriever design for retrieving soil from small lysimeters
    (IOP Publishing, 2024-06-06) Gunaratnam A; McCurdy M; Grafton M; Jeyakumar P; Davies CE; Bishop P
  • Thumbnail Image
    Item
    Assessing the Leaf Blade Nutrient Status of Pinot Noir Using Hyperspectral Reflectance and Machine Learning Models
    (MDPI (Basel, Switzerland), 2023-03-08) Lyu H; Grafton M; Ramilan T; Irwin M; Sandoval E; Díaz-Varela RA
    Monitoring grape nutrient status, from flowering to veraison, is important for viticulturists when implementing vineyard management strategies, in order to produce quality wines. However, traditional methods for measuring nutrient elements incur high labour costs. The aim of this study is to explore the potential of predicting grapevine leaf blade nutrient concentration based on hyperspectral data. Leaf blades were collected at two Pinot Noir commercial vineyards at Martinborough, New Zealand. The leaf blade spectral data were obtained with a handheld spectroradiometer, to evaluate surface reflectance and derivative spectra in the spectrum range between 400 and 2400 nm. Afterwards, leaf blades nutrient concentrations (N, P, K, Ca, and Mg) were measured, and their relationships with the hyperspectral data were modelled by machine learning models; partial least squares regression (PLSR), random forest regression (RFR), and support vector regression (SVR) were used. Pearson correlation and recursive feature elimination, based on cross-validation, were used as feature selection methods for RFR and SVR, to improve the model’s performance. The variable importance score of PLSR, and permutation variable importance of RFR and SVR, were used to determine the most sensitive wavelengths, or spectral regions related to each biochemical variable. The results showed that the best predictive performance for leaf blade N concentration was based on PLSR to raw reflectance data (R2 = 0.66; RMSE = 0.15%). The combination of support vector regression with the Pearson correlation selected method and second derivative reflectance provided a high accuracy for K and Ca modelling (R2 = 0.7; RMSE = 0.06%; R2 = 0.62; RMSE = 0.11%, respectively). However, the modelling performance for P and Mg, by different feature groups and variable selection methods, was poor (R2 = 0.15; RMSE = 0.02%; R2 = 0.43; RMSE = 0.43%, respectively). Thus, a larger dataset is needed for improving the prediction of P and Mg. The results indicated that for Pinot Noir leaf blades, raw reflectance data had potential for the prediction of N concentration, while the second-derivative spectra were more suitable to predict K and Ca. This study led to the provision of rapid and non-destructive measurements of grapevine leaf nutrient status.
  • Thumbnail Image
    Item
    ASSESSMENT OF NITROGEN FERTILIZERS UNDER CONTROLLED ENVIRONMENT – A LYSIMETER DESIGN
    (12/04/2019) Gnaratnam A; McCurdy M; Grafton M; Jeyakumar P; Bishop P; Davies C; Currie, L; Christensen, C
    This paper introduces a closed system lysimeter design to measure fertilizer performance on ryegrass. The lysimeter will measure plant mass growth, gas emissions and leachate in a controlled climate environment based on a long term 90 day spring climate from the Taranaki. A range of commercial fertilizers will be compared to bespoke fertilizers manufactured under this project. This work, although undertaken in laboratory conditions will help quantify the impacts of nitrogenous fertilizers on the environment by mimicking actual conditions in a controlled setting. The study should provide data on the effectiveness of novel fertilizers manufactured within the programme; and other slow and controlled fertilizers, in reducing nitrogen leaching and greenhouse gas (GHG) emissions on pasture. Nitrogenous fertilizers readily leach as nitrates are highly soluble and GHG are emitted through volatilisation of ammonia and nitrous oxide. Reduced leaching and volatilisation increases fertilizer efficiency as less is wasted and more is attenuated in the plant. The aims of the research are to increase the effectiveness and efficiency of nitrogen fertilizer use in New Zealand. This should benefit farmers by reducing the amount of fertilizer applied, ideally reducing fertilizer cost, or at no extra cost by improved plant attenuation. This would also have an environmental benefit through reduced leaching and GHG emissions.
  • Thumbnail Image
    Item
    Capability of ground fertiliser placement when spread from a fixed wing aircraft
    (13/04/2016) Chok S; Grafton M; Yule IJ; Manning M
    Aerial topdressing using differential rate application technology improves fertiliser spreading on hill country farms. However, the system’s ability to place fertiliser accurately and precisely within an area needs to be determined. Accuracy was determined by comparing measured and intended application rates. Precision was indicated by the coefficient of variation (CV), which is the standard deviation of the measured application rate over the mean of this rate. Two trials were carried out, where aircraft deposited fertiliser at two application rates and fertiliser was captured using cone-shaped collectors. The average measured application rate for both trials was less than the intended rate. The CV ranged from 35 to 57%, and was lower than CV’s from pilot-operated hopper systems (78%). A one-way analysis of variance test found the difference between measured application rate in the high and low application zone was statistically significant. The results indicate work is required to improve the accuracy and precision of the differential rate system, however, the system shows promise. Keywords: differential rate application technology, aerial spreading, fertiliser placement
  • Thumbnail Image
    Item
    COMPARATIVE EVALUATION OF CONTROLLED RELEASE FERTILISERS FOR NITRATE LEACHING
    Raveendrakumaran B; Grafton M; Jeyakumar P; Bishop P; Davies C; Christensen, C; Horne, D; Singh, R
    A lysimetric study was carried out with an objective of evaluating the leaching behaviour of different fertilisers on spinach growth on Manawatu sandy soil. The fertiliser treatments applied were urea, two controlled release fertilisers called ‘g’and ‘SmartN’ at the rates of 50 kg N/ha (50N), 100 kg N/ha (100N) and 200 kg N/ha (200N). The 200 kg N/ha urea application was made in 10 split doses at a rate of 20 kg N/ha in 7-day intervals, whereas 200N application of ‘g’ and ‘SmartN’ were made twice at a rate of 100 kg N/ha at the time of planting and six weeks after planting. The control treatment did not receive any fertiliser application (0N). The application of Urea and ‘g’ at all rates except ‘g’-50N produced significantly higher nitrate leaching losses (19.8 to 27.7 kgN/ha) compared to the control (9.1 kgN/ha), while SmartN at all rates produced no significant increase in nitrate leaching. The total nitrate leached per ton of dry matter production was significantly reduced by the application of N fertilisers compared to the control (135.1 kgNO3 - -N/MgDM). On an average, 16.4 kg NO3 - -N/MgDM was leached from the fertilised treatments. Dry matter production increased at 200N application rates with all three fertilisers, but urea-200N produced the highest dry matter yield of 2377 kg/ha. In conclusion, frequent split applications of urea (urea - 200N) increased dry matter yield thereby significantly reduced nitrate leaching.
  • Thumbnail Image
    Item
    Comparing the Carbon Storage Potential of Naturally Regenerated Tea Trees with Default New Zealand Carbon Look-Up Tables: A Case Study
    (MDPI, 12/04/2023) Wilson T; Grafton M; Irwin M; Hatano, R
    The New Zealand Emissions Trading Scheme allows landowners to be remunerated for the carbon sequestration capabilities of eligible forests established post 1990. For afforested areas of 100 hectares or fewer, carbon sequestration is estimated with the use of default carbon look-up tables administered by the Ministry for Primary Industries. However, a disparity exists between exotic pines (Pinus radiata), where carbon sequestration predictions are regionally differentiated, and native species, where carbon sequestration estimations are neither distinguished by species or locality. This paper aims to highlight this inequality by comparing the calculated carbon storage of endemic tree species with the ‘Indigenous Forest’ category in the carbon look-up tables. The carbon storage of 12-year-old naturally regenerated tea trees (Leptospermum scoparium and Kunzea ericoides) was calculated using allometric measurements and compared to the look-up tables. The results suggest that carbon look-up tables underestimate the carbon sequestration of native tea trees by 81.8%. A bimodal data distribution suggests that carbon sequestration is heavily dependent on light interception levels. It is recommended that carbon sequestration data for specific native species in different environments are collected and integrated into such tables.
  • Thumbnail Image
    Item
    DEMONSTRATING THE COMPATIBILITY OF A NEW SPREADMARK TEST WITH THE CURRENT METHOD
    (12/04/2019) Wilson T; Grafton M; Currie, L; Christensen, C
    The New Zealand Spreadmark test which although proven to accurately measure the Coefficient of Variation (CV) of spreading equipment, entails a laborious procedure which is expensive to implement. This study aims to validate the accuracy of a newly developed test method based on the current one that hastens the process, making it increasingly cost effective. The proposed solution reduces the amount of trays used to collect and measure the fertiliser spread pattern. The proposed method reduces the number of trays by half, placing them one meter apart compared to the current industry standard of half a meter. An electronic tray weighing system developed by EuroAgri streamlines the process. This allows the scales to be, zeroed, after each pass by removing the need to empty trays. Collated data of previous Spreadmark tests sourced from certified Spreadmark testers. This had the support of the Fertiliser Quality Council that manages the scheme used in the study. Tray weights of each successive 0.5 and 1.0 metres were, averaged to imitate tray spaces of 1.0 metre. The 1.0 metre tray spacing showed a strong correlation to the 0.5 meter spaces, maintaining the normal distribution pattern of the spread fertilizer albeit in a slightly lower definition. Coupled with the electronic scales that reduces human error, this forms an accurate and efficient method of undertaking testing. This new system could have marked effects upon the future of spreader testing in New Zealand, including higher proportions of conforming spreaders (due to increased time and cost effectiveness) leading to lower field coefficient of variation (CV). As a result, fertiliser efficacy would increase, as would financial returns.
  • Thumbnail Image
    Item
    Evaluation of Point Hyperspectral Reflectance and Multivariate Regression Models for Grapevine Water Status Estimation
    (MDPI AG, 12/08/2021) Wei H-E; Grafton M; Bretherton M; Irwin M; Sandoval E
    Monitoring and management of plant water status over the critical period between flower-ing and veraison, plays a significant role in producing grapes of premium quality. Hyperspectral spectroscopy has been widely studied in precision farming, including for the prediction of grapevine water status. However, these studies were presented based on various combinations of transformed spectral data, feature selection methods, and regression models. To evaluate the performance of different modeling pipelines for estimating grapevine water status, a study spanning the critical period was carried out in two commercial vineyards at Martinborough, New Zealand. The modeling used six hyperspectral data groups (raw reflectance, first derivative reflectance, second derivative reflectance, continuum removal variables, simple ratio indices, and vegetation indices), two variable selection methods (Spearman correlation and recursive feature elimination based on cross-validation), an ensemble of selected variables, and three regression models (partial least squares regression, random forest regression, and support vector regression). Stem water potential (used as a proxy for vine water status) was measured by a pressure bomb. Hyperspectral reflectance was undertaken by a handheld spectroradiometer. The results show that the best predictive performance was achieved by applying partial least squares regression to simple ratio indices (R2 = 0.85; RMSE = 110 kPa). Models trained with an ensemble of selected variables comprising multicombination of transformed data and variable selection approaches outperformed those fitted using single combinations. Although larger data sizes are needed for further testing, this study compares 38 modeling pipelines and presents the best combination of procedures for estimating vine water status. This may lead to the provision of rapid estimation of vine water status in a nondestructive manner and highlights the possibility of applying hyperspectral data to precision irrigation in vineyards.
  • Thumbnail Image
    Item
    Evaluation of the use of two-stage calibrated PlanetScope images and environmental variables for the development of the grapevine water status prediction model
    (24/05/2023) Wei H-E; Grafton M; Bretherton M; Irwin M; Sandoval E
    Abstract Grapevine water status (GWS) assessment between flowering and veraison plays an important role in viticulture management in terms of producing high-quality grapes. Although satellites and uncrewed aerial vehicles (UAV) have successfully monitored GWS, these platforms are practically limited because data transfer is delayed due to post processing and UAV operation is weather dependent. This study focuses on addressing two issues: the unreliability of GWS estimation using satellite images with low-moderate spatial resolution and the inaccessibility of real-time satellite data. It aims to predict the temporal variation of GWS based on a prediction model using spectral information (calibrated PlanetScope (PS) images), soil/topography data (apparent electrical conductivity, elevation, slope), weather parameters (rainfall and potential evapotranspiration), cultivation practices (irrigation, fertigation, plucking, and trimming), and seasonality (day of the year) as predictors. Stem water potential (Ψstem) was used as a proxy for GWS. Two-stage calibration, including an initial calibration of UAV images with measured Ψstem and a subsequent calibration of satellite images with calibrated UAV data, was applied to calibrate the PS images. Three machine learning models (random forest regression, support vector regression, and multilayer perceptron) were used in the calibration and modeling process. The results showed that a two-stage calibration can generate reliable reference data, with a root mean square error of 113 kPa and 59 kPa on the test sets during the first and second calibration stage, respectively. The prediction model described the temporal variation of block Ψstem when compared with the measured Ψstem. In the similarity analysis, the Pearson correlation coefficient was 0.89 and 0.87 between predicted and reference Ψstem maps across four dates for the two study vineyards. This study supports the concept of developing an approach to predict grapevine Ψstem, which would enable growers to acquire Ψstem variation in advance during the growing season, leading to improved irrigation scheduling and optimal grape quality.
  • Thumbnail Image
    Item
    Formulation and characterization of polyester-lignite composite coated slow-release fertilizers
    (Springer Nature Switzerland AG, 26/09/2022) Gunaratnam A; Bishop P; Jeyakumar P; Grafton M; Davies CE; McCurdy M
    Two polyester-lignite composite coated urea slow-release fertilizers (SRFs; Poly3 and Poly5) were developed and their physicochemical properties were studied. Both these SRFs significantly (p < 0.05) extended the urea release compared to uncoated urea; Poly3 and Poly5 by 117 and 172 h, respectively. The urea release characteristics of Poly5 were further enhanced by linseed oil application (Poly5-linseed). The SEM images demonstrated the coatings were in contact with the urea and encase urea particles completely with the average coating thickness of 167.2 ± 15 µm. The new interactions between polyester and lignite in the composite coating were confirmed by the FTIR analysis. Polyester-calcium carbonate (Polyester-CaCO3) coated SRFs (Calc3 and Calc5) were developed using CaCO3 as a filler in place of lignite and the urea dissolution rate was compared with Poly3 and Poly5. The urea release times for the polyester-CaCO3 formulations, 48 and 72 h, were significantly (P < 0.05) lower than the polyester-lignite formulation, showing that lignite imparted greater control over release time than CaCO3. Findings from this work showed that polyester-lignite composites can be used as a coating material for SRFs.
  • Thumbnail Image
    Item
  • Thumbnail Image
    Item
    Hyperspectral Imaging Spectroscopy for Non-Destructive Determination of Grape Berry Total Soluble Solids and Titratable Acidity
    (MDPI AG, 2024-05-07) Lyu H; Grafton M; Ramilan T; Irwin M; Sandoval E; Krasuki K; Weirzbicki D
  • Thumbnail Image
    Item
    Integrating soil moisture measurements into pasture growth forecasting in New Zealand’s hill country
    Hajdu I; Yule I; Bretherton M; Singh R; Grafton M; Nelson, W
    Forecasting pasture growth in hill country landscapes requires information about soil water retention characteristics, which will help to quantify both water uptake, and its percolation below the root zone. Despite the importance of soil moisture data in pasture productivity predictions, current models use low-resolution estimates of water input into their soil water balance equations and plant growth simulations. As a result, they frequently fail to capture the spatial and temporal variability of soil moisture in hill country soils. Wireless Sensor Networks (WSN) are promising in-situ measurement systems for monitoring soil moisture dynamics with high temporal resolution in agricultural soils. This paper presents the deployment of a soil moisture sensing network, utilising WSN technology and multi-sensor probes, to monitor soil water changes over a hill country farm in the northern Wairarapa region of the North Island. Processed capacitance-based raw data was converted to volumetric water content by means of a factory calibration function to assess sensor accuracy and to calculate soil water storage within the pasture root zone. The derived volumetric soil moisture data was examined in terms of its dependence on the variability and influences of hill country landscape characteristics such as aspect. The integration of spatially distributed sensors and multi-depth soil moisture measurements from various hillslope positions showed that slope and aspect exerted a significant impact on soil moisture values. Furthermore, considerable differences were identified in soil water profile responses to significant rainfall events and subsequent soil water redistribution. Initial indications are that high-resolution time series of accurate multi-depth soil moisture measurements collected by a WSN are valuable for investigating root zone water movement. Sensor evaluation and data analysis suggest that these devices and their associated datasets are able to contribute to an improved understanding of drying and wetting cycles and soil moisture variability. Potentially, this will create an opportunity to generate improved pasture growth predictions in pastoral hill country environments.
  • Thumbnail Image
    Item
    Intellispread®: Precision aerial topdressing
    (researchfeatures, 2022-02-21) Grafton M; Irwin ME
    Aerial topdressing – the aerial application of fertilisers over farmland using specialist agricultural aircraft – is an integral part of New Zealand’s agricultural heritage. The procedure was born and developed there, so it makes sense that New Zealand researchers are behind much of its development. Dr Miles Grafton and Matthew Irwin from Massey University on North Island, believe that increasing the efficacy of aerial topdressing is possible by reducing the role of a currently crucial part of the procedure: the pilot.
  • Thumbnail Image
    Item
    Iron-rich sand promoted nitrate reduction in a study for testing of lignite based new slow-release fertilisers
    (Elsevier, 20/12/2022) Abhiram G; Grafton M; Jeyakumar P; Bishop P; Davies C; McCurdy MM
    The N losses and agronomic performances of newly developed slow-releasing fertilisers (SRFs; Epox5 and Poly5) were tested against conventional N fertilisers, urea and diammonium phosphate (DAP), in a climate-controlled lysimeter system. The dry matter (DM) yield and N losses of SRFs were not significantly different from urea and DAP. However, nitrate leaching and nitrous oxide (N2O) losses were unexpectedly low and therefore, it was inferred that nitrate underwent a chemical transformation. It was observed that a thick fibreglass wick interrupted drainage and created an anaerobic condition in the soil. The subsoil was found to have a high extractable total iron and it was postulated that iron played a role in the observed low level of N losses. An investigation was carried out with a factorial design using sand types and rates of N application as the main factors. Two types of sand; with high and low iron concentration and four levels of N applications; 0 (control), 50, 100 and 200 kg N ha-1 were employed in a leaching column and nitrate and N2O losses were measured. The nitrate leaching was significantly (P < 0.05) affected by sand types wherein a lower nitrate level was recorded for high‑iron concentration sand than for low-iron concentration sand at all N application levels. The N2O emission was significantly (P < 0.05) lower for high-iron sand than for low-iron sand for the 200 N treatment, but not significantly different between sand types for other treatments. These observations provide evidence for the involvement of iron in nitrate transformation under anaerobic conditions and it was hypothesised path was dissimilar nitrate reduction (DNR). Further studies are recommended, to identify the underlying mechanism responsible for nitrate reduction with iron-rich sand.
  • Thumbnail Image
    Item
    STUDY THE INFLUENCE OF SOIL MOISTURE AND PACKING INCREMENTAL LEVEL ON SOIL PHYSICAL AND HYDRAULIC PROPERTIES
    (14/07/2020) Gunaratnam A; Grafton M; Jeyakumar P; Bishop P; Davies C; McCurdy M; Christensen, C; Horne, D; Singh, R
    Reconstructed soil packing is an alternative for monolithic soil columns in lysimeter studies. The excavated soil is packed in uniform layers to represent the natural soil conditions. Reconstructed soil packing alters the physical properties, including bulk density and porosity, thus can distort the hydraulic properties of the soil, so consistency of the method used is critical. Therefore, the selection of a suitable packing method is imperative. This preliminary study comes under the broad research programme: “developing and testing new fertilizer formulations in lysimeters”. This work was aimed to study the effect of incremental packing methods on the hydraulic properties of soil to select the best combination for testing fertilizers. The selected soil matrix for this lysimeter study was composed of 10 cm topsoil and 30 cm washed builders’ sand. For this study, four different soil packs were trialled in lysimeters with the combination of two soil moisture conditions (dry/damp and wet) and two packing depth increments (5 and 10 cm). The flow rate and saturated hydraulic conductivity were measured. Subsequently, several pore volumes of water (around 5 – 6) was allowed to pass through the soil column and the soil subsidence level was measured for each packing method. Both soil moisture condition and packing increment level have influenced the flow rate and saturated hydraulic conductivity of the soil matrix. The saturated hydraulic conductivity of the dry-5 cm, dry-10 cm, wet-5 cm and wet-10 cm packing were 3.99, 6.70, 3.56 and 6.53 cm hr- 1 , respectively. Soil subsidence was also influenced by both the soil moisture condition and increment level. The highest soil subsidence was exhibited by dry-10 cm packing (13 mm) and lowest by wet-5 cm (2 mm) (p<0.05). This preliminary study showed that both moisture condition and increment level influence the soil hydraulic property and compaction level. Further study needs to be conducted to understand the influence of soil moisture and incremental level on other physical and hydraulic properties of soil packing.
  • Thumbnail Image
    Item
    The Nitrogen Dynamics of Newly Developed Lignite-Based Controlled-Release Fertilisers in the Soil-Plant Cycle
    (MDPI AG, 29/11/2022) Gunaratnam A; Grafton M; Jeyakumar P; Bishop P; Davies C; McCurdy M
    The effect of newly developed controlled-release fertilisers (CRFs); Epox5 and Ver-1 and two levels of Fe2+ applications (478 and 239 kg-FeSO4 ha−1) on controlling nitrogen (N) losses, were tested on ryegrass, in a climate-controlled lysimeter system. The Epox5 and Ver-1 effectively decreased the total N losses by 37 and 47%, respectively, compared to urea. Nitrous oxide (N2O) emissions by Ver-1 were comparable to urea. However, Epox5 showed significantly higher (p < 0.05) N2O emissions (0.5 kg-N ha−1), compared to other treatments, possibly due to the lock-off nitrogen in Epox5. The application of Fe2+ did not show a significant effect in controlling the N leaching loss and N2O emission. Therefore, a dissimilatory nitrate reduction and chemodenitrification pathways were not pronounced in this study. The total dry matter yield, N accumulation, N use efficiency and soil residual N were not significantly different among any N treatments. Nevertheless, the N accumulation of CRFs was lower in the first month, possibly due to the slow release of urea. The total root biomass was significantly (p < 0.05) lower for Epox5 (35%), compared to urea. The hierarchical clustering of all treatments revealed that Ver-1 outperformed other treatments, followed by Epox5. Further studies are merited to identify the potential of Fe2+ as a controlling agent for N losses.
  • Thumbnail Image
    Item
    USING MONTE CARLO SIMULATIONS TO ACHIEVE THE BEST RESPONSE FROM NITROGEN ON GRAZED PASTURE UNDER A LEGISLATED NITROGEN CAP IN NEW ZEALAND: A REVIEW
    (27/10/2022) Grafton M; Bishop PA; Bretherton MR; Marek, G; Migliaccio, K
    Pastoral and crop farming systems have traditionally used the application of nitrogen (N) to achieve an optimal economic production response. This nitrogen response is estimated from an exponential function that approaches an as ymptote, which is typical of most fertilizer response curves. The optimal economic N response is often achieved when appli cation rates are greater than plant utilization rates, often resulting in leaching, nitrogen run-off, and volatilization of ni trogenous compounds. These losses can have an impact on freshwater quality and contribute to greenhouse gas (GHG) emissions. In New Zealand, urine from N-fertilized pasture grazed by dairy cattle has been shown to be the most problematic source of N losses. As part of New Zealand’s National Environmental Standards (NES), a synthetic N cap of 190 kgN ha-1yr-1 on grazed pasture and crops has been implemented to reduce nutrient enrichment of fresh water. This study reviewed the use of multiple split applications of N to improve N fertilizer use efficiency and pasture response and used Monte Carlo simulations to demonstrate improved response to split N applications rather than a single optimal application based on economic response. In addition, spreading accuracy also became less important as all the low-application variation occurred along the steepest part of the response curve where this variation results in added yield.
  • Thumbnail Image
    Item
    USING PROXIMAL HYPERSPECTRAL SENSING TO MEASURE SOIL OLSEN P AND pH
    (12/04/2019) Grafton M; Kaul T; Palmer A; Bishop P; White M; Currie, L; Christensen, C
    This paper reports on work undertaken to use a large data set of hyperspectral data measured on dry soil samples to obtain regression analysis which allows predictions of pH and Olsen P to be obtained from an independent data set. The large data set was obtained from 3,190 soil samples taken from the Ravensdown Primary Growth Partnership to a depth of 7.5cm. The spectra were measured using an Analytical Spectral Device which recorded 2,150 wavebands of 1nm resolution between 350nm and 2,500nm. Values for Olsen P and pH were provided from chemical analysis by Analytical Research Laboratories. The spectra were regressed using “R” statistical software which has the power to handle the data and report the wavebands with the most significance for the model. The data set for the prediction came from a stratified nested, grid soil sampling exercise which was used to find Olsen P stability at varying depths. This set had 400 samples from each of two data sets from different areas on Patitapu Station using a grid sample protocol. The 100 most significant wavebands from the PGP data set were used to regress the Patitapu data which were combined. These were regressed using “R” (Version 3.41, The R Foundation) and Statdata (Palisade, New York), which produced the same result. The partial least square regression of pH was very significant and was predicted well. Olsen P had a very significant correlation which was quite noisy, correlating the log10 of Olsen P was also undertaken and it would appear something is being measured that is associated with Olsen P. This work shows that it is possible to measure soil nutrient by proximal hyperspectral analysis which is transferable to an independent data set.
  • Thumbnail Image
    Item
    Using Remote and Proximal Sensing Data and Vine Vigor Parameters for Non-Destructive and Rapid Prediction of Grape Quality
    (MDPI AG, 2023-11-19) Lyu H; Grafton M; Ramilan T; Irwin M; Wei H-E; Sandoval E; Zhang C; Liu D
    The traditional method for determining wine grape total soluble solid (TSS) is destructive laboratory analysis, which is time consuming and expensive. In this study, we explore the potential of using different predictor variables from various advanced techniques to predict the grape TSS in a non-destructive and rapid way. Calculating Pearson’s correlation coefficient between the vegetation indices (VIs) obtained from UAV multispectral imagery and grape TSS resulted in a strong correlation between OSAVI and grape TSS with a coefficient of 0.64. Additionally, seven machine learning models including ridge regression and lasso regression, k-Nearest neighbor (KNN), support vector regression (SVR), random forest regression (RFR), extreme gradient boosting (XGBoost), and artificial neural network (ANN) are used to build the prediction models. The predictor variables include the unmanned aerial vehicles (UAV) derived VIs, and other ancillary variables including normalized difference vegetation index (NDVI_proximal) and soil electrical conductivity (ECa) measured by proximal sensors, elevation, slope, trunk circumference, and day of the year for each sampling date. When using 23 VIs and other ancillary variables as input variables, the results show that ensemble learning models (RFR, and XGBoost) outperform other regression models when predicting grape TSS, with the average of root mean square error (RMSE) of 1.19 and 1.2 ◦Brix, and coefficient of determination (R2 ) of 0.52 and 0.52, respectively, during the 20 times testing process. In addition, this study examines the prediction performance of using optimized soil adjusted vegetation index (OSAVI) or normalized green-blue difference index (NGBDI) as the main input for different machine learning models with other ancillary variables. When using OSAVI-based models, the best prediction model is RFR with an average R2 of 0.51 and RMSE of 1.19 ◦Brix, respectively. For NGBDI-based model, the RFR model showed the best average result of predicting TSS were a R2 of 0.54 and a RMSE of 1.16 ◦Brix, respectively. The approach proposed in this study provides an opportunity to grape growers to estimate the whole vineyard grape TSS in a non-destructive way.