The performance of a multi-view machine vision method was documented at an orchard level, relative to packhouse count. High repeatability was achieved in night-time imaging, with an absolute percentage error of 2% or less. Canopy architecture impacted performance, with reasonable estimates achieved on hedge, single leader and conventional systems (3.4, 5.0, and 8.2 average percentage error, respectively) while fruit load of trellised orchards was over-estimated (at 25.2 average percentage error). Yield estimations were made for multiple orchards via: (i) human count of fruit load on ~5% of trees (FARM), (ii) human count of 18 trees randomly selected within three NDVI stratifications (CAL), (iii) multi-view counts (MV-Raw) and (iv) multi-view corrected for occluded fruit using manual counts of CAL trees (MV-CAL). Across the nine orchards for which results for all methods were available, the FARM, CAL, MV-Raw and MV-CAL methods achieved an average percentage error on packhouse counts of 26, 13, 11 and 17%, with SD of 11, 8, 11 and 9%, respectively, in the 2019–2020 season. The absolute percentage error of the MV-Raw estimates was 10% or less in 15 of the 20 orchards assessed. Greater error in load estimation occurred in the 2020–2021 season due to the time-spread of flowering. Use cases for the tree level data on fruit load was explored in context of fruit load density maps to inform early harvesting and to interpret crop damage, and tree frequency distributions based on fruit load per tree.

Forecast of tree fruit yield requires prediction of harvest time fruit size as well as fruit number. Mango (Mangifera indica L.) fruit mass can be estimated from correlation to measurements of fruit length (L), width (W) and thickness (T). On-tree measurements of individually tagged fruit were undertaken using callipers at weekly intervals until the fruit were past commercial maturity, as judged using growing degree days (GDD), for mango cultivars ‘Honey Gold’, ‘Calypso’ and ‘Keitt’ at four locations in Australia and Brazil during the 2020/21 and 21/22 production seasons. Across all cultivars, the linear correlation of fruit mass to LWT was characterized by a R² of 0.99, RMSE of 29.9 g and slope of 0.5472 g/cm³, while the linear correlation of fruit mass to $L(\frac{\left(W+T\right)}{2}$)², mimicking what can be measured by machine vision of fruit on tree, was characterized by a R² of 0.97, RMSE of 25.0 g and slope of 0.5439 g/cm³. A procedure was established for the prediction of fruit size at harvest based on measurements made five and four or four and three weeks prior to harvest (approx. 514 and 422 GDD, before harvest, respectively). Linear regression models on weekly increase in fruit mass estimated from lineal measurements were characterized by an R² > 0.88 for all populations, with an average slope (rate of increase) of 19.6 ± 7.1 g/week, depending on cultivar, season and site. The mean absolute percentage error for predicted mass compared to harvested fruit weight for estimates based on measurements of the earlier and later intervals was 16.3 ± 1.3% and 4.5 ± 2.4%, respectively. Measurement at the later interval allowed better accuracy on prediction of fruit tray size distribution. A recommendation was made for forecast of fruit mass at harvest based on in-field measurements at approximately 400 to 450 GDD units before harvest GDD and one week later.

A forward estimate of mango (Mangifera indica L.) harvest timing is required for farm management (e.g., for organization of harvest labour and marketing). This forward estimate can be based on accumulated growing degree days (GDD) from an early stage of flowering to fruit harvest maturity, with fruit maturity judged on a destructive assessment of flesh colour and dry matter content. The current study was undertaken to improve GDD targets for Australian mango cultivars, to improve estimation of harvest maturity, and to document a methodology recommended for future work characterizing fruit maturation GDD for other mango cultivars. An alternate algorithm on GDD calculation involving use of a function that penalizes high temperatures as well as low temperatures was demonstrated to better predict harvest maturity in warmer climates. Across multiple locations and seasons, the required heat units (GDD, Tb = 12 °C, TB = 32 °C; where TB is upper base temperature of 32 °C and Tb is lower base temperature of 12 °C) to achieve maturity from asparagus stage of flowering was documented as 2185, 1728, and 1740 for the cultivars Keitt, Calypso and Honey Gold, respectively. GDD difference between the asparagus and two-thirds floral opening stages of flowering was 188 ± 18 for Calypso, 184 ± 12 for Honey Gold, 238 ± 21 for Keitt and 175 ± 10 for KP. Colour specifications for a colour card set suitable for maturity assessment of all cultivars was also proposed. A flesh colour harvest maturity card specification of 9 was proposed for the cultivar Honey Gold and 13 for the cultivar Keitt.

Spatially enabled yield forecasting is a key component of farm Management Information Systems (MISs) for broadacre grain production, enabling management decisions such as variable rate fertilization. However, such a capability has been lacking for soft (fleshy)-tree-fruit harvest load, with relevant tools for automated assessment having been developed only recently. Such tools include improved estimates of the heat units required for fruit maturation and in-field machine vision for flower and fruit count and fruit sizing. Feedback on the need for and issues in forecasting were documented. A mango ‘harvest forecast engine’ was designed for the forecasting of harvest timing and fruit load, to aid harvest management. Inputs include 15 min interval temperature data per orchard block, weekly manual or machine-vision-derived estimates of flowering, and preharvest manual or machine-vision-derived estimates of fruit load on an orchard block level across the farm. Outputs include predicted optimal harvest time and fruit load, on a per block and per week basis, to inform harvest scheduling. Use cases are provided, including forecast of the order of harvest of blocks within the orchard, management of harvest windows to match harvesting resources such as staff availability, and within block spatial allocation of resources, such as adequate placement of harvest field bin and frost fans. Design requirements for an effective harvest MIS software artefact incorporating the forecast engine are documented, including an integrated database supporting spatial query, data analysis, processing and mapping, an integrated geospatial database for managing of large spatial–temporal datasets, and use of dynamic web map services to enable rapid visualization of large datasets.

Forecasting the mango harvest window from flowering aids cultivar selection and harvest planning. This study evaluated the cross-site and cross-season generalizability of accumulated growing degree days (GDD) calculated using either a lower base temperature (Tb) alone or both lower and upper temperature thresholds (Tb and TB), relative to calendar days, for Australian mango cultivars. Operational values of Tb = 12 °C and TB = 32 °C were recommended for all assessed cultivars. The coefficient of variation across site-season populations for calendar days, GDD calculated using Tb alone, and GDD calculated using both Tb and TB was 13.5%, 2.5%, and 1.3%, respectively, confirming the robustness of the bounded GDD estimate across growing environments. Cultivar-specific GDD values associated with minimum and maximum fruit maturity were established using two complementary approaches. The first was based on ripening time (GDDr), with time-to-ripen at 20 °C durations of 12 and 6 days used to define lower and upper maturity boundaries (minGDDr and maxGDDr). The second was based on internal attributes at harvest (GDDi), including flesh color, dry matter (DM), soluble solids concentration (SSC), titratable acidity (TA), and flesh firmness. Minimum maturity was based on cultivar-specific flesh-color and DM targets and an SSC:TA criterion of 7.0. The mean firmness of cohorts at maxGDDr ranged from 2.2 to 3.6 kgf; a 3 kgf individual-fruit benchmark was therefore used for maxGDDi assessment. There was close agreement between GDDr and GDDi estimates of minimum maturity. The average of minGDDr and minGDDi for ‘B74’, ‘Honey Gold’, and ‘Keitt’ differed from previously published preliminary values by only 9–13 °C d, equivalent to approximately one calendar day, while values for ‘Nam Dok Mai’, ‘1201’, and ‘4069’ had not previously been estimated. Overall, minGDD values were established for nine cultivars, while complete harvest windows were resolved within the current study for four. For the remaining five cultivars, upper-boundary estimates were based on a single assessment approach, legacy observations, or remained unresolved.