Balázs Székely

Wildfire risk under climate change and urban expansion emphasizes the need for integrated risk mapping. This study develops a multi-factor Fire Susceptibility Index (FSI) based on vegetation, terrain and anthropogenic factors applied to the Pest County and Budapest region of Hungary. It combines satellite-derived vegetation, moisture, and thermal indices; topographic variables; and proximity to roads and settlements. We then compare FSI patterns between the extreme 2022 drought and a milder 2024, applying Principal Component Analysis (PCA) to identify dominant risk gradients. The model was quantitatively validated against historical fire occurrence data (2018–2024) from the NASA FIRMS dataset through Frequency Ratio (FR) analysis. The results of the PCA revealed that the first component (accounting for approximately 28.6% of the total variance) corresponded to a terrain-fuel/dryness gradient, while the second component (12%) reflected an elevation/access gradient. The validation results demonstrated that over 90% of historical ignitions occurred within FSI Class 2, yielding FR values > 1.2 across both study years. This finding serves to substantiate the model’s capacity to discern the “ignition shelf” within transitional peri-urban zones. While the 2022 drought produced widespread high-susceptibility zones in steep, forested uplands, the actual ignition frequency was highest in flatter, accessible areas, highlighting a disparity between landscape hazard and ignition probability. The mean FSI and the proportion of high-risk areas decreased from 2022 to 2024, therefore, the integrated FSI, PCA, and FR framework provides a robust, validated approach for wildfire risk assessment, which is effective in distinguishing between high-hazard fuel-terrain complexes and high-frequency ignition zones.

Long Short-Term Memory (LSTM) neural networks have proven their excellence in basin-level discharge prediction, provided there is an adequate amount of high-quality time series data available for training, including meteorological forcings and streamflow gauge measurements. Such data-driven black-box models can successfully learn the complex behavior of delayed hydraulic responses; however, they cannot yet be easily applied in water management practice, and model transfer attempts to ungauged catchments have not been entirely successful.In our previous work, we explored an approach to characterizing near-surface flow regimes, starting from a full catchment model and then applying a single LSTM network layer within a semi-distributed subbasin setup reflecting downstream topography. Application to the Tarna River catchment area in Hungary (2,116 km2) showed that transfer learning from the full catchment model (achieving an NSE of 0.91 on the training set and 0.66 on an independent test set) to a downstream chain of gauged Hydrological Response Units (HRUs) is a powerful tool for investigating a semi-distributed HRU network. The entire setup, however, involves a much higher level of complexity, and the available detailed meteorological data and gauge measurements in only two-thirds of the subbasins did not provide sufficient information for the single LSTM model to fully predict the HRU network processes.Because these models apply “virtual water amounts” stored in the hidden cells of the LSTM network for discharge estimation, their internal variables lack direct physical interpretability. In the present research, we investigate how data fusion during calibration with Gravity Recovery and Climate Experiment (GRACE) data, downscaled using soil water content and evapotranspiration products from the ECMWF Reanalysis (ERA5) database, can improve predictive performance, and help to verify our working hypothesis regarding the theoretical connection between Near Surface Water Content (NSWC) and LSTM cell states. These results can also validate interpretations derived from our model concerning baseflow contributions and recharge-discharge classification of subbasins, while promising realistic transferability of the pre-trained lumped catchment model to all subbasins and broader general applicability of the proposed method. We hypothesize that the daily change dynamics of Terrestrial Water Storage (TWS) and NSWC – the latter playing a decisive role in gravitational flows within the Critical Zone – are strongly correlated.Accordingly, we propose using downscaled TSW estimates to (1) introduce a new term into the loss function based on our working hypothesis relating median LSTM cell state values to the normalized dynamics of NSWC, and (2) add a new input dimension approximating total runoff as precipitation minus evapotranspiration and infiltration.Furthermore, the current model extension, still based on 0.1 ° gridded input data, prepares the ground for future developments that incorporate high-spatial-resolution satellite remote sensing data, such as Sentinel-2 NDWI, to support local-scale hydrological applications efficiently. Integrating satellite data products with different temporal and spatial resolutions is not a straightforward calibration step for rainfall-runoff models, as pixel-wise normalization of measurements requires complex physically based geostatistical methods compatible with model logic to avoid performance deterioration.

Rapid urbanization in Nairobi (Kenya) is rapidly degrading its urban forests, threatening the city’s long-held identity as the “green city in the sun.” The loss of tree cover in Karura, Ngong Road Forest, and Oloolua reflects a broader problem of deforestation driven by development pressure and weak policy enforcement. This issue is particularly specific because Nairobi’s unique combination of rapid urban growth, legal gaps, and intense competition for land makes forest decline occur more severely here than in many other African cities. It also represents a distinct case where national legislation such as the Forest Act (2005) and EMCA (2015) exist but their implementation within an urban setting remains inconsistent. The data for this research included multi-temporal satellite imagery accessed through Google Earth Engine, focusing on Landsat and Sentinel datasets spanning 2000 to 2024. These datasets were used to generate vegetation indices such as NDVI to quantify forest cover change across the three major urban forests. Complementary policy documents, county urban planning records, and environmental legislation were analyzed to contextualize the observed changes. The methodology combined remote sensing analysis and GIS mapping using platforms such as Google Earth Engine and QGIS. NDVI computation, supervised classification, and change detection techniques were applied to assess temporal and spatial forest cover decline. This geospatial work will be integrated with qualitative policy evaluation to identify the governance gaps driving the ecological trends. The results showed a clear downward trend in forest cover over the last five decades, with sharper losses occurring during periods of accelerated urban expansion. We anticipated demonstrating misalignment between policy intentions and actual land-use outcomes, particularly were development overrides environmental protections. These results will likely reveal the need for stronger urban forest governance. This research is important because it offers a science-based understanding of how policy failures directly shape ecological degradation in growing cities. It contributes to the broader field of global environmental change by linking remote sensing evidence with governance analysis. Ultimately, the study provides an outlook for improving urban sustainability, guiding policymakers and planners in protecting Nairobi’s remaining forests while addressing future urban growth pressures.

The Mohács Plain, a gently terraced low-relief alluvial plain on the right bank of the Danube in southern Hungary, was the site of one of the most important battles in Hungarian history in 1526. There are only a few written accounts of the battle in which the Hungarian forces suffered a defeat against the Ottomans, therefore the exact location of the battle is still debated.In addition to the numerous studies tackling this problem we introduce another aspect: the geomorphometric analysis that contributes to our understanding. As it is a flat area, the focus is primarily on examining microtopographic features and their relative age ordering, in order to construct a feasible 16th century drainage model of the area. The aim of our approach is to determine past and present channels of streams and watercourses that shaped the terrain. These features were analysed using a variety of software, including QGIS and our own sophisticated image processing tools.The created runoff model helped us to identify dried-up streambeds and other avulsions. The elevated terrace encompasses incised channels that previously acted as alluvial ones. This observation may support the idea of effects of a more humid climate resulting in large discharge volumes relative to recent conditions. Image segmentation has yielded results that can be used to distinguish patterns in the terrain model including smaller subunits and patterns within them.By applying these methods, we aim to create a reconstruction of the 16th century landscape. By comparing this reconstruction with historical descriptions and archaeological observations, we expect to assist historians and archaeologists in pinpointing the exact location of the actions of Battle of Mohács.

In this study, we investigate the spatial and density distribution of zodiacal cloud dust in the inner solar system, with the focus being on Mercury's orbit and the formation and persistence of the circumsolar dust ring discovered in 2018 in the immediate vicinity of the Sun. Previous knowledge of Mercury's orbital environment suggests that, due to its proximity to the Sun and hence its extremely high perturbation effects, this environment may not contain matter in the longer term. This picture seems to have become more uncertain. In our model, we have considered the constrained three-body problem of the Sun-Mercury system along the Lagrangian libration points, which due to the degree of stability and the 1:1 mean motion resonance and the continuous perturbative effects of the perturbed dust particle motion within the stable region we assumed a horseshoe orbit. According to our model, the horseshoe orbit undergoes a strong deformation due to the proximity of the Sun at the near-solar boundary of the stable region, where the Poynting-Robertson effect causes dust particles of different grain sizes to continuously fall out of the stable region and spiral into the Sun. To represent the fate of the different grain sizes, a circumsolar dust ring was set up in three scenarios, in all three of which a smaller compact ring of larger grains appeared.Using comet 67P/Churyumov-Gerasimenko as a mass reference, the number of comets needed to fill the toroidal volume of the orbit of Mercury and the time scale required to do so were quantified. As a final result, our model predicts that the instrumentally discovered circumsolar dust ring is distributed along a highly deformed horseshoe track, which can persist in a continuously replenished dynamical system such that the dust grains that constitute it are replaced on a century-to-millennium scale.

The fluvial dynamics of the Danube in the Mohács Island region (Southern Hungary) were analyzed using historical maps spanning the last 500 years. Battle of Mohács, the defeat of the Hungarian armies by the Ottoman Empire took place here in 1526, in environmental conditions that are quite different from today. To reconstruct these conditions, the behaviour of the Danube, the main influencing factor has been studied.Numerous historical maps, like Tabula Hungariae (1528), Marsigli’s (1726), Müller’s (1709), Lipszky’s (1803) and Korabinszky’s (1804) maps together with the three Military Surveys of the Habsburg Empire (1782-1887) have been transformed or georeferenced and integrated in QGIS for analysis. River branches have been digitized and compared to the present-day situation represented in satellite imagery and a digital elevation model.Apart from comparing the channels of different centuries, sinuosity values of the river branches and oxbow lakes were calculated. The results provide insights into the river's meandering behaviour and its environmental evolution. Although meandering is the dominant in this section, historical evidence indicates a period when the Danube exhibited an island-building phase. During this time, unlike today, the eastern branch of the river marked the main channel, shaping the floodplain. Quantitative analyses demonstrated a general declining trend in sinuosity due to the narrowing of bends, river self-regulation, and recently, human-induced interventions such as river regulation. The persistence of island-building period was likely linked to variations in sediment yield and flow patterns. Historical maps suggest that inflowing streams may have had more discharge during this period, contributing to higher sediment deposition at reduced velocities, particularly when the eastern branch served as the main channel. The eventual reversal of the main branch is attributed to flooding events that redefined the fluvial network. To ensure the reliability of the reconstructions, the accuracy of old maps was also evaluated. The oldest maps depicted rather the character of the river branches, but advances in cartography significantly reduced distortions, enabling the reconstruction of former river courses with greater precision. Comparisons with contemporaneous maps and topographical models highlighted minor deviations and resulted in revision of a few former interpretations. The study also identified key areas where flooding or marshland characteristics could have influenced land use and military strategies, underscoring the broader implications of environmental history in shaping the region's past. This work highlights the critical role of historical cartography and environmental analysis in understanding long-term fluvial transformations and their socio-environmental impacts.

In tropical countries like Indonesia, forest fires are common natural or human-induced disasters that occur throughout the Southeast Asian region all year around. It’s also known for its rich biodiversity and extensive peatlands, which are particularly vulnerable due to circumstances and risk factors. Thus, the research aims to shed light on forest fire susceptibility mapping and assessing high-risk zones based on land use, vegetation type, and climatic factors. This study utilized remote sensing through Google Earth Engine (GEE) and Geographic Information Systems (GIS) to evaluate forest fire risk in Sumatra, Indonesia, a region facing challenges such as rapid palm oil plantation development, timber concessions, and active peatland fires. Tropical rainforests, peatlands, and diverse ecosystems characterize Sumatra. The region is often affected by natural forest fires (e.g., El Niño events) and human-induced activities (e.g., illegal land clearing for plantation purposes). Forest fire locations in the study area were identified using historical hotspot data from 2014 to 2022. Fire risk assessments are typically generated using spectral indices to classify the spatial distribution of damage caused by fires; first, forest fire susceptibility through a combined index calculated by adding classified fire risk, forest loss index, normalized slope, aspect, temperature, and relative humidity. This index indicates potential fire risk and demonstrates how to integrate various geospatial datasets, such as vegetation indices, topography, and climate data, within GEE to assess forest fire risk. Then, we evaluated the feasibility of using remote sensing data to identify fire causes by validating forest fire occurrence factors. This study demonstrates how detailed risk assessment provides an effective method of managing forest fires in Riau Province, Sumatra, Indonesia, which can contribute to reducing the frequency and severity of fires, and improve sustainable forest management and governance.

Recurring and co-occurring extreme climate events exacerbate adverse effects on human livelihoods, regional and local economy, and the environment. Previous studies have extensively researched on the frequency, intensity, and duration of single climate extremes. However, recurring and co occurrence compound extremes remain scantly addressed in the East Africa Region. Here, we examine spatial variations of the precipitation and temperature extremes events from 1991 to 2022 (32 years) in East Africa, where agriculture is the main economic mainstay. We used high-resolution (0.25° x 0.25°) precipitation and temperature ERA5-reanalysis data. Three agriculturally relevant precipitation events: consecutive dry days (CDD), consecutive wet days (CWD), annual total precipitation that is wet-days annual amount (RR ≥ 1 mm)(PRCPTOT), and three core temperature metrics: summer days with temperature > 25°C (SU25), extremely hot days with maximum temperature > 35°C (SU35) and diurnal temperature range (DTR) are examined. Our results show that the mean annual CDD ranges between 0 and 240 days in DR Congo, Uganda, Kenya, and the Ethiopian Highlands. The CWD annual averages were the longest, and the maximum was observed in some parts of DR Congo, Ethiopian, and Kenya highlands (365 days). However, minimum CWD events were experienced in the whole of Somalia and arid and semi-arid lands (ASALs) of Kenya, Southern Sudan, and Tanzania. The highest PRCPTOT was experienced in high altitudes and rainforest biomes. Mean annual SU25 were low, predominating in mountainous regions with less than 100 days. Most parts of Kenya show the annual DTR between 10 °C to 12 °C, and few areas with values between 8 °C to 10 °C and between 12 °C and 15 °C. Rwanda and Burundi had values between 8 °C and 10 °C while Tanzania experienced values between 8 °C to 10 °C and between 10 °C and 12 °C. These agriculturally relevant climate extremes threaten people’s livelihood, which is highly dependent on rainfed agriculture. Therefore, contextual-specific adaptation strategies are imperative in minimizing socioeconomic loss and damaging adverse effects in the agriculture and water sectors. Early warning systems should be enforced over East Africa to minimize compounded climate risks.Keywords: Climate Extremes; East Africa region; ERA5; Precipitation; Temperature.

Lake Velence is a shallow soda lake in Hungary whose water budget is mainly driven by precipitation and evaporation. The lake has shown a deteriorating tendency recently, including extremely low lake levels and poor water quality, which indicates its vulnerability against changing climatic conditions. At the same time several water usage conflicts appeared in the catchment area. Until recently, the groundwater component in the lake's water budget and the hydrogeological processes in the catchment area have not been taken into consideration. Recent hydrogeological studies, however, show groundwater discharge into the lake. Thus, further investigating this question is of high importance, hence groundwater could reduce climatic vulnerability.Our ongoing work aims at developing a model-based evaluation technique, utilizing all map-based geophysical information and time series of different satellite data products, having sufficient spatial resolution and providing information about parameters strongly connected to subsurface processes, showing up on the surface. The basic DEM raster layer is imported from Copernicus GLO-30 dataset, having vertical precision