Stream Network Research Articles

Seasonality Drives Carbon Emissions Along a Stream Network

AbstractHeadwater stream networks contribute substantially to the global carbon dioxide terrestrial flux because of high turbulence and coupling with terrestrial environments. Heterogeneity within headwater stream networks, both spatially and temporally, makes measuring and upscaling these emissions challenging because measurements of carbon dioxide in streams are often limited to a few monitoring points. We modified a stream network model to reflect real measurements made under base flow and high flow conditions at Martha Creek in Stabler, WA in the US Pacific Northwest. We found that under high flow conditions, the stream network had much greater total carbon emissions than during low flow conditions (1.22 Mg C day−1 vs. 0.034 Mg C day−1). We attribute this increase to a larger overall stream network area (0.04 vs. 0.01 km2) and discharge (1.9 m3 s−1 vs. 0.005 m3 s−1) in November versus August. Our results demonstrate the need to understand the nonperennial stream reaches when calculating carbon emissions. We compared the stream network emissions with the terrestrial net ecosystem exchange (NEE) estimated by local eddy covariance measurements per watershed area (−5.5 Mg C day−1 in August and −2.2 Mg C day−1 in November). Daily stream emissions in November accounted for a much larger percentage of NEE than in August (54% vs. 0.62%). We concluded that the stream network can emit a large percentage of the forest NEE in the winter months, and annual estimates of stream network emissions must consider the flow regime throughout the year.

Two stream deep neural network based framework to detect abnormal human activities

Two stream deep neural network based framework to detect abnormal human activities

Adaptive Fragment-Based Parallel State Recovery for Stream Processing Systems

Today, large-scale cloud organizations are deploying datacenters and “edge” clusters globally to provide their users with low-latency access to their services. Running stream applications across these geo-distributed sites are emerging as a daily requirement, such as making business decisions from marketing streams, identifying spam campaigns from social network streams, and analyzing existing genomes in different labs and countries to track the sources of a potential epidemic. However, while the progress has been encouraging, the existing efforts have dominantly centered around <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">stateless stream processing</i> , leaving another urgent trend- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">stateful stream processing</i> -much less explored. A driving need is that next-generation stream processing systems need to store and update states during processing, and most importantly, successfully recover large distributed states when faults and failures happen. Existing studies exhibit major limitations including: (1) they mostly inherit MapReduce's “single master/many workers” architecture, where the central master is responsible for all scheduling activities and easily becomes a scalability bottleneck; (2) they offer state recovery mainly through the use of three approaches: replication recovery, checkpointing recovery, and DStream-based lineage recovery, which are either slow, resource-expensive or failing to handle multiple failures; and (3) they are not adaptive to heterogeneous hardware settings in the cloud. In this paper, we present A-FP4S, a novel adaptive fragments-based parallel state recovery mechanism for stream processing systems to manage and recover large distributed states for a massive number of stream applications. The novelty of A-FP4S is that we organize stream operators into a distributed hash table (DHT) based peer-to-peer (P2P) overlay. Then we divide each node's local state into many fragments and periodically store them in each node's multiple neighbors (the leaf set nodes of DHT), ensuring that different sets of available fragments can reconstruct failed states in parallel. By doing that, this failure recovery mechanism is extremely scalable to the size of the lost state, significantly reduces the failure recovery time, and can tolerate multiple node failures. A-FP4S is adaptive to heterogeneous hardware settings (e.g., CPU speed, disk/file-system speed, network bandwidth) by automatic parameter tuning over phases. Compared to Apache Storm, A-FP4S achieves a significant 31.8% to 50.5% reduction in recovery latency. It can scale to many simultaneous failures and successfully recover the state, even more than half of the operators fail or get lost. Large-scale experiments using real-world datasets demonstrate A-FP4S's attractive scalability and adaptivity properties.

Cuckoo Counter: Adaptive Structure of Counters for Accurate Frequency and Top-k Estimation

Frequency estimation and top-k flows identification are fundamental problems in network traffic measurement. Sketch, as a basic probabilistic data structure, has been extensively investigated and used in different management applications. However, few of them is suitable for both estimating frequency and finding top-k flows due to the unbalanced distribution of real-world network streams. By introducing a pre-filtering stage to isolate elephant and mice flows, the recently proposed Augmented Sketch (ASketch) significantly improves accuracy for both tasks. However, it suffers from serious performance degradation because of frequent flow exchanges. In this paper, we propose Cuckoo Counter (CC), an adaptive structure that consists of several buckets organized in a specific way. The size of the entry in each bucket is carefully designed to match the actual distribution of streams. During processing, CC hashes a flow to buckets and uses the idea of cuckoo hashing to relocate the flow if an overflow or collision happens, which contributes to fully utilizing memory. Therefore, the replacement strategy helps CC precisely record elephant flows and cover more mice flows, and also guarantees the throughput. Extensive experimental results show that CC has the highest (Freq.) accuracy, excellent (Heavy hitter / change) accuracy, highest (Top-k) precision, and competitive throughput compared to the state-of-the-art. Specifically, CC improves the throughput and accuracy by around 1 and 2 orders of magnitude respectively compared to the well-known ASketch.

Flood risk assessment of the river Benue catchment in Adamawa State Nigeria

Floods are one of the most devastating natural disasters that occur globally, causing loss of life and property. Adamawa, a state located in north-eastern Nigeria, is prone to flooding due to its location within the floodplains of the River Benue. The research examines the spatial distribution of flood risk, the role of altitude in determining flood risk levels, and the implications for population exposure. The investigation was carried out in the following local government areas, namely Yola-South, Demsa, Numan, Lamurde, Guyuk, Shelleng and Song. The study used a combination of remote sensing and geographic information systems (GIS) techniques to analyze the flood risks extent and exposure within the catchment. SRTM 30 Meters resolution Digital Elevation Model (DEM) data was used to derive catchment boundaries, stream networks and watershed characteristics. 500 years return period flood hazard raster along with a population raster of the settlements within the region of interest were also used. The findings reveal varying flood risk extents, with areas at lower elevations exhibiting higher flood risks. Additionally, the study identifies specific wards within each locality that are situated in different flood risk zones, highlighting the need for localized flood risk assessments. The results demonstrate a correlation between flood risk levels and the percentage of the population exposed to these risks. The study aligns with existing literature on flood risk, altitude, and population vulnerability, emphasizing the importance of considering these factors in flood risk management strategies. The findings contribute to a better understanding of flood risk patterns and can guide policymakers and stakeholders in implementing targeted interventions and developing effective disaster response plans. However, it is essential to contextualize the findings within the specific study areas and complement them with site-specific assessments and further research.

Drivers of Geomorphic Heterogeneity in Unconfined Non‐Perennial River Corridors

AbstractRiver corridors along non‐perennial stream networks provide diverse physical and ecological functions that are thought to be related to the spatial and temporal variability of geomorphic units, also known as geomorphic heterogeneity. While studies on the characteristics and drivers of geomorphic heterogeneity have been developed in perennial streams, similar studies in ephemeral streams are lacking. Given the ubiquity of non‐perennial streams globally, we aim to answer questions regarding the magnitude and drivers of geomorphic heterogeneity in ephemeral river corridors as well as how geomorphic unit assemblages reflect processes related to flood disturbance. Geomorphic units were mapped in 30 unconfined river corridors within six non‐perennial watersheds in Utah and Arizona, USA. Landscape heterogeneity metrics—Shannon's Diversity Index, Shannon's Evenness Index, and patch density—were used to quantify geomorphic heterogeneity within each reach. Additionally, variables that potentially constrain or drive heterogeneity were quantified, including floodplain shape, grain size, large wood abundance, and proxies for flood disturbance. While heterogeneity positively correlated with metrics for morphology and disturbance, statistical models suggest that morphologic context, particularly river corridor width, was a more important predictor for estimating geomorphic heterogeneity. Still, geomorphic units reflected aggradation processes indicative of a range of flood energies, suggesting a strong tie between heterogeneity and disturbance. Results suggest that geomorphic heterogeneity may be resilient to changes in flood disturbance frequency or magnitude, but future studies investigating long‐term temporal heterogeneity are needed.

Power efficient UAV placement and resource allocation for adaptive video streaming in wireless networks

Unmanned aerial vehicle (UAV) enabled wireless communication that can transmit/receive data to/from ground users, especially at cell boundaries or in disaster-stricken areas where the ground-based infrastructure has been damaged, is an emerging area of research. These aerial vehicles require energy for communication and are typically energy constrained. The communication system becomes more energy hungry when it comes to video streaming of high quality. Because for higher video quality streaming, higher bitrate and more bandwidth is required. Higher transmission power and rate will result in increased streaming quality. We concentrate on UAV-aided power efficient adaptive video streaming wireless networks with multiple ground users, taking into account the available energy budget and information causality constraints. The objective of this research work is to maximize adaptive video streaming utility per unit of power in a UAV-enabled wireless communication network by optimizing the position of UAV and allocation of communication resources, such as bandwidth, and transmit power, while competing with system-designed constraints, such as energy budget, transmission rate, and allocated bandwidth. The problem that is addressed is non-convex which is approached by applying successive convex approximation, Dinkelbach method and coordinates decent techniques, and an efficient iterative algorithm is devised for its solution. Extensive simulations show that the suggested optimization strategy outperforms the results of the previously proposed works.

Climate-warming-driven changes in the cryosphere and their impact on groundwater–surface-water interactions in the Heihe River basin

Abstract. The Heihe River basin in northwest China depends heavily on both anthropogenic and natural storage (e.g., surface reservoirs, rivers and groundwater) to support economic and environmental functions. The Qilian Mountain cryosphere in the upper basin is integral to recharging these storage supplies. It is well established that climate warming is driving major shifts in high-elevation water storage through loss of glaciers and permafrost. However, the impacts on groundwater–surface-water interactions and water supply in corresponding lower reaches are less clear. We built an integrated hydrologic model of the middle basin, where most water usage occurs, in order to explore the hydrologic response to the changing cryosphere. We simulate the watershed response to loss of glaciers (glacier scenario), advanced permafrost degradation (permafrost scenario), both of these changes simultaneously (combined scenario) and projected temperature increases in the middle basin (warming scenario) by altering streamflow inputs to the model to represent cryosphere-melting processes, as well as by increasing the temperature of the climate forcing data. Net losses to groundwater storage in the glacier scenario and net gains in the permafrost and combined scenarios show the potential of groundwater exchanges to mediate streamflow shifts. The result of the combined scenario also shows that permafrost degradation has more of an impact on the system than glacial loss. Seasonal differences in groundwater–surface-water partitioning are also evident. The glacier scenario has the highest fraction of groundwater in terms of streamflow in early spring. The permafrost and combined scenarios meanwhile have the highest fraction of streamflow infiltration in late spring and summer. The warming scenario raises the temperature of the combined scenario by 2 ∘C. This results in net groundwater storage loss, a reversal from the combined scenario. Large seasonal changes in evapotranspiration and stream network connectivity relative to the combined scenario show the potential for warming to overpower changes resulting from streamflow. Our results demonstrate the importance of understanding the entire system of groundwater–surface-water exchanges to assess water resources under changing climatic conditions. Ultimately, this analysis can be used to examine the cascading impact of climate change in the cryosphere on the resilience of water resources in arid basins downstream of mountain ranges globally.

An ablation study on part-based face analysis using a Multi-input Convolutional Neural Network and Semantic Segmentation

Face-based recognition methods usually need the image of the whole face to perform, but in some situations, only a fraction of the face is visible, for example wearing sunglasses or recently with the COVID pandemic we had to wear facial masks. In this work, we propose a network architecture made up of four deep learning streams that process each one a different face element, namely: mouth, nose, eyes, and eyebrows, followed by a feature merge layer. Therefore, the face is segmented into the part of interest by means of ROI masks to keep the same input size for the four network streams. The aim is to assess the capacity of different combinations of face elements in recognizing the subject. The experiments were carried out on the Masked Face Recognition Database (M2FRED) which includes videos of 46 participants. The obtained results are 96% of recognition accuracy considering the four face elements; and 92%, 87%, and 63% of accuracy for the best combination of three, two, and one face elements respectively.

Integrating spatially-and temporally-heterogeneous data on river network dynamics using graph theory

The study of non-perennial streams requires extensive experimental data on the temporal evolution of surface flow presence across different nodes of channel networks. However, the consistency and homogeneity of available datasets is threatened by the empirical burden required to map stream network expansions and contractions. Here, we developed a data-driven, graph-theory framework aimed at representing the hierarchical structuring of channel network dynamics (i.e., the order of node activation/deactivation during network expansion/retraction) through a directed acyclic graph. The method enables the estimation of the configuration of the active portion of the network based on a limited number of observed nodes, and can be utilized to combine datasets with different temporal resolutions and spatial coverage. A proof-of-concept application to a seasonally-dry catchment in central Italy demonstrated the ability of the approach to reduce the empirical effort required for monitoring network dynamics and efficiently extrapolate experimental observations in space and time.

A world-class example of a Late Palaeozoic glaciated landscape in Chad

Continental glacial deposits are rarely preserved in nature. Recent studies of the LPIA reveal the presence of periglacial deposits in Northern Chad covering an area of 20,000 km2. Using satellite maps, this project managed to examine the entirety of the Ennedi Plateau. 3D reconstructions of this mountain range reveal complex glacial morphological features. This paper presents large-scale subglacial features interbedded with proglacial fluvial features and discussed the variables at play during the progressive, repetitive, and perhaps cyclical nature of the Carboniferous glaciation. Using remote sensing techniques, this paper shows that there are three major ice advance–retreat cycles that led to the deposition of mega-scale glacial lineations with a complex fluvial channel belt. Palaeo-geomorphological mapping of satellite images in northern Chad reveals that the Late Carboniferous Chadian ice sheet covered at least 20,000 km2, a much greater extent than previously thought (ca. 6000 km2). Glacial lineations (GLs) are dominant and extensive on the plateau. They are arranged in widespread ice stream networks and developed at multiple stratigraphic levels. This broadly indicates palaeo-ice flow to the north. We report a newly discovered pristine channel belt, covering at least 300 km2 of a plateau in the Ennedi Plateau. A swarm of N–S trending channels, up to 250 m wide, is recognised. These are of both braided and meandering characters and are organised into channel belts of up to 500 m width. The mapped palaeo-morphology shows complex crosscutting relationships. The channel belt is intercalated with zones of GLs; hence, ice activity preceded and succeeded fluvial activity. Preserved planform sinuous channel geometries show distinctive point bar deposits and scroll bar geometries therein, which testify to a terrestrial, subaerial environment, rather than to their evolution in the subglacial realm, e.g. as eskers. Therefore, we find that cross-cutting, amalgamated channel systems record the complex phases of meltwater release during glacial retreat in the advance–retreat cycles of the Late Palaeozoic Ice Age in Chad.

Interaction Between Tectonics and Climate Encoded in the Planform Geometry of Stream Networks on the Eastern Tibetan Plateau

AbstractStream networks are highly abundant across Earth's surface, reflecting the tectonic and climatic history under which they have developed. Recent studies suggest that branching angles are strongly correlated with climatic aridity. However, the impact of tectonic forcing, especially in tectonically active regions, remains ambiguous. Here we analyze branching angles between headwater channels of the major river networks on the eastern Tibetan Plateau, a region with complex tectonics, variable climate, and diverse landscapes. We find that spatial variations in tectonic uplift (as reflected in channel gradients) shape the branching geometry of stream networks on the steep eastern margin, while in the flat interior of the eastern Tibetan Plateau, branching angles are mainly controlled by climatic aridity. This leads to the conclusion that, in the steep margin of the eastern Tibetan Plateau, climatic impacts on branching angles are overprinted by stronger tectonic controls.

Enhanced Spatial Stream of Two-Stream Network Using Optical Flow for Human Action Recognition

Introduction: Convolutional neural networks (CNNs) have maintained their dominance in deep learning methods for human action recognition (HAR) and other computer vision tasks. However, the need for a large amount of training data always restricts the performance of CNNs. Method: This paper is inspired by the two-stream network, where a CNN is deployed to train the network by using the spatial and temporal aspects of an activity, thus exploiting the strengths of both networks to achieve better accuracy. Contributions: Our contribution is twofold: first, we deploy an enhanced spatial stream, and it is demonstrated that models pre-trained on a larger dataset, when used in the spatial stream, yield good performance instead of training the entire model from scratch. Second, a dataset augmentation technique is presented to minimize overfitting of CNNs, where we increase the dataset size by performing various transformations on the images such as rotation and flipping, etc. Results: UCF101 is a standard benchmark dataset for action videos, and our architecture has been trained and validated on it. Compared with the other two-stream networks, our results outperformed them in terms of accuracy.

Riverine organic matter functional diversity increases with catchment size

A large amount of dissolved organic matter (DOM) is transported to the ocean from terrestrial inputs each year (~0.95 Pg C per year) and undergoes a series of abiotic and biotic reactions, causing a significant release of CO2. Combined, these reactions result in variable DOM characteristics (e.g., nominal oxidation state of carbon, double-bond equivalents, chemodiversity) which have demonstrated impacts on biogeochemistry and ecosystem function. Despite this importance, however, comparatively few studies focus on the drivers for DOM chemodiversity along a riverine continuum. Here, we characterized DOM within samples collected from a stream network in the Yakima River Basin using ultrahigh-resolution mass spectrometry (i.e., FTICR-MS). To link DOM chemistry to potential function, we identified putative biochemical transformations within each sample. We also used various molecular characteristics (e.g., thermodynamic favorability, degradability) to calculate a series of functional diversity metrics. We observed that the diversity of biochemical transformations increased with increasing upstream catchment area and landcover. This increase was also connected to expanding functional diversity of the molecular formula. This pattern suggests that as molecular formulas become more diverse in thermodynamics or degradability, there is increased opportunity for biochemical transformations, potentially creating a self-reinforcing cycle where transformations in turn increase diversity and diversity increase transformations. We also observed that these patterns are, in part, connected to landcover whereby the occurrence of many landcover types (e.g., agriculture, urban, forest, shrub) could expand DOM functional diversity. For example, we observed that a novel functional diversity metric measuring similarity to common freshwater molecular formulas (i.e., carboxyl-rich alicyclic molecules) was significantly related to urban coverage. These results show that DOM diversity does not decrease along stream networks, as predicted by a common conceptual model known as the River Continuum Concept, but rather are influenced by the thermodynamic and degradation potential of molecular formula within the DOM, as well as landcover patterns.

Hydrological Object‐Based Flow Direction Model for Constructing a Lake‐Stream Topological System

AbstractThe topological system of lakes and streams is the basis for deriving a multitude of patterns and processes in hydrology, geomorphology and ecology. Flow direction is key to constructing a lake‐stream topological system. However, lakes and streams are important hydrological objects that should be considered since they are linked during the process of material and energy transportation. Therefore, a flow direction algorithm that can clearly express the topological relationship of lakes and streams, was proposed here. Lakes and streams were taken as the hydrological objects in stream networks and coupled into a flow direction algorithm. A topological system was established based on the proposed algorithms. Six areas on the Tibetan Plateau were chosen as sample areas with the MERIT digital elevation model (DEM) as the DEM data source. Streams derived from flow direction data in the HydroSHEDS data set (HYDROSTREAMS) and MERIT hydro data set (MERITSTREAMS), as well as streams derived from flow direction results of the D‐infinity algorithm and priority‐flood algorithm were compared with streams derived from the proposed algorithm. The topological relationship between streams and lakes was clearly expressed by the proposed algorithm. In lake areas, the proposed algorithm showed better performance, and no parallel streams were found. In nonlake areas, the stream networks derived from these algorithms were stable. The proposed algorithm offers an opportunity to treat lakes and streams differently during simulation, thus adding to the authenticity of the simulation.

Modeling Complex Concentration-Discharge Relationships with Generalized Additive Models

Concentration-discharge relationships in water chemical time series can provide important insights into sources, mobilization, and delivery of solutes and particulates into stream networks. The observed relationships are often complex, including nonlinear and hysteretic patterns reflecting seasonal, climatic, and land management changes in biogeochemical release and hydrological transport of solutes and particulates to streams. Using standard single concentration-discharge (c-q) slopes can obscure this wealth of information. In this study, we suggest a new approach using generalized additive models for evaluation of complex c-q patterns in low-frequency water quality data (only monthly or biweekly observations). We used these models to estimate c-q slopes together with their uncertainty, to provide evidence of changes in c-q behaviors and their controls. We estimated c-q slopes for a selection of Swedish streams and evaluated their nonlinear, seasonal, and temporal structure as indicators of changing hydrological or biochemical drivers.

Study on Drainage Network Pattern and Morphometric Analysis by Using Hydrology Support Algorithm

Abstract The degree of accuracy in the extraction of the drainage network and morphometric analysis of the basin plays a vital role in assessing geomorphological characteristics and hydrological processes. The extraction of the exact flow network from SRTM-DEM quantifies the drainage network, and morphometric parameters, which is the main objective of this study. The Thamirabarani river basin in Tamil Nadu, India is taken as the study area. The stream order threshold algorithm is used for stream rasterization to get the pattern of the basin’s drainage network. The hydrology-based algorithm uses a different range of threshold values to determine the stream pattern, which is comparable to the stream vectors of the toposheet. The drainage network distribution with a threshold value of 10000 gives many minor streams which imply dissimilarities with the stream network of toposheet, whereas the threshold value of 20000 shows remarkable similarity to an existing river system. The model with threshold 20000 provides a Stranhler-based drainage network with a basin area of 5664.52 km2. The basin is the fifth-order stream network with many lower-order streams. The drainage pattern is resulted in dendritic and sub-dendritic. The morphometric parameters of the basin are evaluated with the help of an aspect, slope, and contour. The slope of the study area is flat in the east and steeper in the west, along the area of Western Ghats. A bifurcation ratio ranges from 1.05 to 2.42, indicating that the drainage network is not influenced by structural disturbances in that area. The comprehensive evaluation of flow patterns along with morphometric properties obtained from this study provides data support for estimation of the volume of water in the basin, reservoir operation, and construction of hydrological structures. Also, this result can be used to compute watershed attributes normally used in hydrologic and water resources studies. The hydrology-based algorithm can be employed in other river basins to extract drainage network patterns.

Habitat development and species arrival drive succession of the benthic invertebrate community in restored urban streams

BackgroundUrban streams are characterised by species-poor and frequently disturbed communities. The recovery of heavily polluted urban streams is challenging but the simple community structure makes recolonisation patterns more transparent. Therefore, they are generally applicable model systems for recolonisation of restored streams. Principal questions of stream restoration concern the drivers and patterns of recolonisation processes. Rarely, recolonisation of restored streams is recorded for a sufficient time to observe patterns of habitat and community development in detail. Over 10 years, we monitored benthic habitat changes and macroinvertebrate communities of eight restored sites in an urban stream network that was formerly used as an open sewer and thus, almost uninhabitable for macroinvertebrates prior to restoration. We analysed changes in environmental variables and communities with a selection of multi-variate analyses and identified indicator species in successional stages.ResultsProportions of stony substrate and conductivity decreased over time since restoration, while the riparian vegetation cover increased along with the amount of sandy substrate. The communities fluctuated strongly after restoration but began to stabilise after around eight years. TITAN analysis identified 9 species, (e.g. the mayfly Cloeon dipterum and the beetle Agabus didymus), whose abundances decreased with time since restoration, and 19 species with an increasing abundance trend (e.g. several Trichopteran species, which colonised once specific habitats developed). Woody riparian vegetation cover and related variables were identified as major driver for changes in species abundance. In the last phase of the observation period, a dry episode resulted in complete dewatering of some sites. These temporarily dried sections were recolonised much more rapidly compared to the recolonisation following restoration.ConclusionsOur results underline that community changes following urban stream restoration are closely linked to the evolving environmental conditions of restored streams, in particular habitat availability initialised by riparian vegetation. It takes about a decade for the development of a rich and stable community. Even in streams that were almost completely lacking benthic invertebrates before restoration, the establishment of a diverse macroinvertebrate community is possible, underlining the potential for habitat restoration in formerly heavily polluted urban areas.

Interaction graph, topical communities, and efficient local event detection from social streams

Social networks have become an essential part of daily life, and hence every real-world activity finds its place in this virtual world. The present paper proposes a methodology to find localized micro-events from the social network stream. The method is named CommunityINDICATOR. A concept of ‘separation of concerns’ from the software design principle is incorporated in the methodology to reduce the execution time drastically from existing state-of-the-art methods of event detection. In order to reduce the execution time, the algorithm first generates an interaction graph from the social stream and applies community detection followed by a clustering algorithm onto it to detect micro-level events. Experiments have been conducted on Twitter data stream of 5 different cities on three different continents with the size of 2 million tweets. We have used well known quality metrics such as precision, recall, F1-score, accuracy, and execution time to compare performance with other state-of-the-art methodologies. The proposed CommunityINDICATOR provides up to 30% higher accuracy than EvenTweet and SEDTWik in similar execution times. An improvement of 11% to 51% and 17% to 57% in execution time is observed for the proposed algorithm in comparison with TwiiterNews and EventX, respectively, for different datasets.

Evaluating deep learning architecture and data assimilation for improving water temperature forecasts at unmonitored locations

Deep learning (DL) models are increasingly used to forecast water quality variables for use in decision making. Ingesting recent observations of the forecasted variable has been shown to greatly increase model performance at monitored locations; however, observations are not collected at all locations, and methods are not yet well developed for DL models for optimally ingesting recent observations from other sites to inform focal sites. In this paper, we evaluate two different DL model structures, a long short-term memory neural network (LSTM) and a recurrent graph convolutional neural network (RGCN), both with and without data assimilation for forecasting daily maximum stream temperature 7 days into the future at monitored and unmonitored locations in a 70-segment stream network. All our DL models performed well when forecasting stream temperature as the root mean squared error (RMSE) across all models ranged from 2.03 to 2.11°C for 1-day lead times in the validation period, with substantially better performance at gaged locations (RMSE = 1.45–1.52°C) compared to ungaged locations (RMSE = 3.18–3.27°C). Forecast uncertainty characterization was near-perfect for gaged locations but all DL models were overconfident (i.e., uncertainty bounds too narrow) for ungaged locations. Our results show that the RGCN with data assimilation performed best for ungaged locations and especially at higher temperatures (&amp;gt;18°C) which is important for management decisions in our study location. This indicates that the networked model structure and data assimilation techniques may help borrow information from nearby monitored sites to improve forecasts at unmonitored locations. Results from this study can help guide DL modeling decisions when forecasting other important environmental variables.