The saturation property of vegetation indices posed a known limitation and this study was motivated to understand the saturation property of three widely used vegetation indices in mixed crop-forest ecosystem where limited knowledge existed. Normalized Difference Vegetation Index (NDVI), Simple Ratio Index (SRI) and Transformed Vegetation Index (TVI) were computed from sentinel-2 bands and; variations among bands and among vegetation indices were evaluated. The study employed green Leaf Area Index (gLAI) Version 1 product, derived from PROBA-V daily data for discriminating the saturation property of the indices. Although the study applied various methods of image preprocessing and processing, best curve fitting and correlation analysis were the key ones. The three vegetation indices: NDVI, SRI, and TVI computed from sentinel-2 bands: four (red) and five (red edge) coupled with bands 8 and 8a showed some levels of saturation. Nonetheless, TVI computed from bands 8a and 4 is the best outperforming combination, i.e., the least saturated one and it is an interesting output in a sense that a single index with significantly lower values of noise equivalent green Leaf Area Index as well as having strong association with gLAI is obtained that could be very useful for quantification of gLAI in similar ecosystems. For the rest of the bands and vegetation indices combination of the indices via setting thresholds could be one possible solution.
In this paper, spatial and statistical analysis methods were integrated to analyze the spatial accessibility, user population, and coverage of the Light Rail Transit of Addis Ababa. Stations were found to be the least accessible due to five types of land use and road network-related causes: government and social institutions with large area compounds, low road network density, open spaces devoid of road network, natural barriers such as water courses and large area public squares. The study explicitly revealed stations with overlaps in their service areas as well as stations having neighborhoods which are uncovered with the current service areas, i.e., 4.7 km2 and 19.9 km2 uncovered areas are found within walking distances of 1200–1500 m and 1500–2000 m, respectively; nonetheless, stations have fairly adequate coverage within 400 m and 800 m walking distance. Depending on the causes, stations were grouped into three categories. Service area-based determination of potential users is estimated to be 888,502, while the total number of the daily average actual users is around 67,624. Generally an increase in the accessible population is associated with an increase in the service area coverage. However, the population accessibility and service area coverage do not directly imply more number of actual users. Stations with large overlap of service area have reduced number of actual users due to the fact that the accessible population is divided between shared stations. Therefore, unnecessary large service area overlap is a wastage that does not contribute to more number of users. Future studies in the current study area might focus on considering the nexus between the accessibility and modal share as well as the long-term dynamics of the population of users.
Field-scale prediction methods that use remote sensing are significant in many global projects; however, the existing methods have several limitations. In particular, the characteristics of smallholder systems pose a unique challenge in the development of reliable prediction methods. Therefore, in this study, a fast and reproducible new approach to wheat prediction is developed by combining predictors derived from optical (Sentinel-2) and radar (Sentinel-1) sensors using a diverse set of machine learning and deep learning methods under a small dataset domain. This study takes place in the wheat belt region of Ethiopia and evaluates forty-two predictors that represent the major vegetation index categories of green, water, chlorophyll, dry biomass, and VH polarization SAR indices. The study also applies field-collected agronomic data from 165 farm fields for training and validation. According to results, compared to other methods, a combined automated machine learning (AutoML) approach with a generalized linear model (GLM) showed higher performance. AutoML, which reduces training time, delivered ten influential parameters. For the combined approach, the mean RMSE of wheat yield was from 0.84 to 0.98 ton/ha using ten predictors from the test dataset, achieving a 99% confidence interval. It also showed a correlation coefficient as high as 0.69 between the estimated yield and measured yield, and it was less sensitive to the small datasets used for model training and validation. A deep neural network with three hidden layers using the ten influential parameters was the second model. For this model, the mean RMSE of wheat yield was between 1.31 and 1.36 ton/ha on the test dataset, achieving a 99% confidence interval. This model used 55 neurons with respective values of 0.1, 0.5, and 1 × 10−4 for the hidden dropout ratio, input dropout ratio, and l2 regularization. The approaches implemented in this study are fast and reproducible and beneficial to predict yield at scale. These approaches could be adapted to predict grain yields of other cereal crops grown under smallholder systems in similar global production systems.
24 April 2025 in Environmental and Sustainability Indicators