Mahsa Mozaffari

Maîtrise
Université du Québec en Abitibi-Témiscamingue

Amélioration de la cartographie de la texture du sol à l'aide de la télédétection haute résolution et de l'apprentissage automatique
"Improving Soil Texture Mapping Using High-Resolution Remote Sensing and Machine Learning"

Directeur: Maxence Martin
Codirecteur: Osvaldo Valeria

LinkedIn 

CURRENT RESEARCH THEME

Soils play a crucial role in agriculture, forestry, and environmental planning, providing essential ecosystem services. However, existing soil maps often lack the spatial resolution needed for local applications, limiting their effectiveness in supporting informed decision-making. Digital Soil Mapping (DSM) uses remote sensing, climate, and topographic data to model soil properties. Large-scale DSM products provide broad spatial coverage but typically by using low to moderate-resolution data, which limits their effectiveness for local applications. In contrast, local-scale models based on high resolution data capture fine-scale variability but may lack generalizability. A hierarchical approach integrating large-scale predictions with high-resolution data in a local-scale model offers a promising solution to the limitations of the previous two model types. This study aims to evaluate how high-resolution remote sensing data contribute to soil texture predictions and to assess whether integrating SIIGSOL predictions improves model accuracy. To achieve this, we integrate Sentinel-1 (radar), Sentinel-2 (optical), and LiDAR-derived terrain indices with soil texture predictions from SIIGSOL, a 100m resolution digital soil mapping system for Quebec, Canada. The analysis focuses on a single depth level in three study areas in Quebec. The study areas were chosen for their diverse soil and geological characteristics, providing the necessary range of conditions to evaluate the methodology effectively. A Random Forest model is used to predict soil texture, first based solely on high-resolution remote sensing data, and then incorporating the SIIGSOL map to assess the benefits of integrating large-scale soil predictions within a local modeling framework. Preliminary results indicate that integrating high-resolution remote sensing data with SIIGSOL predictions enhances the accuracy of soil texture predictions. These findings suggest that multi-scale data integration will enhance land management measures, contributing to sustainable development and environmental planning in Quebec.

ACADEMICS

• M. Sc. Ecology and Management of Forest Ecosystems, University of Quebec in Abitibi-Témiscamingue (UQAT), Rouyn-Noranda, QC, Canada [2023 - to date]
• M. Sc. Remote Sensing Engineering, K. N. Toosi University of Technology, Tehran, Iran [2021]
• B. Sc. Geomatics Engineering, University of Zanjan, Zanjan, Iran [2018]

RESEARCH EXPERIENCES

• Natural hazards assessment and time series analysis using InSAR techniques
• Remote sensing applications in vegetation monitoring, land-use/land-cover mapping, and soil moisture estimation
• Urban environmental studies including urban heat island monitoring and atmospheric analysis

TECHNICAL SKILLS

• Programming: Python, R, MATLAB
• Remote Sensing and GIS: GEE, ENVI, ArcGIS, QGIS, eCognition, SNAP, SARproz, StaMPS, GMTSAR, ISCE, PolSARpro
• Operating Systems: Windows, Linux

POSTER PRESENTATIONS

• Investigating the uncertainties of digital soil texture map using high-resolution remote sensing data. Mahsa Mozaffari, Osvaldo Valeria, Jean-Daniel Sylvain, Maxence Martin. 17th annual CEF conference, Université du Québec en Outaouais [May 2024].
• Improve high-resolution regional mapping of soil properties and their uncertainty by using artificial intelligence approach . Mahsa Mozaffari, Osvaldo Valeria, Mickaël Germain, Jean-Daniel Sylvain, Maxence Martin. 25th AFD Chair conference, University of Quebec in Abitibi-Témiscamingue, Rouyn-Noranda, Quebec (Nov 2023).

MEMBERSHIP

• Institut de Recherche sur les Forêts (IRF, UQAT)
• Centre d'étude de la forêt (CEF)
• Chaire UQAT-UQAM en aménagement forestier durable (CAFD)

LANGUAGES

• Persian [Mother tongue]
• English [Advanced]
• French [Intermediate]