Hassan Afzaal

Soil roughness is one of the most challenging issues in the agricultural domain and plays a crucial role in soil quality. The objective of this research was to develop a computerized method based on stereo vision technique to estimate the roughness formed on the agricultural soils. Additionally, soil till quality was investigated by analyzing the height of plow layers. An image dataset was provided in the real conditions of the field. For determining the soil surface roughness, the elevation of clods obtained from tillage operations was computed using a depth map. This map was obtained by extracting and matching corresponding keypoints as super pixels of images. Regression equations and coefficients of determination between the measured and estimated values indicate that the proposed method has a strong potential for the estimation of soil shallow roughness as an important physical parameter in tillage operations. In addition, peak fitting of tilled layers was applied to the height profile to evaluate the till quality. The results of this suggest that the peak fitting is an effective method of judging tillage quality in the fields.

This study evaluated the potential of using machine vision in combination with deep learning (DL) to identify the early blight disease in real-time for potato production systems. Four fields were selected to collect images (n = 5199) of healthy and diseased potato plants under variable lights and shadow effects. A database was constructed using DL to identify the disease infestation at different stages throughout the growing season. Three convolutional neural networks (CNNs), namely GoogleNet, VGGNet, and EfficientNet, were trained using the PyTorch framework. The disease images were classified into three classes (2-class, 4-class, and 6-class) for accurate disease identification at different growth stages. Results of 2-class CNNs for disease identification revealed the significantly better performance of EfficientNet and VGGNet when compared with the GoogleNet (FScore range: 0.84–0.98). Results of 4-Class CNNs indicated better performance of EfficientNet when compared with other CNNs (FScore range: 0.79–0.94). Results of 6-class CNNs showed similar results as 4-class, with EfficientNet performing the best. GoogleNet, VGGNet, and EfficientNet inference time values ranged from 6.8–8.3, 2.1–2.5, 5.95–6.53 frames per second, respectively, on a Dell Latitude 5580 using graphical processing unit (GPU) mode. Overall, the CNNs and DL frameworks used in this study accurately classified the early blight disease at different stages. Site-specific application of fungicides by accurately identifying the early blight infected plants has a strong potential to reduce agrochemicals use, improve the profitability of potato growers, and lower environmental risks (runoff of fungicides to water bodies).

Proximal sensing techniques can potentially survey soil and crop variables responsible for variations in crop yield. The full potential of these precision agriculture technologies may be exploited in combination with innovative methods of data processing such as machine learning (ML) algorithms for the extraction of useful information responsible for controlling crop yield. Four ML algorithms, namely linear regression (LR), elastic net (EN), k-nearest neighbor (k-NN), and support vector regression (SVR), were used to predict potato (Solanum tuberosum) tuber yield from data of soil and crop properties collected through proximal sensing. Six fields in Atlantic Canada including three fields in Prince Edward Island (PE) and three fields in New Brunswick (NB) were sampled, over two (2017 and 2018) growing seasons, for soil electrical conductivity, soil moisture content, soil slope, normalized-difference vegetative index (NDVI), and soil chemistry. Data were collected from 39–40 30 × 30 m² locations in each field, four times throughout the growing season, and yield samples were collected manually at the end of the growing season. Four datasets, namely PE-2017, PE-2018, NB-2017, and NB-2018, were then formed by combing data points from three fields to represent the province data for the respective years. Modeling techniques were employed to generate yield predictions assessed with different statistical parameters. The SVR models outperformed all other models for NB-2017, NB-2018, PE-2017, and PE-2018 dataset with RMSE of 5.97, 4.62, 6.60, and 6.17 t/ha, respectively. The performance of k-NN remained poor in three out of four datasets, namely NB-2017, NB-2018, and PE-2017 with RMSE of 6.93, 5.23, and 6.91 t/ha, respectively. The study also showed that large datasets are required to generate useful results using either model. This information is needed for creating site-specific management zones for potatoes, which form a significant component for food security initiatives across the globe.

Agricultural management practices are responsible for almost two-thirds of the variations in potato tuber yield. In order to answer the research question about the remaining variability of the tuber yield, we hypothesized that climate extremes partly explain the missing component of variations of the tuber yield. Therefore, this research attempts to bridge this knowledge gap in order to generate a knowledge base for future strategies. A climate extreme dataset of the Prince Edward Island (PEI) was computed by averaging the data of five meteorological stations. In detail, changing patterns of 20 climate extreme indices were computed with ClimPACT2 software for 30 years (1989-2018) data of PEI. Statistical significance of the trends and their slope values were determined with the Mann-Kendall test and Sen’s slope estimates, respectively. Average of daily mean temperature (TMm), mean daily minimum temperature (TNm) and the occurrence of continuous dry days (CDD), significantly increased by 0.77 °C, 1.17 °C and 3.33 days., respectively, during the potato growing seasons (May-October) of the past three decades. For this period daily temperature range (DTR), frost days (FD), cold days (TX10p), cold nights (TN10p) and warmest days (TXx) showed decreasing trends of −1.01 °C, −3.75 days, −5.67 days, −11.40 nights, and −2.00 days, respectively. The principal component analysis showed that DTR, TXx, CDD, and TNm were the main factors affecting seasonal variations of tuber yield. The multiple regression model attributed ~39% of tuber yield variance to DTR, TXx, CDD, and TNm. However, these indices explained individually 21%, 19%, 16%, and 4% variation to the tuber yield, respectively. The remaining variation in the tuber yield explained by other yield affecting factors. The information generated from this study can be used for future planning about agricultural management strategies in the Island, for example, the provision of water resources for supplemental irrigation of crops during dry months.

Climate change induced uneven patterns of rainfall emphasize the use of supplemental irrigation in rainfed agriculture. The Penman–Monteith method was used to calculate supplemental irrigation for water budgeting of a potato crop in Prince Edward Island, Canada. Cumulative gaps between rainfall and crop evapotranspiration (ETc) during August and September of the study years were due to high crop coefficient factor, justifying the need for supplemental irrigation. Pressurized irrigation systems, including sprinklers, fertigation, and drip irrigation were installed, to evaluate the impact of scheduled supplemental irrigation in offsetting deficits in irrigation water requirements in comparison with conventional practice of rainfed cultivation (control). A two-way ANOVA examined the effect of irrigation methods and year on potato tuber yield, water productivity, tuber quality, and payout. Sprinkler and fertigation systems performed better than drip and control treatments. In terms of payout returns and potato tuber quality (percentage of marketable potatoes), the sprinkler treatment performed significantly better than the other treatments. However, for water productivity, fertigation treatment performed significantly better than control and sprinkler treatments during both years. The use of supplemental irrigation is recommended for profitable cultivation of potatoes in soil, agricultural, and environmental conditions resembling to those of Prince Edward Island.

Information about potential scenarios and causes of floods is important for future planning. Historical weather data of Fredericton (New Brunswick) and Charlottetown (Prince Edward Island), the two coastal cities of Atlantic Canada, were analyzed using RClimDex, Mann–Kendall test, and Sen’s slope estimates for potential scenarios and causes of floods. Flood hazard analyses were conducted using GIS (Geographical Information System) and ArcSWAT software. The watersheds of Fredericton and Charlottetown were delineated from 25 × 25 m resolution DEMs (Digital Elevation Models) of the two cities followed by percent watershed area calculations for different elevation classes for flood generation. Over the past 100 years, there was a significant decreasing trend in the high intensity precipitation in Charlottetown supported by a significant decrease in the number of heavy precipitation days. However, maximum one-day precipitation and maximum five-day precipitation significantly increased in Charlottetown and Fredericton, respectively. Charlottetown received more annual precipitation than Fredericton. In the last 30 years, there was an event exceeding 50 mm precipitation (considered as a threshold for the return period of urban floods) in Charlottetown; Fredericton experienced such events for more than 1.5 times. For twelve times, these events occurred more than once in a year in Charlottetown as compared to fourteen times in Fredericton. Despite statistically proven similarities in the occurrence of extreme events in the two cities, the visualized flood hazards, and the mapping of watershed characteristics, no devastating floods were reported for Charlottetown. This does not necessarily mean that there had never been risks of flooding in Charlottetown. These findings may help policymakers for future developments.

Accurate estimation of reference evapotranspiration (ETo) provides useful information for water resource management and sustainable agriculture. This study estimates ETo with recurrent neural networks (RNNs), namely long short-term memory (LSTM) and bidirectional LSTM. Four representative meteorological sites (North Cape, Summerside, Harrington, and Saint Peters) were selected across Prince Edward Island (PEI), Canada to form a PEI dataset from mean values of the four sites’ climatic variables for capturing climatic variability from all parts of the province. Based on subset regression analysis, the highest contributing climatic variables, namely maximum air temperature and relative humidity, were selected as input variables for RNNs’ training (2011–2015) and testing (2016–2017) runs. The results suggested that the LSTM and bidirectional LSTM are suitable methods to accurately (R² > 0.90) estimate ETo for all sites except Harrington. Testing period (2016–2017) root mean square errors were recorded in range of 0.38–0.58 mm/day for all sites. No major differences were observed in accuracy of LSTM and bidirectional LSTM. Another objective of this study was to highlight the potential gap between ET_O and rainfall for assessing agriculture sustainability in Prince Edward Island. Analyses of the data highlighted that the cumulative ETo surpassed the cumulative rainfall potentially affecting yield of major crops in the island. Therefore, agriculture sustainability requires viable options such as supplemental irrigation to replenish the crop water requirements as and when needed.

Precise estimation of physical hydrology components including groundwater levels (GWLs) is a challenging task, especially in relatively non-contiguous watersheds. This study estimates GWLs with deep learning and artificial neural networks (ANNs), namely a multilayer perceptron (MLP), long short term memory (LSTM), and a convolutional neural network (CNN) with four different input variable combinations for two watersheds (Baltic River and Long Creek) in Prince Edward Island, Canada. Variables including stream level, stream flow, precipitation, relative humidity, mean temperature, evapotranspiration, heat degree days, dew point temperature, and evapotranspiration for the 2011–2017 period were used as input variables. Using a hit and trial approach and various hyperparameters, all ANNs were trained from scratched (2011–2015) and validated (2016–2017). The stream level was the major contributor to GWL fluctuation for the Baltic River and Long Creek watersheds (R² = 50.8 and 49.1%, respectively). The MLP performed better in validation for Baltic River and Long Creek watersheds (RMSE = 0.471 and 1.15, respectively). Increased number of variables from 1 to 4 improved the RMSE for the Baltic River watershed by 11% and for the Long Creek watershed by 1.6%. The deep learning techniques introduced in this study to estimate GWL fluctuations are convenient and accurate as compared to collection of periodic dips based on the groundwater monitoring wells for groundwater inventory control and management.