Lake Sevan in Armenia is a unique, large, alpine lake given its surface, volume, and geographic location. The lake suffered from progressing eutrophication and, since 2018, massive cyanobacterial blooms repeatedly occurred. Although the lake is comparatively intensely monitored, the feasibility to reliably detect the algal bloom events appeared to be limited by the established in situ monitoring, mostly because algal bloom dynamics are far more dynamic than the realized monitoring frequency of monthly samplings. This mismatch of monitoring frequency and ecosystem dynamics is a notorious problem in lakes, where plankton dynamics often work at relatively short time scales. Satellite-based monitoring with higher overpass frequency, e.g., by Sentinel-3 OLCI with its daily overcasts, are expected to fill this gap. The goal of our study was therefore the establishment of a fast detection of algal blooms in Lake Sevan that operates at the time scale of days instead of months. We found that algal bloom detection in Lake Sevan failed, however, when it was only based on chlorophyll due to complications with optical water properties and atmospheric corrections. Instead, we obtained good results when true-color RGB images were analyzed or a specifically designed satellite-based HAB indicator was applied. These methods provide reliable and very fast bloom detection at a scale of days. At the same time, our results indicated that there are still considerable limitations for the use of remote sensing when it comes to a fully quantitative assessment of algal dynamics in Lake Sevan. The observations made so far indicate that algal blooms are a regular feature in Lake Sevan and occur almost always when water temperatures surpass approximately 20 °C. Our satellite-based method effectively allowed for bloom detection at short time scales and identified blooms over several years where classical sampling failed to do so, simply because of the unfortunate timing of sampling dates and blooming phases. The extension of classical in situ sampling by satellite-based methods is therefore a step towards a more reliable, faster, and more cost-effective detection of algal blooms in this valuable lake.

Machine learning (ML) was used to assess and predict urban air temperature (T_air) considering the complexity of the terrain features in Yerevan (Armenia). The estimation was performed based on the Partial Least-Squares Regression (PLSR) model with a high number (30) of input variables. The relevant parameters include a newly purposed modification of spectral index IBI-SAVI, which turned out to strongly impact T_air prediction together with land surface temperature (LST). Cross-validation analysis on temperature predictions across a station-centered 1000 m circular area revealed quite a high correlation (R²_Val = 0.77, RMSE_Val = 1.58) between the predicted and measured T_air from the test set. It was concluded the remote sensing is an effective tool to estimate T_air distribution where a dense network of weather stations is not available. However, further developments will include incorporation of additional weather parameters from the weather stations, such as precipitation and wind speed, as well as the use of non-parametric ML techniques.

Very-high-resolution thermal infrared data has a good potential to detect and monitor LST variations in urban areas. In this work, an attempt was made to estimate Land Surface Temperature (LST) from very high resolution (VHR) data using machine learning technics in an area of Botanical garden of Yerevan, Armenia; UAV-derived and in-situ measured high precision LST of various land cover (LC) patterns were then compared for September and October. The main purpose of this study was to explore the capabilities of UAV imagery (multispectral/TIR) in assessing LST of the different LC patterns. For this purpose, a UAV survey was performed, and VHR data were collected using multispectral and TIR thermal camera while in-situ measurements of temperature and surveying of the different LC patterns were performed. A significant correlation was detected between LSTs in situ measured and detected by UAV in 06 September (r=0.758; p-value <0.01) and in 27 October (r=0.686; p-value <0.01). When comparing the LSTs of separate land-cover patterns the best results were received for Bare soils (r=0.591; p-value <0.05) and Concrete (r=0.927; p-value <0.01) when the survey and measurements was done in September. Despite its limitations, it can be stated that UAV thermal survey has a good potential to detect and monitor LST variations in urban areas.

Machine Learning (ML) was used to assess and predict urban air temperature (Tair) considering the complexity of the terrain features in Yerevan (Armenia). The estimation was performed based on the PLSR model with a high number (30) of input variables. The relevant parameters include a newly purposed modification of spectral index IBI-SAVI, which turned out to be strongly impacting on Tair prediction together with land surface temperature (LST). Cross-validation analysis on temperature predictions across a station-centered 1000m circular area revealed quite a high correlation (R2Val = 0.77, RMSEVal = 1.58) between predicted and measured Tair from the test set. It was concluded the remote sensing is an effective tool to estimate Tair distribution where a dense network of weather stations is not available. However, further developments will include incorporation of additional weather parameters from the weather stations such as precipitation and wind speed, and the use of non-parametric ML techniques.

This paper investigates the prediction of urban air temperature (Tair) from satellite-derived land surface temperature (LST) in the complex urban and topographic environment of Yerevan, Armenia. Building on previous work that demonstrated the effectiveness of machine learning (ML) approaches for point-based Tair estimation using Partial Least-Squares Regression (PLSR) with multiple environmental variables, this study shifts the focus to the spatial distribution of Tair. Several prediction methods and input variable combinations are evaluated to generate gridded Tair maps, which are assessed for spatial consistency against expected patterns driven by land cover, elevation, local knowledge, and spot observations. In total, five predicting methods were used—one regression approach (PLSR) and four ML methods: random forest (RF), quantile regression forest (QRF), support vector machine (SVM), multilayer perception (MLP). RF and QRF deliver the best overall results, with RF achieving the highest testing R² (0.74) and lowest RMSE (0.56). Spatial patterns are similar for PLSR, RF and QRF, highlighting cooler northern high-altitude areas and warmer southern urban areas. Overall, the results confirm the reliability of the proposed Tair spatial mapping methods in complex urban environments.

Accurate land-cover mapping of urban botanical gardens is essential for biodiversity assessment, ecosystem monitoring, and sustainable landscape management. However, their high structural complexity creates significant challenges for remote sensing-based classification. This study evaluates the influence of spatial resolution on classification performance by comparing 3 m PlanetScope satellite imagery and centimeter-scale unmanned aerial vehicle (UAV) data within a 28 ha heterogeneous area of the Yerevan Botanical Garden (YBG) in Yerevan, Armenia. Three pixel-based machine learning (ML) algorithms, Random Forest (RF), Support Vector Machine (SVM), and Gradient Boosting (GB), were compared with the spatial-context-based U-Net deep learning (DL) semantic segmentation model. The results revealed a strong relationship between sensor resolution and classifier performance. For satellite imagery, U-Net achieved the highest statistical accuracy, with an Overall Accuracy (OA) of 78.56% and a Kappa coefficient (κ) of 0.726, demonstrating the benefit of contextual feature extraction for reducing pixel-level noise. However, spatial evaluation indicated that RF generated land-cover patterns with better preservation of local landscape structures and class boundaries. In contrast, UAV-based classification showed superior performance for pixel-based ML methods, with GB achieving the highest accuracy (OA = 85.9%, κ = 0.839), followed by SVM (OA = 84.7%, κ = 0.825) and RF (OA = 84.3%, κ = 0.821). Area comparison revealed that medium-resolution satellite imagery overestimated continuous tree canopy coverage (30.17% vs. 16.53% from UAV mapping) and underestimated fragmented vegetation classes due to mixed-pixel effects. The findings highlight that classifier selection should be adapted to sensor characteristics and landscape complexity. For satellite imagery, U-Net achieved the highest overall accuracy while RF preserved landscape structure and class boundaries more effectively; for UAV imagery, GB achieved the best overall performance, confirming that spectral-based pixel classifiers perform strongly at fine spatial resolutions. This is probably due to reduced mixed-pixel effects rather than enriched spatial information, as the latter was not leveraged in the classification process used.