I am currently a researcher at the Norwegian Institute for Nature Research NINA. As a researcher specialized in remote sensing, land system science, and landscape ecology, my motivation centers on using technological advancements to tackle ecological challenges. With ecosystems playing a crucial role in climate regulation and biodiversity support, my work aims to bridge the gap in their effective monitoring, management, and restoration. This is particularly urgent in the face of rapid urbanization and environmental change, where my focus is on integrating remote sensing with ecological knowledge to better understand and safeguard terrestrial and marine ecosystems. My goal is to contribute to innovative, practical solutions for sustainable environmental management and policy-making. I am a strong advocate of open science and reproducible research.
Faculty of Biology and Earth Science, Jagiellonian University (Poland)Thesis: Spatiotemporal simulation of land cover change dynamics (Institute of Geography and Spatial Management at the Jagiellonian University Award for outstanding PhD thesis within Geography defended between 2005 and 2007)
2004: MSc in Physics (specialization: Nuclear Physics)
Institute of Physics, Faculty of Physics, Astronomy and Applied Computer Science, Jagiellonian University (Poland) Thesis: Cross-sections of nucleus-proton fragmentation reactions
2002: MSc in Physical Geography within Environmental and Mathematical Studies (specialization: Geographic Information Systems)
Institute of Geography and Spatial Management, Faculty of Biology and Earth Science, Jagiellonian University (Poland) Thesis: Spatial distribution of forests in the Western Beskidy Mountains (Polish Geographical Society Award for outstanding MSc thesis)
2001-2002: Pedagogical training
Teacher Training Center, Jagiellonian University (Poland)
ARCVEG project: Arctic tundra vegetation as a mirror for landscape response to climate change (2022-2024)
The rapidly changing landscapes of polar regions, such as Svalbard, present a fascinating yet critical focus for scientific research. Our project takes a deep dive into this icy realm, employing advanced remote sensing technologies to unravel the dynamics of vegetation productivity. By analyzing a suite of satellite imagery—ranging from MODIS and OCO-2 to Sentinel 5P and Sentinel 2—we're piecing together a detailed picture of how vegetation cover in these extreme environments is shifting. This analysis is further enriched by ground-level data, such as FLoX measurements, offering a unique perspective that bridges the gap between space-based observations and on-the-ground reality. The implications of our findings extend far beyond the icy borders of Svalbard. By understanding these shifts in vegetation cover, we're gaining crucial insights into the broader impacts on global carbon cycles and biodiversity. This research not only sheds light on the immediate effects of environmental change in polar regions but also contributes to our understanding of global ecological balances and their susceptibility to climate change.
The primary objectives of this project are: (1) to leverage recent advancements in remote sensing for comprehensive mapping of land cover and land use changes, utilizing both optical and radar imagery to capitalize on data-rich time series across Central Europe (Czechia, Hungary, Poland, and Slovakia) spanning the past fifty years; (2) to gain a deeper understanding of the driving factors behind these land cover and land use changes in Central Europe; and (3) to evaluate the impact of these changes on biodiversity, as well as carbon pools and fluxes, throughout the region.
TRACE project
Trajectories, causes and effects of land cover and land use changes in Central Europe
Annual MODIS NDVI with inter-annual compositing (from 2020-01-01 to 2021-01-01)
Background
Land provides essential resources to society, including food, fuel, fibres and many other ecosystem services that support production functions, regulate risks of natural hazards, or provide cultural and spiritual services. By using the land, society modifies the provision of these services. Changes in land systems are the main factors of global environmental change and, at the same time, have significant consequences for the local environment and human well-being. Understanding and modelling of complex land systems and land change trajectories, the enhanced human pressures on the earth's limited land resources, as well as the increasingly complex drivers of those changes, have been critical objectives for land change science or land system science (LSS).
Central Europe has experienced drastic changes in political, economic, and societal structures since 1990. The shift from centralised command economies to market-oriented systems has altered economic opportunities, induced technological changes and fostered rapid demographic processes. In 2004, economic and political conditions changed significantly for some Central and Eastern European countries with their accession to the European Union. Socio-economic and political boundary conditions constitute the framework for land use decisions, and the system change in Central and Eastern Europe had a substantial impact on land management
Land cover change in Central and Eastern Europe are only summarised in a few studies. Therefore, we focus our project on methodologies that enable us to detect and estimate land cover and land use changes, causes and consequences.
Project aims and objectives
The major aims and objectives of the TRACE project include:
Mapping and Analysis of Land Cover and Land Use Change Trajectories (Task 1): Employing remote sensing, the project aimed to map land cover and land use changes across Central Europe (Czechia, Hungary, Poland, Slovakia) over the past fifty years, focusing on processes such as agricultural area abandonment and the wildland-urban interface. This involved utilizing optical and radar imagery, along with data-dense time series, to create a detailed spatial database of these
changes.
Understanding the Causes of Changes (Task 2): Through ensemble modeling techniques and the telecoupling framework, the project seeked to explore the socio-economic, political, and institutional drivers behind these land changes. We achived the goal of understand the why behind land cover and land use modifications, incorporating various datasets to support the analysis in regional scale.
Assessing the Effects on Biodiversity and Carbon Cycles (Task 3 and 4): Another objective was to evaluate how land cover and land use changes have affected biodiversity and carbon pools and fluxes across the region. This involved using remote sensing data and methodologies to analyze the impacts of these changes, providing insights into the consequences for ecosystem services and carbon dynamics. We focused on developmented on indecieced that allowed to improved assessment of biodiversity changes in regional scale - Central Europe but also added additional aspect to our analysis and focus on roadless areas.
The TRACE project directly aligns with several United Nations Sustainable Development Goals (UN SDGs), particularly:
SDG 13 (Climate Action): By examining the impact of land use changes on carbon cycles and contributing to knowledge on climate change mitigation through land management.
SDG 15 (Life on Land): The project's focus on biodiversity and ecosystem services relates directly to efforts to protect, restore, and promote sustainable use of terrestrial ecosystems.
SDG 11 (Sustainable Cities and Communities): Through its investigation of the wildland-urban interface, the project contributes to understanding and managing urban expansion in a sustainable manner.
By aiming to provide a comprehensive assessment of land cover and land use changes, their drivers, and their environmental impacts, TRACE supports the broader goals of sustainable development by informing policy and management strategies to balance economic development with environmental conservation and resilience to climate change.
Publications
Szczęch M; Kania M.; Loch J.; Ostapowicz K.; Struś P., 2024, Mapping grasslands' preservation potential: A case study from the northern Carpathians, Land Degradation & Development, 35, 2, https://doi.org/10.1002/ldr.4941
Hoffmann M.T., Ostapowicz K., Bartoń K., Ibisch P.L., Selva N., 2024, Mapping roadless areas in regions with contrasting human footprint, Scientific Reports, 14, 4722, https://doi.org/10.1038/s41598-024-55283-3
More coming soon by the end of 2024
Team
Katarzyna Ostapowicz (Principal Investigator)
Mateusz Szczęch (Postdoctoral researcher)
Konrad Turlej (Postdoctoral researcher)
Monika Hoffmann (Doctoral researcher) Aleksandra Wasik (Master student)
SOCPIX project: Socializing the pixel - detecting and understanding of changing land systems with remote sensing and social sciences (grant from the Polish National Agency for Academic Exchange, 2019-2020, no. PPN/BEK/2018/1/00310/00001) [role: Principal Investigator]
This project aimed to develop concepts and methodologies to (1) integrate remote sensing data required in telecoupling frameworks with socioeconomic information at the pixel level (PIXEL SOCIALIZATION); (2) accurately map and link land use intensification and expansion, displacement, and transition using remote sensing and spatial data; (3) understand how land use changes are driven by complex factors that transcend spatial, institutional, and temporal scales; and (4) examine how various stakeholders organize land use dynamics, influencing regime shifts in land systems and the emergence of frontiers.
RS4FOR project: Forest change detection and monitoring using passive and active remote sensing data (grant from the National Science Center - Poland, 2016-2020, no.2015/19/B/ST10/02127) [role: Principal Investigator]
The main aim of the RS4FOR project was to develop and test approaches that allow improving forest cover change detection and monitoring using different types of remote sensing data (optical data: Landsat 4, 5, 7, 8 (data time series from 1985 to 2017) and Sentinel 2 (data time series from 2015-17), radar data: Sentinel 1 (data time series from 2014-17), and data from airborne laser scanning (ALS) (2013, project ISOK). We focused on both forest cover conversion and modification and three different aspects of forest monitoring: (1) forest cover and its change, (2) prediction models of forest structure and its change and (3) forest health. Our approaches was developed for the temperate forest in mountainous areas.
Use of UAV data for agriculture land abandonment and forest secondary succession mapping (grant from the Faculty of Geography and Geology at the Jagiellonian University - fund for linking science with practice, 2019, No.) [role: Principal Investigator]
In this project, we developed a workflow enabling the automatic detection of land abandonment in agricultural areas and secondary forest succession, utilizing UAV and machine learning techniques.
CON@SK.PL project: Transboundary ecological connectivity – modelling landscapes and ecological flows (grant from the Visegrad Funds, 2017-2018, No. 21640051) [role: Co-Principal Investigator]
The CON@SK.PL project was dedicated to gaining a deeper understanding of habitat connectivity in the Northern Carpathians of Slovakia and Poland. Our international team focused on assessing multispecies connectivity, specifically for the brown bear (Ursus arctos) and European bison (Bison bonasus L.). Utilizing state-of-the-art approaches, we were able to make substantial contributions to addressing critical conservation issues concerning these species.
LIM project: Integration of categorical- and gradient-based approaches in landscape fragmentation and connectivity modelling using GIS&T (grant from the National Science Center - Poland, 2012-2015, no. 2011/03/D/ST10/05568) [role: Principal Investigator]
In the LIM project, we focused on developing new models and measures that enabled the quantitative description of fragmentation and connectivity. Our work involved forests and protected species, including the brown bear and European bison. Throughout the analysis, we employed geographic information technology and spatial data, such as elevation models and satellite land cover maps, alongside various analytical modeling methods. We demonstrated that our workflows could be successfully applied in practical settings. For example, they were instrumental in guiding environmental conservation plans, which involved identifying biodiversity hotspots, preventing biodiversity loss, determining reintroduction sites for endangered species, and delimiting ecological corridors.
FORECOM project: Forest cover changes in mountainous regions - drivers, trajectories and implications (grant from the Swiss Contribution , 2012-2016, No. PSPB 008/2010) [role: Senior researcher]
200 years of land use and land cover changes and their driving forces in the Carpathian basin in Central Europe (grant from the NASA LCLUC program, 2011-2014, No. NNH09DA001N) [role: Senior researcher]
Mountain.TRIP project: Mountain Sustainability: Transforming Research into Practice (grant from the EU: SEVENTH FRAMEWORK PROGRAMME Environment (including Climate Change), 2009-2011, No. FP7-ENV-2009-5.1.0.2) [role: Senior researcher]
Land cover and land use change in the Polish Carpathians (grant from the Polish Ministry of Science and Education, 2008-2010, No. NNH09DA001N) [role: Co-Principal Investigator]
Land use change in the period of political transformation and their relationship with natural and socio-economic conditions in the western part of the Beskidy Mountains (grant from the Polish Ministry of Science and Education, 2004-2007, No. NNH09DA001N) [role: Doctoral researcher]
SOCPIX project
Socializing the pixel - detecting and understanding of changing land systems with remote sensing and social sciences
Background
Land provides essential resources to society, including food, fuel, fibres and many other ecosystem services that support humans, regulate risks of natural hazards, or provide cultural and spiritual services. By using land, society modifies the provision of these services. Changes in land systems are main factors of global environmental change and have large consequences for the local environment and human well-being. Understanding and modelling complex land systems and land change trajectories, the enhanced human pressures on the earth’s limited land resources, as well as the increasingly complex drivers of those changes, have been key objectives for land change science or land system science (LSS).
Project objectives
The main objective of this project was to develop analytical approaches explaining the linkages between the major processes in land systems, i.e., land use intensification and expansion, land use displacement and land use transition within the telecoupling framework with a particular emphasis on the advantage of Earth observations (EO) and spatial data use. This aim will be achieved in a sequence of tasks focusing on different aspects of the emerging land use change of European (the Carpathians), and North American (the Rocky Mountains and Sierra Nevada) mountainous ranges over last five decades and its linkages with other regions.
We developed concepts and methodologies that explained:
how to integrate remote sensing data needed in telecoupling frameworks with socioeconomic information on pixel level (PIXEL SOCIALIZATION),
how to accurately map and link land use intensification and expansion, land use displacement and land use transition using remote sensing and spatial data,
how land use changes are influenced by a complexity of drivers that transcend spatial, institutional and temporal scales,
how land use dynamics are organized by different actors to shape land systems regime shifts and frontier emergence.
Forest change detection and monitoring using passive and active remote sensing data
Project aim & objectives
The main aim of the RS4FOR project was to develop approaches that allowed to improve forest cover change detection and monitoring using various types of Earth Observation:
optical imagery: Landsat MSS, TM, ETM+ and OLI (time series 1978- 2020) and Sentinel-2 (data time series from 2016-2020),
radar imagery: Sentinel-1 (data time series from 2016-2020),
We focused on both forest cover conversion and modification, and on four aspects of forest monitoring:
forest cover and its change,
prediction models of forest structure and its change,
forest health,
old-growth forests.
Our approaches were developed for temperate forests in mountainous areas. The test area was located in the Polish Carpathians.
Publications
Zielonka A., Drewnik M., Musielok Ł., Dyderski M. K., Struzik D., Smułek G., Ostapowicz K., 2021, Biotic and Abiotic Determinants of Soil Organic Matter Stock and Fine Root Biomass in Mountain Area Temperate Forests—Examples from Cambisols under European Beech, Norway Spruce, and Silver Fir (Carpathians, Central Europe), Forests, 12, 823
Grabska E., Frantz D., Ostapowicz K., 2020, Evaluation of machine learning algorithms for forest stand species mapping using Sentinel-2 imagery and environmental data in the Polish Carpathians, Remote Sensing of Environment, 251, 112103
Grabska E., Hostert P., Pflugmacher D., Ostapowicz K., 2019, Forest Stand Species Mapping Using the Sentinel-2 Time Series, Remote Sensing, 11, 10, 1197
More coming soon by the end of 2024
Team
Katarzyna Ostapowicz (Principal Investigator)
Katarzyna Staszyńska (Postdoctoral researcher)
Ewa Grabska (Doctoral researcher)
Anna Zielonka (Doctoral researcher)
Collaboration
Humboldt-Universitat zu Berlin, Geography Department, Earth Observation Lab., Germany
Patrick Hostert
Dirk Pflugmacher
David Frantz
Integration of categorical- and gradient-based approaches in landscape fragmentation and connectivity modelling using GIS&T
Background
The world around us is changing more and more heavily influenced by human activity. We transform the landscape around us and thereby change the conditions of existence of different species and their habitats. For example, forests fragmentation is shifting which influences connectivity of forest patches and affects animals’ movement. Therefore, for efficient spatial management or development of effective conservation mechanisms, it is essential to have the right tools and workflows that allow for complex evaluation of fragmentation and connectivity at the landscape and species level.
Project aim & objectives
In LIM project, we focused on new models and measures development which would enable quantitatively describe habitats’ fragmentation and connectivity. We worked with two protected species; brown bear and European bison.
Outcomes
At each step of the analysis, we used geographic information technology, among others, spatial data such as elevation models or satellite land cover maps and various analytical modelling methods e.g., network analysis. We have shown that our workflows can be successfully used in practice, e.g., to identify new directions in environment conservation plans that include identification of biodiversity hotspots and prevention of biodiversity loss, the location of reintroduction sites for endangered species or delimitation of ecological corridors.
Publications
Ziółkowska E., Ostapowicz K., Radeloff V.C., Selva N., Kuemmerle T., Śmietana W., 2016, Assessing differences in connectivity based on habitat versus movement models for brown bears in the Carpathians, Landscape Ecology, 31, 1863-1882
Ziółkowska E., Perzanowski K., Bleyhl B., Ostapowicz K., Kuemmerle T., 2016, Understanding unexpected reintroduction outcomes: why do European bison do not colonize suitable habitat in the Carpathians? Biological Conservation, 195, 106-117
Ziółkowska E., Ostapowicz K., Kuemmerle T., Radeloff V., 2014, Effects of different matrix representations and connectivity measures on habitat network assessments, Landscape Ecology, 29, 9, 1551-1570
Team
Katarzyna Ostapowicz (Principal Investigator)
Elżbieta Ziółkowska (Doctoral researcher)
Collaboration
Humboldt-Universitat zu Berlin, Geography Department, Biogeography Lab., Germany
Tobias Kummerle
Benjamin Bleyhl
University of Wisconsin-Madison, Department of Forest and Wildlife Ecology, SILVIS Lab., USA
Volker C. Radeloff
Polish Academy of Science, Poland
Nuria Selva
Kajetan Perzanowski
The northernmost hyperspectral FLoX sensor dataset for monitoring of high-Arctic tundra vegetation phenology and Sun-Induced Fluorescence (SIF)
Information about forest stand species distribution is essential for biodiversity modelling, forest disturbances, fire hazard and drought monitoring, biomass and carbon estimation, detection of non-native and invasive species, as well as for planning forest management strategies. High temporal and spectral resolution remote sensing data from the Sentinel-2 mission enables the derivation of accurate and timely maps of tree species in forests in a cost-efficient way. However, there is still a lack of studies regarding forest stand species mapping for large areas like the Polish Carpathian Mountains (approx. 20,000 km2). In this study, we aimed to develop a workflow to obtain forest stand species maps with machine learning algorithms applied to multi-temporal Sentinel-2 products and environmental data at regional scale. Using variable importance techniques - Variable Importance Using Random Forests (VSURF) and Recursive Feature Elimination (RFE) - we assessed three Sentinel-2 Best Available Pixel composites (April, July and October), eight annual spectral-temporal metrics (STM; mean, minimum, maximum, standard deviation, range, first quartile, third quartile and interquartile range), and four environmental topographic variables (elevation, slope, aspect, distance to water bodies), i.e. 114 variables in total. Following a variable importance assessment, we produced maps of eleven tree species with the use of three Machine Learning algorithms: Random Forest (RF), Support Vector Machines (SVM) and Extreme Gradient Boosting (XGB) on nine different variable subsets, i.e. in total 27 classifications. The results showed that SVM outperformed the other two classifiers - the highest overall accuracy exceeded 85% for SVM classification of all variables (86.9%), and 64 variables (85.6%). Including elevation information improved the accuracies. From the best five classifications we created a final ensemble map (overall accuracy of 86.6%) and a precision map based on the Simpson Index, which indicates where the five models agree. This ensemble approach allowed us to determine that the lowest precision occurred in foothills and basins with lower forest cover, in the areas with lack of good quality imagery, and at the borders of stands with homogenous species composition. On the other hand, the highest precision occurred in regions with homogenous stands with high forest and canopy cover. The study demonstrates the potential of Sentinel-2 imagery and topographic data in mapping forest stand species in large mountainous areas with high accuracy. Furthermore, it demonstrates the usefulness of the ensemble approach, which enables to assess the classification precision.
Tømmervik H., Julitta T., Nilsen L.; Park T., Burkart A., Ostapowicz K., Karlsen S.R., Parmentier F., Pirk N., Bjerke J.W., 2023, Data in Brief, 50, 109581
Evaluation of machine learning algorithms for forest stand species mapping using Sentinel-2 imagery and environmental data in the Polish Carpathians
Information about forest stand species distribution is essential for biodiversity modelling, forest disturbances, fire hazard and drought monitoring, biomass and carbon estimation, detection of non-native and invasive species, as well as for planning forest management strategies. High temporal and spectral resolution remote sensing data from the Sentinel-2 mission enables the derivation of accurate and timely maps of tree species in forests in a cost-efficient way. However, there is still a lack of studies regarding forest stand species mapping for large areas like the Polish Carpathian Mountains (approx. 20,000 km2). In this study, we aimed to develop a workflow to obtain forest stand species maps with machine learning algorithms applied to multi-temporal Sentinel-2 products and environmental data at regional scale. Using variable importance techniques - Variable Importance Using Random Forests (VSURF) and Recursive Feature Elimination (RFE) - we assessed three Sentinel-2 Best Available Pixel composites (April, July and October), eight annual spectral-temporal metrics (STM; mean, minimum, maximum, standard deviation, range, first quartile, third quartile and interquartile range), and four environmental topographic variables (elevation, slope, aspect, distance to water bodies), i.e. 114 variables in total. Following a variable importance assessment, we produced maps of eleven tree species with the use of three Machine Learning algorithms: Random Forest (RF), Support Vector Machines (SVM) and Extreme Gradient Boosting (XGB) on nine different variable subsets, i.e. in total 27 classifications. The results showed that SVM outperformed the other two classifiers - the highest overall accuracy exceeded 85% for SVM classification of all variables (86.9%), and 64 variables (85.6%). Including elevation information improved the accuracies. From the best five classifications we created a final ensemble map (overall accuracy of 86.6%) and a precision map based on the Simpson Index, which indicates where the five models agree. This ensemble approach allowed us to determine that the lowest precision occurred in foothills and basins with lower forest cover, in the areas with lack of good quality imagery, and at the borders of stands with homogenous species composition. On the other hand, the highest precision occurred in regions with homogenous stands with high forest and canopy cover. The study demonstrates the potential of Sentinel-2 imagery and topographic data in mapping forest stand species in large mountainous areas with high accuracy. Furthermore, it demonstrates the usefulness of the ensemble approach, which enables to assess the classification precision.
Grabska E., Frantz D., Ostapowicz K., 2020, Remote Sensing of Environment, 251, 112103
Assessing differences in connectivity based on habitat versus movement models for brown bears in the Carpathians
Connectivity assessments typically rely on resistance surfaces derived from habitat models, assuming that higher-quality habitat facilitates movement. This assumption remains largely untested though, and it is unlikely that the same environmental factors determine both animal movements and habitat selection, potentially biasing connectivity assessments. We evaluated how much connectivity assessments differ when based on resistance surfaces from habitat versus movement models. In addition, we tested how sensitive connectivity assessments are with respect to the parameterization of the movement models. We parameterized maximum entropy models to predict habitat suitability, and step selection functions to derive movement models for brown bear (Ursus arctos) in the northeastern Carpathians. We compared spatial patterns and distributions of resistance values derived from those models, and locations and characteristics of potential movement corridors. Brown bears preferred areas with high forest cover, close to forest edges, high topographic complexity, and with low human pressure in both habitat and movement models. However, resistance surfaces derived from the habitat models based on predictors measured at broad and medium scales tended to underestimate connectivity, as they predicted substantially higher resistance values for most of the study area, including corridors. Our findings highlighted that connectivity assessments should be based on movement information if available, rather than generic habitat models. However, the parameterization of movement models is important, because the type of movement events considered, and the sampling method of environmental covariates can greatly affect connectivity assessments, and hence the predicted corridors.
Ziółkowska E., Ostapowicz K., Radeloff V.C., Selva N., Kuemmerle T., Śmietana W., 2016, Landscape Ecology, 31, 1863-1882
Postal and office address:Norwegian Institute for Nature Research NINA, FRAM – High North Research Centre for Climate and the Environment, PO Box 6606 Langnes, NO-9296 Tromsø, Norway Phone: +47 46505709
Email: katarzyna.ostapowicz@nina.no
Annual MODIS NDVI with inter-annual compositing (from 2020-01-01 to 2021-01-01)
