In arable farming (e.g. sugar-beet, onion, potato, carrot, etc.), plant root activities primarily occur within the top 0.3 m of soil layer. Therefore, measurement of soil spatial variability and identifying field-scale heterogeneities of this top layer is very important from the perspective of site specific crop management. Apparent soil electrical conductivity (EC_a), which is related to different soil physical properties, such as clay content, moisture content, bulk density, pH and salinity, can be used to determine the soil spatial variability in a convenient way. Electromagnetic induction based EM38 instrument when placed on ground can measure soil EC_a upto a depth of 0.75 m in horizontal dipole mode and 1.5 m in vertical dipole mode. Numerous researchers used EM38 to measure depth weighted EC_a for discrete soil depth intervals; however, there is no simple technique that can be used to measure EC_a in top 0.3 m soil layer. To visualize EC_a variations in a field, EC_a mapping is also an important aspect of precision agriculture. Furthermore, the accuracy of interpolation methods for spatially varying soil properties has been analysed in several studies. However, a large discrepancy exists among the findings of the researchers.
The objectives of this study were to find a simple method to measure root-zone (top 0.3m) soil EC_a with the help of an EM38 and to check if it can be representative of the measured soil physical properties. Further intent of this study was to investigate different deterministic, geo-statistical and hybrid interpolation techniques for EC_a and soil properties mapping. To evaluate the accuracy of maps using relatively a fine (e.g. 10.5x10.5 m) and a coarse (21x21 m) grid size was also a sub-objective of this study.
A simple method has been developed based on the electromagnetic induction theory to determine the EC_a of root-zone soil layer. The EM38 was placed in both horizontal and vertical modes at 0, 15, 30. 60, 90 and 120 cm above soil surface to get the depth weighted EC_a. From this depth weighted EC_a profile, root-zone EC_a was calculated. Two equations were derived for both horizontal and vertical dipole modes. The EC_a measured from top 0.3 m in horizontal and vertical dipole modes was correlated with soil physical properties such as, clay content, moisture content, bulk density, pH and electrical conductivity. Mapping techniques investigated in this study were comprised on deterministic (inverse distance weighting, spline), geostatistical (ordinary kriging, universal kriging) and hybrid (Co-kriging) interpolation techniques. Root mean square error (RMSE) was used to compare the accuracy of different interpolation techniques for EC_a and different soil properties mapping. To evaluate the accuracy of maps between a fine and a coarse grid size, cross validation technique was used.
In horizontal dipole mode, about 57.70, 19.00, 11.15 and 7.27% response was calculated whereas in vertical dipole mode 50.20, 19.60, 13.40 and 9.25% response was calculated from 0-0.3, 0.3-0.6, 0.6-0.9 and 0.9-1.2 m depth respectively. As EC_a in horizontal mode can give better response from top soil, therefore, 57.70% of EC_a (H) is the root-zone EC_a. Positive correlation between EC_a and clay content, MC, pH and BD conclude that these soil properties contribute to soil EC_a. All techniques showed comparable results, however, Co-kriging outperformed slightly over other techniques. Ordinary kriging gave better predictions for clay (RMSE = 2.03) whereas universal kriging interpolated pH with lowest RMSE (RMSE = 0.226). The fine grid size (10.Sm) gave better result (R² = 0.67) than a coarse grid size (R² = 0.55).
Our technique is very simple and can be used easily by taking a few EM38 readings at various height above ground on a field level. Different relations can be developed for different types of soils with varying soil properties. Root-zone EC_a measurement and mapping can be used for better soil management in arable farming.

Evapotranspiration (ET) is the total amount of water released by a crop or pasture canopy in the form of Transpiration (T) and Evaporation (E). ET accounts for up to 96% of the water loss depending on the types of vegetation cover and climatic conditions (Wilcox et al., 2003). Transpiration is related to the productivity of crops and pasture; whereas, evaporation is the loss of water directly from soil surface. Estimating separately the components of ET in the field is challenging yet it is highly significant in terms of improving crop water use and irrigation efficiency, since a anywhere between 30 and 80% of the water flux can be associated with the all important evaporation component (Wilcox et al., 2003). Sap flow monitoring, using micro-lysimeters or isotopic analysis are the usual options for separately measuring transpiration and evaporation in plants but these are incompatible with in-situ field deployment. In this study a portable and convenient method for separately determining the evapotranspiration components has been developed. When coupled with widely-used active optical sensors, the device can be used to develop, in-situ, relationships between spectro-optical indices such as NDVI and evapotranspiration coefficients (Kc, Kcb and Ke) that characterize actual ET water loss relative to evaporative demand.

Monitoring pasture growth rate is an important component of managing grazing livestock production systems. In this study, we demonstrate that a pasture growth rate (PGR) model, initially designed for NOAA AVHRR normalised difference vegetation index (NDVI) and since adapted to MODIS NDVI, can provide PGR at spatial resolution of ~2 m with an accuracy of ~2 kg DM/ha.day when incorporating in-situ sensor data. A PGR model based on light-use efficiency (LUE) was combined with 'in-situ' measurements from proximal weather (temperature), plant (fraction of absorbed photosynthetically active radiation, fAPAR) and soil (relative moisture) sensors to calculate the growth rate of a tall fescue pasture. Based on an initial estimate of LUEmax for the candidate pasture, followed by a process of iterating LUEmax to reduce prediction errors, the model was capable of estimating PGR with a root mean square error of 1.68 kg/ha.day (R² = 0.96, P-value ≈ 0). The iterative process proved to be a convenient means of estimating LUE of this pasture (1.59 g DM/MJ APAR) under local conditions. The application of the LUE-PGR approach to developing an in-situ pasture growth rate monitoring system is discussed.

Accurate yield forecasting in high value fruit tree crops provides vital management information to growers as well as supporting improved decision making, including postharvest handling, storage and forward selling. Current research evaluated the 8 spectral band WorldView 3 (WV-3) with a spatial resolution of 1.2 m, as a tool for exploring the relationship between individual tree canopy reflectance and a number of tree growth parameters, including yield. WV-3 imagery was captured on the 7th of April, 2016, over two Macadamia ('Macadamia integrifolia') and three Avocado ('Persea americana') orchards growing near the Queensland township of Bundaberg, Australia. Using the extent of each block, the WV-3 imagery was sub-setted and classified into 8 Normalised Difference Vegetation index (NDVI) classes. From these classes 6 replicate trees were selected to represent high, medium and low NDVI regions (n=18) and subsequently ground truthed for a number of yield parameters during April and May, 2016. The measured parameters were then correlated against 20 structural and pigment based vegetation indices derived from the 8 band spectral information corresponding to each individual tree canopy (12.6 m2). The results identified a positive relationship between derived vegetation indices (VI) and fruit weight (kg/tree) R2 > 0.69 for Macadamia and R2 > 0.68 for Avocado; and fruit number R2 > 0.6 for Macadamia and R2 > 0.61 for Avocado. The algorithm derived between the optimum VI and yield for each block was then applied across the entire block to derive a yield map. The results show that remote sensing of tree canopy condition can be used to measure yield parameters in Macadamia and Avocado grown in the Bundaberg region.

Satellite imagery has been demonstrated to be an effective technology for producing accurate pre-harvest estimates in many agricultural crops. For Australian sugarcane, yield forecasting models have been developed from a single date SPOT satellite image acquired around peak crop growth. However, a failure to acquire a SPOT image at this critical growth stage, from continued cloud cover or from competition for the satellite, can prevent an image being captured and therefore a forecast being made for that season. In order to reduce the reliance on a single image capture and to improve the accuracies of the forecasts themselves, time series yield prediction models have been developed for eight sugarcane growing regions using multiple years of free Landsat satellite images. In addition to the forecasting of average regional yield, an automated computational and programming procedure enabling the derivation of crop vigour variability (GNDVI) maps from the freely available Sentinel 2 satellite imagery was developed. These maps, produced for 15 sugarcane growing regions during the 2017 growing season, identify both variations in crop vigour across regions and within every individual crop. These outputs were made available to collaborating mills within each growing region. This paper presents the accuracies achieved from the time series yield forecasting models versus actual 2017 yields for the respective regions, as well as provides an example of the derived mapping outputs.

The normalized difference vegetation index (NDVI), derived from ground based or satellite borne, passive sensors is often used to estimate the fraction of absorbed photosynthetically active radiation (fAPAR) of a plant canopy. It is well documented that the measured NDVI from passive sensors is affected by the sun and/or view geometry due to the non-Lambertian properties of plant canopies. Despite this the fAPAR-NDVI relationships are often found to be independent of the solar elevation angle (ᶿs) because the ᶿs-dependent absorption of the Red wavelengths within the canopy, which dominates the fAPAR, cancels out the ᶿs-dependency of the NIR scattering which dominates the NDVI measurement. Active optical sensors (AOS), which have their own illuminating light source measure NDVI (NDVI AOS) without any interference of solar geometry. However as fAPAR of a plant canopy does change with solar elevation angle (ᶿs), the fAPAR-NDVIAOS relationship too changes with varying ᶿs. The objective of this study was to explore a correction factor which can eliminate the ᶿs-dependency in fAPAR-NDVIAOS relationship. Data were collected using LightScout quantum bar and CropCircle™ for Tall fescue ('Festuca arundinacea' var. Fletcher) at ᶿs ranging from 40° to 80°. A ᶿs-dependent vegetation index, NDVI*AOS that introduces simple trigonometric correction factors to the measured Red and NIR irradiance for nadir-viewing active optical sensor provides a fAPAR-NDVI relationship that is independent of ᶿs. When the solar elevation angle is introduced this way into the NDVIAOS the fAPAR can then be calculated from the NDVIAOS for any solar elevation angle within the range of 40-80°.

Quantifying sugarcane production is critical for a wide range of applications, including crop management and decision making processes such as harvesting, storage, and forward selling. This study explored a novel model for predicting sugarcane yield in Bundaberg region from time series Landsat data. From the freely available Landsat archive, 98 cloud free (<40%) Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) images, acquired between November 15th to July 31st (2001-2015) were sourced for this study. The images were masked using the field boundary layer vector files of each year and the GNDVI was calculated. An analysis of average green normalized difference vegetation index (GNDVI) values from all sugarcane crops grown within the Bundaberg region over the 15 year period identified the beginning of April as the peak growth stage and, therefore, the optimum time for satellite image based yield forecasting. As the GNDVI is an indicator of crop vigor, the model derived maximum GNDVI was regressed against historical sugarcane yield data, which showed a significant correlation with R² = 0.69 and RMSE = 4.2 t/ha. Results showed that the model derived maximum GNDVI from Landsat imagery would be a feasible and a modest technique to predict sugarcane yield in Bundaberg region.

The estimation of actual crop evapotranspiration (ETc) from any given land cover or crop type is important for irrigation water management and agricultural water consumption analysis. The main parameter used for such estimations is the crop coefficient (Kc). Spectral reflectance indices, such as the normalized difference vegetation index (NDVI) and the crop coefficient of a specific crop or pasture canopy are important indicators of 'vigour', namely the photosynthetic activity and rate of biomass accumulation. Measuring both parameters simultaneously, with a view to understanding how they interact, or for creating optical, surrogate indicators of Kc is very difficult because Kc itself is difficult to measure. In this study a portable enclosed chamber was used to measure ETc of a pasture and subsequently calculated Kc from reference evapotranspiration (ETo) data derived from a nearby automatic weather station (AWS). Calibration of the chamber confirms the suitability of the device to measure the amount of water vapour produced by local plant evapotranspiration, producing a calibration factor (C) close to 1 (C=1.02, R2=0.87). The coincident NDVI values were measured using a portable active optical sensor. In a test involving a pasture (Festuca arundinacea var. Dovey) at two different stages of growth in two consecutive growing seasons, the NDVI and crop coefficients were observed to be strongly correlated (R2=0.80 and 0.77, respectively). A polynomial regression (R2=0.84) was found to be the best fit for the combined, multi-temporal Kc-NDVI relationship. The main advantages of this method include the suitability of operating at a smaller scale (<1 m2), in real time and repeatability.

The fraction of absorbed photosynthetically active radiation (fAPAR) for plant canopies is often inferred from top-of-canopy, spectral reflectance, vegetation indices like the normalized difference vegetation index (NDVI). Such measures are derived using passive optical sensors and solar illumination of the canopy. However both the passive sensor-derived NDVI and the accompanying fAPAR measurements are affected by the solar elevation angle (θs). In many cases the effect of θs on both NDVI and fAPAR measurements is similar and the effect of θs is often cancelled out. The new class of active optical sensors (AOS) that contain their own radiant light sources to produce equivalent measurements of NDVI are not influenced by θs even though the accompanying values of fAPAR, as derived using a passive sensor are. This means the fAPAR–NDVIAOS relationship will invariably be sensitive to θs. By way of example, this paper investigates the correlations between the NDVIAOS and fAPAR under conditions of varying solar illumination angle for a tall fescue (Festuca arundinacea) pasture. The NDVIAOS was observed to retain a strong linear correlation with fAPAR (R2 ≥ 0.85) but fAPAR was highly sensitive to θs. Subsequently, simple models can be utilized to predict the fAPAR-NDVIAOS relationship for any solar elevation angle between 30 and 80°.

In current best commercial practice, pre-harvest fruit load predictions of mango orchards are provided based on a manual count of fruit number on up to 5% of trees within each block. However, the variability in fruit number per tree (coefficient of variation, CV, from 27 to 93% across ten orchards) was demonstrated to be such that the best case commercial sampling practice was inadequate for reliable estimation (to an error of 54–82 fruit/tree, and percentage error, PE, of 10% at a probability of 0.95). These results highlight the need for alternative methods for estimation of orchard fruit load. Pre-harvest fruit load was estimated for a case study orchard of 469 trees using (i) count of a sample of trees, (ii) in-field machine vision and (iii) correlation to a tree spectral index estimated using high resolution satellite imagery. A count of 5% of trees (23) in the trial orchard resulted in a PE of 31% (error of 37 fruit/tree), with a count of 157 trees required to achieve a PE of 10% (error of 12 fruit/tree). Sampling effort to achieve a PE of 10% was decreased by only 10% by sampling from aspatial k-means tree classifications based on machine vision derived fruit counts of all trees. Clustering based on tree attributes of canopy volume and trunk circumference was not helpful in decreasing sampling effort as these attributes were poorly correlated to fruit load (R² =0.21 and 0.17, respectively). In-field multi-view machine vision-based estimation of fruit load per tree achieved a R²= 0.97 and a RMSE = 14.8 fruit/tree against harvest fruit count per tree for a set of 18 trees (average = 88; SD = 82 fruit/tree), using a faster region convolutional neural network trained the previous season. The relationship between WorldView-3 (WV3) satellite spectral reflectance characteristics of sampled trees and fruit number was characterised by a R² = 0.66 and a RMSE = 56.1 fruit/tree. For this orchard, for which the actual fruit harvest was 56,720 fruit, the estimate based on a manual count of 5% of trees was 47,955 fruit, while estimates based on 20 iterations of stratified sampling (of 5% of trees in each cycle) had variation (SD) of 9597. The machine vision method resulted in an estimate of 53,520 (SD = 1960) fruit and the remote sensing method, 51,944 (SD = 26,300) fruit for the orchard.