cv

PhD Student

Faculty of Soil Science & Geomorphology, University of Tübingen

Soil Erosion modeling using Deep Learning and Remote Sensing in Precision Agriculture.

• Soil Erosion
• Deep Learning
• Remote Sensing
• Precision Agriculture

PhD

University of Tübingen, Tübingen, Germany

Soil Science

Rapid flood mapping from aerial imagery using fine-tuned SAM and ResNet-backboned U-Net

Hydrology and Earth System Sciences

This work presents a fine-tuned SAM and ResNet-based U-Net approach for rapid and accurate flood mapping using aerial imagery.

A GLUE-based assessment of WaTEM/SEDEM for simulating soil erosion, transport, and deposition in soil conservation optimised agricultural watersheds

SOIL

This study assesses the performance of WaTEM/SEDEM using a GLUE-based uncertainty framework for simulating soil erosion, transport, and deposition in optimized agricultural watersheds.

Soil pH and latitude as a major predictor of C:N:P stoichiometry in Germany

CATENA

This study identifies soil pH and latitude as key predictors of soil C:N:P stoichiometry in Germany using advanced machine learning and statistical modeling.

Comparing UAV-Based Hyperspectral and Satellite-Based Multispectral Data for Soil Moisture Estimation Using Machine Learning

Water

This paper compares UAV hyperspectral and satellite multispectral data for soil moisture estimation using various machine learning algorithms.

Erosion-SAM: Semantic segmentation of soil erosion by water

CATENA

A deep learning framework, Erosion-SAM, is developed for semantic segmentation of soil erosion features from remote sensing imagery.

Random Forest-Based Soil Moisture Estimation Using Sentinel-2, Landsat-8/9, and UAV-Based Hyperspectral Data

Remote Sensing

A random forest model is developed to estimate soil moisture using data from Sentinel-2, Landsat-8/9, and UAV hyperspectral imagery.

Assessing the Role of Environmental Covariates and Pixel Size in Soil Property Prediction: A Comparative Study of Various Areas in Southwest Iran

Land

This paper investigates the effects of pixel size and environmental covariates on soil property prediction performance.

Assessment of Land Suitability Potential Using Ensemble Approaches of Advanced Multi-Criteria Decision Models and Machine Learning for Wheat Cultivation

Remote Sensing

An ensemble approach combining advanced decision models and machine learning to assess land suitability for wheat cultivation.

BIPE: A Bi-Layer Predictive Ensemble Framework for Forest Fire Susceptibility Mapping in Germany

Remote Sensing

This study introduces BIPE, a bi-layer predictive ensemble framework for mapping forest fire susceptibility across Germany.

Assessing soil moisture levels using visible UAV imagery and machine learning models

Remote Sensing Applications: Society and Environment

This paper explores the potential of visible UAV imagery combined with ML models for soil moisture estimation.

Evaluating the Accuracy of Precipitation Products Over Utah, United States, Using the Google Earth Engine Platform

Desert

The study evaluates various satellite precipitation products over Utah using Google Earth Engine analysis.

Optimization and Feasibility Study of using Rainwater Harvesting Systems in Ardabil

Journal of Water and Soil

This Persian-language study evaluates the feasibility and optimization of rainwater harvesting systems in Ardabil.

Estimation of Evapotranspiration in Landscape by WUCOLS, PF and IPOS Methods

Journal of Water and Soil

A comparative estimation of landscape evapotranspiration using WUCOLS, PF, and IPOS methods. (In Persian)

Reliability, Optimization and Economic Analysis of the Rainwater Harvesting System from the Rooftop

Iranian Journal of Irrigation and Water Engineering

A study on the reliability, optimization, and cost-effectiveness of rooftop rainwater harvesting systems. (In Persian)

Designing of Rainwater Harvesting Systems Using Drone Images

Engineering Journal

This study uses drone images to design and optimize rainwater harvesting systems. (In Persian)

Assessing reliability of rainwater harvesting systems for meeting water demands in different climatic zones of Iran

Modeling Earth Systems and Environment

This paper evaluates the reliability of rainwater harvesting systems across different Iranian climatic zones.

Urban runoff management in sustainable development and the environment

Journal of Environmental Science Studies

This Persian-language paper discusses urban runoff management strategies for sustainable development.

Erosion-SAM

The Erosion-SAM project focuses on the automatic detection of soil erosion and deposition in aerial imagery. After heavy rainfall events, high-resolution aerial images were captured over affected areas. Erosion and deposition features were then manually segmented. These annotated images were used to fine-tune the Segment Anything Model (SAM) through transfer learning, addressing the common issue of limited labeled data in soil erosion studies. The resulting Erosion-SAM model can accurately identify erosion and deposition features from aerial images without additional training. Its outputs are valuable for training and validating machine learning models, as well as for environmental monitoring, natural resource management, and insurance assessments.

• Fine-tuning
• Transfer Learning
• Soil Erosion
• Segment Anything Model (SAM)

Soil Erosion Modeling

This project employs a hybrid modeling approach to estimate soil erosion. First, erosion and deposition are modeled using the WATSEDM framework. The resulting outputs, along with additional datasets, are then used as inputs for a deep learning model. For validation, outputs from our previous project, (Erosion-SAM) are utilized. This approach can provide a more accurate and robust prediction of soil erosion patterns by combining the strengths of physically-based models with the adaptive capabilities of machine learning.

• Soil Erosion
• Hybrid Modeling
• WaTEM/SEDEM
• Deep Learning

Rainfall Erosivity Forecasting

In this project, we investigate the relationship between rainfall data and monthly rainfall erosivity for the period with available high-resolution precipitation records. We then extend this relationship backward in time to reconstruct historical erosivity. Using this long-term time series, we forecast future erosivity patterns with a ConvLSTM model. This approach allows for a better understanding of past trends, provides a foundation for predicting future soil erosion risks, and supports more informed land management and conservation planning.

• Rainfall Erosivity
• Time Series Analysis
• ConvLSTM
• Forecasting

Rapid Flood Mapping

The Rapid Flood Mapping project compares two deep learning approaches for detecting flooded areas from aerial imagery: a fine-tuned Segment Anything Model (SAM) and a U-Net with a ResNet backbone. High-resolution aerial images captured shortly after major flood events were used to train and evaluate both models. This project demonstrates the potential of AI-driven flood mapping for rapid disaster assessment, emergency response planning, infrastructure risk analysis, and environmental monitoring, enabling authorities and organizations to make faster and more informed decisions in the aftermath of floods.

• Fine-tuning
• Transfer Learning
• Flood
• Segment Anything Model (SAM)
• U-Net
• ResNet

Soil Moisture Estimation

This project compares the effectiveness of UAV-based hyperspectral data and satellite-based multispectral data for estimating surface soil moisture using machine learning. High-resolution hyperspectral imagery was collected using UAVs, while multispectral data was obtained from Sentinel-2 and Landsat-8/9 satellites. Several machine learning models were trained and evaluated on both datasets. The study demonstrates that UAV-based hyperspectral sensing, combined with machine learning, can provide higher-resolution and more accurate soil moisture maps, which are valuable for precision agriculture, hydrology, and resource management.

• Soil Moisture
• Hyperspectral
• Multispectral
• Machine Learning
• Random Forest

Potential of Visible UAV Imagery in Soil Moisture Estimation

This project investigates the potential of high-resolution visible UAV imagery to estimate surface soil moisture. By capturing aerial images immediately after irrigation and rainfall events, the project extracts spectral indices that are sensitive to soil moisture variations. These indices are then analyzed to assess the relationship between image-derived data and in-situ soil moisture measurements. The study demonstrates that UAV-based visible imagery can provide a low-cost and rapid method for monitoring soil moisture, offering valuable insights for precision agriculture, irrigation management, and sustainable land use planning.

• Soil Moisture
• Visible Imagery
• UAV
• Machine Learning

WaTEM/SEDEM Model Assessment using GLUE

This project evaluates the performance of the WaTEM/SEDEM soil erosion model using the Generalized Likelihood Uncertainty Estimation (GLUE) framework. By applying the model to agricultural watersheds optimized for soil conservation, the study assesses its ability to simulate soil erosion, transport, and deposition processes. The GLUE approach allows for a comprehensive uncertainty analysis, providing insights into model reliability and identifying key parameters influencing model outputs. The findings contribute to improving soil erosion modeling practices and inform land management strategies aimed at reducing soil loss and enhancing conservation efforts.

• Soil Erosion
• WaTEM/SEDEM
• GLUE
• Uncertainty Analysis

Land Suitability Assessment for Wheat Cultivation

This project assesses land suitability for wheat cultivation using an ensemble approach that combines advanced multi-criteria decision models and machine learning techniques. By integrating various environmental, climatic, and soil factors, the study evaluates the potential of different land areas for optimal wheat growth. The ensemble method enhances the accuracy and reliability of suitability assessments, providing valuable insights for agricultural planning, resource management, and sustainable land use practices. The results support decision-making processes aimed at improving crop yields and ensuring food security.

• Land Suitability
• Wheat Cultivation
• Multi-Criteria Decision Models
• Machine Learning
• Ensemble Approach

Forest Fire Susceptibility Mapping

This project develops a bi-layer predictive ensemble (BIPE) framework for mapping forest fire susceptibility across Germany. By integrating multiple machine learning algorithms and environmental factors, the BIPE approach enhances the accuracy of fire risk predictions. The study identifies high-risk areas, providing valuable information for forest management, fire prevention strategies, and emergency response planning. The results contribute to improving our understanding of forest fire dynamics and support efforts to mitigate the impacts of wildfires on ecosystems and communities.

• Forest Fire
• Susceptibility Mapping
• Machine Learning

Precipitation Product Accuracy Evaluation using Google Earth Engine

This project evaluates the accuracy of various satellite-based precipitation products over Utah using the Google Earth Engine platform. By comparing satellite data with ground-based observations, the study assesses the performance of different precipitation datasets in capturing spatial and temporal variability. The research provides insights into the strengths and limitations of each product, informing their appropriate use in hydrological modeling, climate studies, and water resource management. The findings contribute to improving the reliability of precipitation data for various applications.

• Precipitation Products
• Google Earth Engine
• Accuracy Evaluation
• Satellite Data

Fresh water springs identification

In this project, we aimed to identify submarine freshwater springs using remote sensing data. By analyzing multiple derived layers along with field measurements, we successfully located freshwater springs beneath the Persian Gulf. This approach not only enhances our understanding of underwater hydrogeology but also provides valuable information for coastal water resource management, environmental monitoring, and sustainable exploitation of submarine freshwater sources.

• Freshwater Springs
• Remote Sensing
• Hydrogeology
• Environmental Monitoring