Optimizing machine learning algorithms for spatial prediction of gully erosion susceptibility with four training scenarios.

Gully erosion causes high soil erosion rates and is an environmental concern posing major risk to the sustainability of cultivated areas of the world. Gullies modify the land, shape new landforms, and damage agricultural fields. Gully erosion mapping is essential to understand the mechanism, development, and evolution of gullies. In this work, a new modeling approach was employed for gully erosion susceptibility mapping (GESM) in the Golestan Dam basin of Iran. The measurements of 14 gully erosion (GE) factors at 1042 GE locations were compiled in a spatial database. Four training datasets comprised of 100%, 75%, 50%, and 25% of the entire database were used for modeling and validation (for each data set in the common 70:30 ratio). Four machine learning models-maximum entropy (MaxEnt), general linear model (GLM), support vector machine (SVM), and artificial neural network (ANN)- were employed to check the usefulness of the four training scenarios. The results of random forest (RF) analysis indicated that the most important GE effective factors were distance from the stream, elevation, distance from the road, and vertical distance of the channel network (VDCN). The receiver operating characteristic (ROC) was used to validate the results. Our study showed that the sample size influenced the performance of the four machine learning algorithms. However, the ANN had a lower sensitivity to the reduction of sample size. In addition, validation results revealed that ANN (AUROC = 0.85.7-0.90.4%) had the best performance based on all four sample data sets. The results of this research can be useful and valuable guidelines for choosing machine learning methods when a complete gully inventory is not available in a region.

Identifying sources of dust aerosol using a new framework based on remote sensing and modelling.

Dust particles are transported globally. Dust storms can adversely impact both human health and the environment, but they also impact transportation infrastructure, agriculture, and industry, occasionally severely. The identification of the locations that are the primary sources of dust, especially in arid and semi-arid environments, remains a challenge as these sites are often in remote or data-scarce regions. In this study, a new method using state-of-the-art machine-learning algorithms - random forest (RF), support vector machines (SVM), and multivariate adaptive regression splines (MARS) - was evaluated for its ability to spatially model the distribution of dust-source potential in eastern Iran. To accomplish this, empirically identified dust-source locations were determined with the ozone monitoring instrument aerosol index and the Moderate-Resolution Imaging Spectroradiometer (MODIS) Deep Blue aerosol optical thickness methods. The identified areas were divided into training (70%) and validation (30%) sets. Measurements of the conditioning factors (lithology, wind speed, maximum air temperature, land use, slope angle, soil, rainfall, and land cover) were compiled for the study area and predictive models were developed. The area-under-the-receiver operating characteristics curve (AUC) and true-skill statistics (TSS) were used to validate the maps of the models' predictions. The results show that the RF algorithm performed best (AUC = 89.4% and TSS = 0.751), followed by the SVM (AUC = 87.5%, TSS = 0.73) and the MARS algorithm (AUC = 81%, TSS = 0.69). The results of the RF indicated that wind speed and land cover are the most important factors affecting dust generation. The region of highest dust-source potential that was identified by the RF is in the eastern parts of the study region. This model can be applied to other arid and semi-arid environments that experience dust storms to promote management that prevents desertification and reduces dust production.

Environmental filtering, predominance of strong competitor trees and exclusion of moderate-weak competitor trees shape species richness and biomass.

Strong competitor (i.e. big-sized) trees are globally crucial for promoting aboveground biomass. Still, we do not fully understand the simultaneous influences of different levels of competitor (i.e. strong, moderate, medium and weak) trees at stand level in shaping forest diversity and biomass along a climatic gradient. We hypothesized that few strong competitor trees shape the positive relationship between tree species richness and aboveground biomass better than moderate, medium and weak competitor trees along a climatic gradient. Using the forest inventory data (i.e. tree diameter, height and crown diameter), we quantified strong (i.e. 99th percentile; top 1%), moderate (i.e. 75th percentile; top 25%), medium (i.e. 50th percentile) and weak (i.e. 25th percentile) competitor trees as well as species richness and aboveground biomass of 248 plots (moist temperate, semi-humid, and semi-arid forests) across 12 sites in Iran. The main results from three piecewise structural equation models (i.e. tree diameter, height and crown based models) showed that, after considering the simultaneous fixed effects of climate and random effects of sites or forest types variation, strong competitor trees possessed strong positive effects on tree species richness and biomass whereas moderate, medium and weak competitor trees possessed negligible positive to negative effects. Also, different levels of competitor trees promoted each other in a top-down way but the effects of strong competitor trees on moderate, medium and weak competitor trees were relatively weak. This study suggests that the simultaneous interactions of different tree sizes at stand level across forest sites should be included in the integrative ecological modeling for better understanding the role of different levels of competitor trees in shaping positive forest diversity - functioning relationship in a changing environment.

Impacts of climatic and edaphic factors on the diversity, structure and biomass of species-poor and structurally-complex forests.

Understanding the impacts of multiple climatic and edaphic factors on forest diversity, structure and biomass is crucial to predicting how forests will react to global environmental change. Here, we addressed how do forest structural attributes (i.e. top 1% big, top 25% big medium and small trees; in terms of tree height, diameter, and crown), species richness, and aboveground biomass respond to temperature-related and water-related climatic factors as well as to edaphic factors. By assuming disturbance as a constant factor in the study forests, we hypothesize that water-related and temperature-related climatic factors play contrasting roles whereas edaphic factors play an additional role in shaping forest diversity, structure and aboveground biomass in species-poor and structurally-complex forests. We used forest inventory and environmental factors data from 248 forest plots (moist temperate, semi-humid, and semi-arid) across 12 sites in Iran. We developed multiple linear mixed-effect models for each response variable by using multiple climatic and edaphic factors as fixed effects whereas sites as a random effect. Top 1% big, top 25% big, medium, and small trees enhanced with mean annual temperature but declined with water-related climatic (i.e. mean annual precipitation, cloud cover, potential evapotranspiration, and wet day frequency) factors, whereas soil texture (i.e. sand content) and pH were of additional importance. Species richness increased with precipitation and cloud cover but decreased with temperature, potential evapotranspiration, soil fertility and sand content. Aboveground biomass increased along temperature gradient but decreased with potential evapotranspiration, clay and sand contents. Temperature seemed to be the main driver underlying the increase in forest structure (i.e. diameter-related attributes) and biomass whereas precipitation did so for species richness. We argue that the impacts of multiple climatic factors on forest structural attributes, diversity and biomass should be properly evaluated in order to better understand the responses of species-poor forests to climate change.

Data on assessment of corrosion-scaling potential and chemical parameters of groundwater quality for industrial and agricultural sectors in the Piranshahr Watershed in the West Azerbaijan province, Iran.

The present work sets out to 1) evaluate the corrosion-scaling potential of groundwater resources for the industrial sector as well as to 2) examine groundwater chemical parameters for the agricultural sector in the Piranshahr Watershed in the West Azerbaijan province, Iran, using geostatistical analyses and the Wilcox diagram in a GIS environment. A total of 145 spring locations as representatives of groundwater potentiality were recorded by a handheld GPS device and the corrosion and scaling potential states were further scrutinized. The latter was carried out on the basis of examining the chemical parameters at each sample location including alkalinity, pH, temperature, Na+, Ca++, Mg++, TH, HCO 3 - , Co 3 - 2 , Cl - , SO 4 - 2 , Electrical conductivity, and Total dissolved solids. The corrosion and scaling potential of groundwater was then evaluated by using Langelier saturation index (LSI), Larson-Skold index (LS), Ryznar stability index (RSI), Aggressive index (AI), and Puckorius scaling index (PSI). Also, the groundwater quality state for agriculture was assessed by the Wilcox diagram on the basis of Electrical conductivity and Sodium adsorption ratio parameters. The provided data are beneficial for researchers, policymakers, and authorities for taking pragmatic actions. Also, the compiled data can be used in the context of corrosion/scaling and groundwater quality assessment can be generalized around the world.