Mapping smallholder fruit plantations using optical data is challenging due to morphological landscape heterogeneity and crop types having overlapping spectral signatures. Furthermore, cloud covers limit the use of optical sensing, especially in subtropical climates where they are persistent. This research assessed the effectiveness of Sentinel-1 (S1) and Sentinel-2 (S2) data for mapping fruit trees and co-existing land-use types by using support vector machine (SVM) and random forest (RF) classifiers independently. These classifiers were also applied to fused data from the two sensors. Feature ranks were extracted using the RF mean decrease accuracy (MDA) and forward variable selection (FVS) to identify optimal spectral windows to classify fruit trees. Based on RF MDA and FVS, the SVM classifier resulted in relatively high classification accuracy with overall accuracy (OA) = 0.91.6% and kappa coefficient = 0.91% when applied to the fused satellite data. Application of SVM to S1, S2, S2 selected variables and S1S2 fusion independently produced OA = 27.64, Kappa coefficient = 0.13%; OA= 87%, Kappa coefficient = 86.89%; OA = 69.33, Kappa coefficient = 69. %; OA = 87.01%, Kappa coefficient = 87%, respectively. Results also indicated that the optimal spectral bands for fruit trees mapping are green (B3) and SWIR_2 (B10) for S2, whereas for S1, the vertical-horizontal (VH) polarization band. Including the textural metrics from the VV channel improved crop discrimination and co-existing land use cover types. The fusion approach proved robust and well suited for accurate smallholder fruit plantation mapping.

Fruit-tree crops generate food and income for local households and contribute to South Africa’s gross domestic product. Timely and accurate phenotyping of fruit-tree crops is essential for innovating and achieving precision agriculture in the horticulture industry. Traditional methods for fruit-tree crop classification are time-consuming, costly, and often impossible to use for mapping heterogeneous horticulture systems. The application of remote sensing in smallholder agricultural landscapes is more promising. However, intercropping systems coupled with the presence of dispersed small agricultural fields that are characterized by common and uncommon crop types result in imbalanced samples, which may limit conventionally applied classification methods for phenotyping. This study assessed the influence of balanced and imbalanced multi-class distribution and data-sampling techniques on fruit-tree crop detection accuracy. Seven data samples were used as input to adaptive boosting (AdaBoost), gradient boosting (GB), random forest (RF), support vector machine (SVM), and eXtreme gradient boost (XGBoost) machine learning algorithms. A pixel-based approach was applied using Sentinel-2 (S2). The SVM algorithm produced the highest classification accuracy of 71%, compared with AdaBoost (67%), RF (65%), XGBoost (63%), and GB (62%), respectively. Individually, the majority of the crop types were classified with an F1 score of between 60% and 100%. In addition, the study assessed the effect of size and ratio of class imbalance in the training datasets on algorithms’ sensitiveness and stability. The results show that the highest classification accuracy of 71% could be achieved from an imbalanced training dataset containing only 60% of the original dataset. The results also showed that S2 data could be successfully used to map fruit-tree crops and provide valuable information for subtropical crop management and precision agriculture in heterogeneous horticultural landscapes.

There is considerable interest in accurately estimating water quality parameters in turbid (Case 2) and eutrophic waters such as the Western Basin of Lake Erie (WBLE). Lake Erie is a large, open freshwater body that supports diverse ecosystem, and over 12 million people in the mid-western part of the United States depend on it for drinking water, fisheries, navigational, and recreational purposes. The increasing utilization of the freshwater has deteriorated the water severely and currently the lake is experiencing recurring harmful algal blooms (HABs). Improving the water quality of Lake Erie requires the use of robust monitoring tools that help water quality managers understand sources and pathways of influxes that trigger HABs. Satellite-based remote sensing sensor such as the moderate resolution imaging spectroradiometer (MODIS) may provide frequent and synoptic view of the water quality indices. In this study, data set from field measurements was used to evaluate the performance of 14 existing ocean color algorithms. Results indicated that MODIS data consistently underestimated the chlorophyll a concentrations in the WBLE, with the largest source of errors from dissolved organic matter and xanthophyll accessory pigments in this data set. Most of the global algorithms, including OC4v4 and the Baltic model, generated near-identical statistical parameters with an average R² of ~0.57 and RMSE ~2.9 μg/l. MODIS performed poorly (R² ~0.18) when its NIR/red bands were used. A slightly improved model was developed using similar band ratio approach generating R² of ~0.62 and RMSE ~1.8 μg/l.

The complex composition and distribution of colour producing agents (CPAs) in turbid aquatic environments such as the Western Basin of Lake Erie (WBLE) presents a challenge to the application of remote sensing data for differentiating among in-water constituents and estimating their concentrations independently. In this study, multivariate procedures are applied to lab-based spectrophotometer data to estimate the concentration of chlorophyll-a and suspended matters in the WBLE. Principal Component Analysis of first-derivative transformed hyper-spectral data from the spectrophotometer extracted three significant spectral components for each cruise, explaining up to 88% of the spectral variability. Spectral matching using reference spectra indicated that two of the extracted patterns represent signatures of in-water constituents that govern the optical properties of the WBLE, namely, cyanobacteria and diatoms associated with green algae. The spectrophotometer data clearly revealed known spectral features associated with phytoplankton, such as the absorption minima near 550 and 700 nm, which can be attributed to the minimum of absorption and fluorescence of chlorophyll-a, respectively. The method also extracted the absorption peaks due to chlorophyll-a, near 670 nm, and due to phycocyanin, near 620 nm. Principal component regression of chlorophyll-a on the PC scores indicated that 63.4% of variation of chlorophyll-a in the WBLE can be explained by two components. Factors 2 and 3 explain 60% of the joint spatiotemporal variability of suspended matters in the WBLE. The results illustrate the potential of multivariate technique applied to remote sensing data in isolating the patterns that represent constituents in turbid Case 2 waters.

Understanding long-term land use/land cover (LULC) change patterns is vital to implementing policies for effective environmental management practices and sustainable land use. This study assessed patterns of change in LULC in the Vaal Dam Catchment area, one of the most critically important areas in South Africa, since it contributes a vast portion of water to the Vaal Dam Reservoir. The reservoir has been used to supply water to about 13 million inhabitants in Gauteng province and its surrounding areas. Multi-temporal Landsat imagery series were used to map LULC changes between 1986 and 2021. The LULC classification was performed by applying the random forest (RF) algorithm to the Landsat data. The change-detection analysis showed grassland being the dominant land cover type (ranging from 52% to 57% of the study area) during the entire period. The second most dominant land cover type was agricultural land, which included cleared fields, while cultivated land covered around 41% of the study area. Other land use types covering small portions of the study area included settlements, mining activities, water bodies and woody vegetation. Time series analysis showed patterns of increasing and decreasing changes for all land cover types, except in the settlement class, which showed continuous increase owing to population growth. From the study results, the settlement class increased considerably for 1986–1993, 1993–2000, 2000–2007, 2007–2014 and 2014–2021 by 712.64 ha (0.02%), 10245.94 ha (0.26%), 3736.62 ha (0.1%), 1872.09 ha (0.05%) and 3801.06 ha (0.1%), respectively. This study highlights the importance of using remote sensing techniques in detecting LULC changes in this vitally important catchment.