The quantification of aboveground biomass using remote sensing is critical for better understanding the role of forests in carbon sequestration and for informed sustainable management. Although remote sensing techniques have been proven useful in assessing forest biomass in general, more is required to investigate their capabilities in predicting intra-and-inter species biomass which are mainly characterised by non-linear relationships. In this study, we tested two machine learning algorithms, Stochastic Gradient Boosting (SGB) and Random Forest (RF) regression trees to predict intra-and-inter species biomass using high resolution RapidEye reflectance bands as well as the derived vegetation indices in a commercial plantation. The results showed that the SGB algorithm yielded the best performance for intra-and-inter species biomass prediction; using all the predictor variables as well as based on the most important selected variables. For example using the most important variables the algorithm produced an R² of 0.80 and RMSE of 16.93 t·ha⁻¹ for E. grandis; R² of 0.79, RMSE of 17.27 t·ha⁻¹ for P. taeda and R² of 0.61, RMSE of 43.39 t·ha⁻¹ for the combined species data sets. Comparatively, RF yielded plausible results only for E. dunii (R² of 0.79; RMSE of 7.18 t·ha⁻¹). We demonstrated that although the two statistical methods were able to predict biomass accurately, RF produced weaker results as compared to SGB when applied to combined species dataset. The result underscores the relevance of stochastic models in predicting biomass drawn from different species and genera using the new generation high resolution RapidEye sensor with strategically positioned bands.

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.

Phaeosphaeria leaf spot (PLS) is considered one of the major diseases that threaten the stability of maize production in tropical and subtropical African regions. The objective of the present study was to investigate the use of hyperspectral data in detecting the early stage of PLS in tropical maize. Field data were collected from healthy and the early stage of PLS over two years (2013 and 2014) using a handheld spectroradiometer. An integration of a newly developed guided regularized random forest (GRRF) and a traditional random forest (RF) was used for feature selection and classification, respectively. The 2013 dataset was used to train the model, while the 2014 dataset was used as independent test dataset. Results showed that there were statistically significant differences in biochemical concentration between the healthy leaves and leaves that were at an early stage of PLS infestation. The newly developed GRRF was able to reduce the high dimensionality of hyperspectral data by selecting key wavelengths with less autocorrelation. These wavelengths are located at 420 nm, 795 nm, 779 nm, 1543 nm, 1747 nm, and 1010 nm. Using these variables (n=6), a random forest classifier was able to discriminate between the healthy maize and maize at an early stage of PLS infestation with an overall accuracy of 88% and a kappa value of 0.75. Overall, our study showed potential application of hyperspectral data, GRRF feature selection, and RF classifiers in detecting the early stage of PLS infestation in tropical maize.

Shoreline analysis in response to the rapid erosion of sandy beaches has evolved along with geospatial and computer technology; it remains an essential task for sustainable coastal management. This severe and rapid erosion has been reported at several sandy beaches worldwide, including Yzerfontein beaches, on the southwest coast of South Africa. We determined this vulnerability from 1937 to 2020 and predicted its change by 2040 by manually delineating shoreline positions from 1937, 1960, and 1977 from aerial photographs and Landsat products between 1985 and 2020 in an automated fashion using the CoastSat toolkit and Google Earth Engine. We then integrated these datasets to calculate the extent of shoreline dynamics over the past eight decades using the Digital Shoreline Analysis System (DSAS). Our results show that the coastline changed dynamically between 1937 and 2020, culminating in an average net erosion of 38 m, with the most extensive erosion occurring between 2015 and 2020. However, coastal projections indicate a slight change in shoreline position over the next two decades. Further studies should integrate additional high resolution remote sensing data and non-remote sensing data (e.g., field surveys) to improve our results and provide a more thorough understanding of the coastal environment and overcome some of remotely-sensed data underlying uncertainties.

Leaf area index (LAI) is an important biophysical trait for forest ecosystem and ecological modeling, as it plays a key role for the forest productivity and structural characteristics. The ground-based methods like the handheld optical instruments for predicting LAI are subjective, pricy and time-consuming. The advent of very high spatial resolutions multispectral data and robust machine learning regression algorithms like support vector machines (SVM) and artificial neural networks (ANN) has provided an opportunity to estimate LAI at tree species level. The objective of the this study was therefore to test the utility of spectral vegetation indices (SVI) calculated from the multispectral WorldView-2 (WV-2) data in predicting LAI at tree species level using the SVM and ANN machine learning regression algorithms. We further tested whether there are significant differences between LAI of intact and fragmented (open) indigenous forest ecosystems at tree species level. The study shows that LAI at tree species level could accurately be estimated using the fragmented stratum data compared with the intact stratum data. Specifically, our study shows that the accurate LAI predictions were achieved for Hymenocardia ulmoides using the fragmented stratum data and SVM regression model based on a validation dataset (R²_Val = 0.75, RMSE_Val = 0.05 (1.37% of the mean)). Our study further showed that SVM regression approach achieved more accurate models for predicting the LAI of the six endangered tree species compared with ANN regression method. It is concluded that the successful application of the WV-2 data, SVM and ANN methods in predicting LAI of six endangered tree species in the Dukuduku indigenous forest could help in making informed decisions and policies regarding management, protection and conservation of these endangered tree species.

Soil organic carbon constitutes an important indicator of soil fertility. The purpose of this study was to predict soil organic carbon content in the mountainous terrain of eastern Lesotho, southern Africa, which is an area of high endemic biodiversity as well as an area extensively used for small-scale agriculture. An integrated field and laboratory approach was undertaken, through measurements of reflectance spectra of soil using an Analytical Spectral Device (ASD) FieldSpec® 4 optical sensor. Soil spectra were collected on the land surface under field conditions and then on soil in the laboratory, in order to assess the accuracy of field spectroscopy-based models. The predictive performance of two different statistical models (random forest and partial least square regression) was compared. Results show that random forest regression can most accurately predict the soil organic carbon contents on an independent dataset using the field spectroscopy data. In contrast, the partial least square regression model overfits the calibration dataset. Important wavelengths to predict soil organic contents were localised around the visible range (400–700 nm). This study shows that soil organic carbon can be most accurately estimated using derivative field spectroscopy measurements and random forest regression.