True Skill Statistic Research Articles

Predictive capacity of nine algorithms and an ensemble model to determine the geographic distribution of tree species

The different models that predict the distribution of species are a useful tool for the evaluation and monitoring of forest resources as they facilitate the planning of their management in a changing climate environment. Recently, a significant number of algorithms have been proposed for this purpose, making it difficult to select the most appropriate to use. The evaluation of performance and predictive stability of these models can elucidate this problem. Distribution data of 17 pine species with high economic importance for Mexico were collected and distribution models were carried out. We carried out a pre-modeling design to select the prediction variables (climatic, edaphic and topographic), after which nine algorithms and an ensemble model were contrasted against one another. The true skill statistic (TSS) and the area under the curve (AUC) were used to evaluate the predictive performance of the models, and the coefficient of variation of the predictions was used to evaluate their stability. The number of predictive variables in the final models fluctuated from 6 to 12; the mean diurnal range and the maximum temperature of warmest month were included in the models for most species. Random forests, the ensemble model, generalized additive models and MaxEnt were the ones that best described the distribution of the species (AUC >0.92 and TSS >0.72); the opposite was found in Bioclim and Domain (AUC<0.75 and <0.82; and TSS<0.5 and <0.55). Support vector machine, Mahalanobis distance, generalized linear models and boosted regression trees obtained intermediate settings. The coefficient of variation indicated that Bioclim, Domain and Support vector machine have low predictive stability (CV>0.055); on the contrary, Maxent and the ensemble model attained high predictive stability (CV<0.015). The ensemble model obtained greater performance and predictive stability in the predictions of the distribution of the 17 species of pines. The differences found in performance and predictive stability of the algorithms suggest that the ensemble model has the potential to model the distribution of tree species.

Dust source susceptibility mapping based on remote sensing and machine learning techniques

Dust source susceptibility modeling and mapping is the first step in managing the impacts of dust on environmental systems and human health. In this study, satellite products and terrestrial data were used to detect dust sources in central Iran using remote sensing and machine learning techniques. After recording 890 sites as dust sources based on field surveys and determining 14 independent variables affecting wind erosion and dust sources, dust source distribution maps were prepared through GLM (Generalized Linear Model), CTA (Classification Tree Analysis), ANN (Artificial Neural Network), MARS (Multivariate Adaptive Regression Spline), RF (Random Forest), Maxent (Maximum Entropy), and ensemble algorithms. Specifically, 70% of dust source sites were used as training data and 30% were used for algorithm performance evaluation through different statistical methods such as partial ROC (Receiver Operator Characteristic), sensitivity, specificity, and TSS (True Skill Statistics). According to the results, following the ensemble model, RF had the highest and GLM had the lowest performance in dust source detection. According to the ensemble model, precipitation with a mean weight of 0.3 followed by evaporation, temperature, and soil moisture with mean weight of 0.173, 0.16, and 0.153, respectively, were the main driving forces in dust susceptibility mapping. This model classified 40.92% of the study area with low potential, 15.37% with medium potential, 25.77% with high potential, and 17.94% with very high potential. The research findings indicate that the integration of remote sensing and prediction algorithms can be used as a useful means for predicting the spatial distribution of dust sources in arid and semi-arid regions.

Seeing from space makes sense: Novel earth observation variables accurately map species distributions over Himalaya

Topical advances in earth observation have enabled spatially explicit mapping of species' fundamental niche limits that can be used for nature conservation and management applications. This study investigates the possibility of applying functional variables of ecosystem retrieved from Moderate Resolution Imaging Spectroradiometer (MODIS) onboard sensor data to map the species distribution of two alpine treeline species, namely Betula utilis D.Don and Rhododendron campanulatum D.Don over the Himalayan biodiversity hotspot. In this study, we have developed forty-nine Novel Earth Observation Variables (NEOVs) from MODIS products, an asset to the present investigation. To determine the effectiveness and ecological significance of NEOVs combinations, we built and compared four different models, namely, a bioclimatic model (BCM) with bioclimatic predictor variables, a phenology model (PhenoM) with earth observation derived phenological predictor variables, a biophysical model (BiophyM) with earth observation derived biophysical predictor variables, and a hybrid model (HM) with a combination of selected predictor variables from BCM, PhenoM, and BiophyM. All models utilized topographical variables by default. Models that include NEOVs were competitive for focal species, and models without NEOVs had considerably poor model performance and explanatory strength. To ascertain the accurate predictions, we assessed the congruence of predictions by pairwise comparisons of their performance. Among the three machine learning algorithms tested (artificial neural networks, generalised boosting model, and maximum entropy), maximum entropy produced the most promising predictions for BCM, PhenoM, BiophyM, and HM. Area under curve (AUC) and true skill statistic (TSS) scores for the BCM, PhenoM, BiophyM, and HM models derived from maximum entropy were AUC ≥0.9 and TSS ≥0.6 for the focal species. The overall investigation revealed the competency of NEOVs in the accurate prediction of species' fundamental niches, but conventional bioclimatic variables were unable to achieve such a level of precision. A principal component analysis of environmental spaces disclosed that niches of focal species substantially overlapped each other. We demonstrate that the use of satellite onboard sensors’ biotic and abiotic variables with species occurrence data can provide precision and resolution for species distribution mapping at a scale that is relevant ecologically and at the operational scale of most conservation and management actions.

Determining the potential distribution of Oryctes monoceros and Oryctes rhinoceros by combining machine-learning with high-dimensional multidisciplinary environmental variables

The African coconut beetle Oryctes monoceros and Asiatic rhinoceros beetle O. rhinoceros have been associated with economic losses to plantations worldwide. Despite the amount of effort put in determining the potential geographic extent of these pests, their environmental suitability maps have not yet been well established. Using MaxEnt model, the potential distribution of the pests has been defined on a global scale. The results show that large areas of the globe, important for production of palms, are suitable for and potentially susceptible to these pests. The main determinants for O. monoceros distribution were; temperature annual range, followed by land cover, and precipitation seasonality. The major determinants for O. rhinoceros were; temperature annual range, followed by precipitation of wettest month, and elevation. The area under the curve values of 0.976 and 0.975, and True skill statistic values of 0.90 and 0.88, were obtained for O. monoceros and O. rhinoceros, respectively. The global simulated areas for O. rhinoceros (1279.00 × 104 km2) were more than that of O. monoceros (610.72 × 104 km2). Our findings inform decision-making and the development of quarantine measures against the two most important pests of palms.

Predicting the invasion risk of rugose spiraling whitefly, Aleurodicus rugioperculatus, in India based on CMIP6 projections by MaxEnt.

Rugose spiraling whitefly (RSW), Aleurodicus rugioperculatus Martin, is a highly polyphagous invasive pest native to Central America. The occurrence of A. rugioperculatus in the Oriental region was first reported from Pollachi, Tamil Nadu, India in 2017. This pest is widely distributed in India, causing severe economic damage to coconut and other horticultural crops. It is a recent invasion in India and information on its potential distribution is lacking. Thus, in the present study we used the latest Coupled Model Intercomparison Project phase 6 (CMIP6) dataset through Maximum Entropy species distribution modelling (version 3.4.1, MaxEnt) to determine the potential distribution of RSW in present and future climate change scenarios in 2050 and 2070 under Shared Socioeconomic Pathway (SSP) 126 and SSP585 emission scenarios. The performance of the model was evaluated using the area under the curve (AUC), true skill statistics (TSS) and the continuous Boyce index (CBI). The MaxEnt model performed well and predicted the potential distribution of A. rugioperculatus with high-accuracy AUC values of 0.991 and 0.989, TSS values of 0.891 and 0.842, and CBI values of 0.972 and 0.934 for training and testing, respectively. Jackknife analysis revealed that A. rugioperculatus distribution was mostly influenced by temperature-based bioclimatic variables contributing 62.1% of the suitability, with precipitation variables contributing the remainder. The most important bioclimatic variables for RSW distribution were annual mean temperature (Bio 1, 28.9%) followed by mean diurnal range (Bio 2, 19.5%) and annual precipitation (Bio 12, 19.1). Potential suitable areas for RSW establishment were mostly found in all coastal and southern states of India. A. rugioperculatus prefers a warm and humid climate, indicating that the tropics, subtropics and temperate regions are ideal for its spread and invasion. Our results highlighted that the suitable habitat area for A. rugioperculatus is predicted to increase and highest probability of invasion and spread in 2050 and 2070 under future climate change scenarios of SSP126 and SSP585 compared to present climatic conditions. This is the first study to use the latest CMIP6 models and it predicts the potential distribution of RSW in India under present and future climate change scenarios. The implementation of strict domestic quarantine measures may prevent the spread and damage of RSW to noncoastal regions of India. The results of the current study should help in timely monitoring and surveillance of RSW and to formulate integrated pest management strategies at the national level to restrict its spread, invasion and damage to new areas. © 2022 Society of Chemical Industry.

Automatic Detection of Interplanetary Coronal Mass Ejections in Solar Wind In Situ Data

AbstractInterplanetary coronal mass ejections (ICMEs) are one of the main drivers for space weather disturbances. In the past, different approaches have been used to automatically detect events in existing time series resulting from solar wind in situ observations. However, accurate and fast detection still remains a challenge when facing the large amount of data from different instruments. For the automatic detection of ICMEs we propose a pipeline using a method that has recently proven successful in medical image segmentation. Comparing it to an existing method, we find that while achieving similar results, our model outperforms the baseline regarding training time by a factor of approximately 20, thus making it more applicable for other datasets. The method has been tested on in situ data from the Wind spacecraft between 1997 and 2015 with a True Skill Statistic of 0.64. Out of the 640 ICMEs, 466 were detected correctly by our algorithm, producing a total of 254 false positives. Additionally, it produced reasonable results on datasets with fewer features and smaller training sets from Wind, STEREO‐A, and STEREO‐B with TSSs of 0.56, 0.57, and 0.53, respectively. Our pipeline manages to find the start of an ICME with a mean absolute error (MAE) of around 2 hr and 56 min, and the end time with a MAE of 3 hr and 20 min. The relatively fast training allows straightforward tuning of hyperparameters and could therefore easily be used to detect other structures and phenomena in solar wind data, such as corotating interaction regions.

Climate change and dispersion dynamics of the invasive plant species Chromolaena odorata and Lantana camara in parts of the central and eastern India

Lantana camara and Chromolaena odorata are categorized as the most obnoxious invasive flora globally. Their ability to combat the regeneration and proliferation of neighbouring flora, expansive nature, and robust adaptability to diverse habitats, drew global attention. Investigating the potential mutual dispersion phenomenon of these two invasive species under the climate change scenario was the primary objective of this study. The present and future (2050) prospective distribution scenarios for these two species were determined using MaxEnt in the eastern and central Indian regions encompassing the states of Jharkhand, Chhattisgarh, and West Bengal. Future projections for 2050 were derived using IPSL-CM5A-LR & MIROC5 and IPSL-CM6A-LR & MIROC6 models for different representative concentration pathways (RCPs 2.6, 4.5, 6.0, and 8.5), and Shared Socioeconomic Pathways (SSPs 126, 245, 370 and 585), respectively. The investigation revealed that currently ∼31% and ∼ 24% of the study area are susceptible to infestation of L. camara and C. odorata, respectively. Compared to the current scenario, the results showed a probable future increase of ∼1.53% in C. odorata infestation and a decrease of ∼4.95% for L. camara. The True Skill Statistics (TSS) and Kappa coefficient (in %) values of 0.71 & 76.50 for L. camara and 0.52 & 63.38 for C. odorata indicated a good model fit. Collectively, both the species exhibited robust resilience to climate change, with C. odorata outcompeting L. camara. Using both RCP and SSP pathways under the multiple climate scenarios offered a comprehensive and novel approach to acquiring greater insights into likely interactions, dominance, and distribution scenarios of these species. The results provide prior information on sensitive sites prone to future invasion, allowing management to formulate preventative measures to control infestation.

Fine-scale ensemble species distribution modeling of eelgrass (Zostera marina) to inform nearshore conservation planning and habitat management

Baseline data on the distribution and extent of biogenic habitat-forming species at a high spatial resolution are essential to inform habitat management strategies, preserve ecosystem integrity, and achieve effective conservation objectives in the nearshore. Model-based approaches to map suitable habitat for these species are a key tool to address this need, filling in gaps where observations are otherwise unavailable and remote sensing methods are limited by turbid waters or cannot be applied at scale. We developed a high resolution (35 m) ensemble species distribution model to predict the distribution of eelgrass (Zostera marina) along the Atlantic coast of Nova Scotia, Canada where the observational coverage of eelgrass occurrence is sparse and nearshore waters are optically complex. Our ensemble model was derived as a performance-weighted average prediction of 7 different modeling methods fit to 6 physical predictors (substrate type, depth, wave exposure, slope, and two bathymetric position indices) and evaluated with a 5-fold spatially-blocked cross-validation procedure. The ensemble model showed moderate predictive performance (Area Under the Receiver-Operating Characteristic Curve (AUC) = 0.803 ± 0.061, True Skill Statistic (TSS) = 0.531 ± 0.100; mean ± SD), high sensitivity (92.0 ± 4.5), and offered some improvement over individual models. Substrate type, depth, and relative wave exposure were the most influential predictors associated with eelgrass occurrence, where the highest probabilities were associated with sandy and sandy-mud sediments, depths ranging 0 m – 4 m, and low to intermediate wave exposure. Within our study region, we predicted a total extent of suitable eelgrass habitat of 38,130 ha. We found suitable habitat was particularly extensive within the long narrow inlets and extensive shallow flats of the South Shore, Eastern Shore, and Bras d’Or Lakes. We also identified substantial overlap of eelgrass habitat with previously identified Ecologically and Biologically Significant Areas that guide regional conservation planning while also highlighting areas of greater prediction uncertainty arising from disagreement among modeling methods. By offering improved sensitivity and insights into the fine-scale regional distribution of a habitat-forming species with associated uncertainties, our ensemble-based modeling approach provides improved support to numerous nearshore applications including conservation planning and restoration, marine spatial and emergency response planning, environmental impact assessments, and fish habitat protection.

Ecological modelling for the conservation of Gluta travancorica Bedd. - An endemic tree species of southern Western Ghats, India

Endemic species are highly adapted to grow exclusively in a specific geographical area. The goal of the current study is to determine the probable habitat distribution range of the narrowly endemic species Gluta travancorica. An ecological niche modelling is carried out, using four different models viz., BioClim, MaxEnt, Random Forest and Deep Neural Networks (DNN). A total of 506 G. travancorica cluster locations were surveyed and used for this study with thirty different ecogeographic, edaphic and bioclimatic environmental parameters. After a preliminary investigation using multi-collinearity analysis, soil parameter variables like pH, cation exchange capacity (CEC), silt and clay content are used for final modelling. Factor analysis of ecological niche revealed the soil parameters like pH, CEC, silt and clay content as the key predictors. The result is validated using true skill statistics, sensitivity, specificity, kappa statistic and AUC-ROC. Results of the present study show that DNN have exceptional prediction performance, demonstrated by its AUC score of 0.959. DNN model projected 32.37% (938.18 km2) of the study region to have a ‘highly suitable habitat’, whereas 67.63% (1960.82 km2) was classified as having ‘low habitat suitability’. Besides, back-to-field assessments have also proven DNN's potential in predicting the habitat suitability of G. travancorica. The study results will facilitate the prioritization of conservation and seedling restoration strategies. The forest range explored in this work is a component of one of the most important global biodiversity hotspots, and it has significant implications for regional biodiversity conservation.

Ensemble modeling approach to predict the past and future climate suitability for two mangrove species along the coastal wetlands of peninsular India

Mangroves support numerous ecosystem services and help in reducing coastal ecological risks, yet they are declining rapidly due to climate change, sea level fluctuations and human activities. It is important to understand their responses to climate and sea level changes and identify conservation target areas at spatio-temporal scales, specifically in regions of rich mangrove biodiversity. In this study, we predicted the potential impact of past (Middle Holocene, ∼6000 years), current and future (2050s, 2070s; RCP 2.6 and RCP 8.5) climate change scenarios on the two dominant species in the coastal mangrove forest wetlands of India, i.e., Rhizophora mucronata and Avicennia officinalis through an ensemble species distribution modeling approach. The ensemble modeling has been carried out by integrating eight single algorithm methods. Based on the receiver operating characteristics of area under the curve (AUC) and true skill statistics (TSS) values the ensemble modeling has yielded the highest predictive performance for SVM for both the species and lowest by CART for R. mucronata and BIOCLIM for A. officinalis. The internal evaluation metrics of the resulting Species distribution models (SDMs) tested its robustness with AUC-0.97 and TSS-0.89 for A. officinalis and AUC-0.99 and TSS-0.90 for R. mucronata. Precipitation of Wettest Month (Bio 13) and Mean Temperature of Warmest Quarter (Bio 10) was the most important variable (54–67%) for the distribution of A. officinalis and Precipitation Seasonality (Bio 15) and Precipitation of Warmest Quarter (Bio 18) for R. mucronata. High precipitation and sea-level highstand during middle Holocene led to the maximum range expansion of suitable habitat for the mangrove species which is also validated in the present study by the fossil pollen datasets. Total mangrove habitat in current and future climatic scenarios decreased in 2.6 and 8.5 Representative Concentration Pathways (RCPs) for 2050 and 2070 which indicates the vulnerability of the species to climate change impacts. Mangrove species are projected to shift their ranges more towards land in future experiencing a decrease in the amount of suitable coastal area available to them throughout the Indian coastline. The plausible cause for this range shift may be due to higher precipitation that is usually associated with longer period of soil inundation and because of the rise in sea level. Our findings will assist in formulating species-specific restoration plans for these keystone species in context of climate change in the Indian Subcontinent.

The current and future distribution of the yellow fever mosquito (Aedes aegypti) on Madeira Island.

The Aedes aegypti mosquito is the main vector for several diseases of global importance, such as dengue and yellow fever. This species was first identified on Madeira Island in 2005, and between 2012 and 2013 was responsible for an outbreak of dengue that affected several thousand people. However, the potential distribution of the species on the island remains poorly investigated. Here we assess the suitability of current and future climatic conditions to the species on the island and complement this assessment with estimates of the suitability of land use and human settlement conditions. We used four modelling algorithms (boosted regression trees, generalized additive models, generalized linear models and random forest) and data on the distribution of the species worldwide and across the island. For both climatic and non-climatic factors, suitability estimates predicted the current distribution of the species with good accuracy (mean area under the Receiver Operating Characteristic curve = 0.88 ±0.06, mean true skill statistic = 0.72 ±0.1). Minimum temperature of coldest month was the most influential climatic predictor, while human population density, residential housing density and public spaces were the most influential predictors describing land use and human settlement conditions. Suitable areas under current climates are predicted to occur mainly in the warmer and densely inhabited coastal areas of the southern part of the island, where the species is already established. By mid-century (2041–2060), the extent of climatically suitable areas is expected to increase, mainly towards higher altitudes and in the eastern part of the island. Our work shows that ongoing efforts to monitor and prevent the spread of Ae. aegypti on Madeira Island will have to increasingly consider the effects of climate change.

An Improved Method for Rainfall Forecast Based on GNSS-PWV

Global navigation satellite systems (GNSS) has been applied to the sounding of precipitable water vapor (PWV) due to its high accuracy and high spatiotemporal resolutions. PWV obtained from GNSS (GNSS-PWV) can be used to investigate extreme weather phenomena, such as the formation mechanism and prediction of rainfalls. In the study, a new, improved model for rainfall forecasting was developed based on GNSS data and rainfall data for the 9-year period from 2010 to 2018 at 66 stations located in the USA. The new model included three prediction factors—PWV value, PWV increase, maximum hourly PWV increase. The two key tasks involved for the development of the model were the determination of the thresholds for each prediction factor and the selection of the optimal strategy for using the three prediction factors together. For determining the thresholds, both critical success index (CSI) and true skill statistic (TSS) were tested, and results showed that TSS outperformed CSI for all rainfall events tested. Then, various strategies by combining the three prediction factors together were also tested, and results indicated that the best forecast result was from the case that any two of the prediction factors were over their own thresholds. Finally, the new model was evaluated using the GNSS data for the 2-year period from 2019 to 2020 at the above mentioned 66 stations, and the probability of detection (POD) and false-alarms rate (FAR) were adopted to measure the model performances. Over the 66 stations, the POD values ranged from 73% to 97% with the mean of 87%, and the FARs ranged from 26% to 77% with the mean of 53%. Moreover, it was also found that both POD and FAR values were related to the region of the station; e.g., the results at the stations that are located in humid regions were better than the ones located in dry regions. All these results suggest the feasibility and good performance of using GNSS-PWV for forecasting rainfall.

Countrywide classification of permanent grassland habitats at high spatial resolution

AbstractEuropean grasslands face strong declines in extent and quality. Many grassland types are priority habitats for national and European conservation strategies. Countrywide, high spatial resolution maps of their distribution are often lacking. Here, we modelled the spatial distribution of 20 permanent grassland habitats at the level of phytosociological alliances across Switzerland at 10x10 m resolution. First, we applied ensemble models to provide distribution maps of the individual habitat types, using training data from various sources. Copernicus Sentinel satellite imagery and variables describing climate, soil and topography were used as predictors. The performance of these models was assessed based on the true skill statistics with a split‐sampling of the data. Second, the individual maps were combined into countrywide maps of the most and second most likely habitat type, respectively, using an expert‐based weighting approach. The performance of the combined map for the most likely habitat type was assessed via an independent testing dataset and a comparison of the predicted habitat‐type proportions with extrapolations from field surveys. Most individual maps had useful to excellent predictive performance (TSS ≥ 0.6). For most grid cells in the combined maps, the most and second most likely habitat types were either ecologically closely related or representing two grassland types along a nutrient gradient. The same was true for omission errors. We found good agreement between the predicted and estimated proportions from field surveys. The area of raised bogs appears to be underestimated, while dry grasslands showed highest agreement. This work highlights the potential of earth observation data at fine spatial and temporal resolution to map habitats at broad scales, thereby providing the foundation for diverse conservation applications. A particular challenge remains in capturing the transition from nutrient‐poor to nutrient‐rich grasslands, which is highly important for biodiversity conservation.

MaxEnt-based prediction of the potential invasion of Lantana camara L. under climate change scenarios in Arunachal Pradesh, India

Lantana camara L. is a weed of global importance due to its widespread distribution and far-reaching impact on ecosystems. In this study, we predicted the current niche suitability of the weed and its potential invasion using the Maximum Entropy (MaxEnt) model for two future climatic years 2050 and 2070 under three representative concentration pathways (RCP 2.6, 4.5, and 8.5) in Arunachal Pradesh, India. Out of the total 144 occurrence records and 22 environmental variables (19 bioclimatic and 3 topographic), 61 occurrence points and seven environmental variables were incorporated into the final model after removing biases, multicollinearity test, and contribution analysis. The model performed well with mean area under curve (AUC) = 0.864 to 0.882 and true skill statistic (TSS) = 0.738. The most contributing variables were the temperature of the wettest quarter, altitude, and precipitation of the warmest quarter. The results predicted a considerable range expansion of L. camara with an area of 14,331.95 km2 (current to 2050 under RCP 4.5) to 25,560.55 km2 (current to 2070 under RCP 8.5) towards the southern parts which are characterized by virgin tropical forests and cultivable areas. So, the findings would be helpful to the planners and managers in evolving appropriate strategies for containing the invasion of L. camara in the state.

Habitat suitability mapping of the Indian giant flying squirrel (Petaurista philippensis Elliot, 1839) in India with ensemble modeling

Flying squirrels are one of the least studied mammalian taxa in South Asia as well as in India owing to their nocturnal, arboreal and cryptic nature. We applied ensemble species distribution modeling using BIOMOD2 (ver 1.0) R-package to predict the suitable habitat of the Indian giant flying squirrel (Petaurista philippensis, Elliot 1839) in Indian biogeographic regions and states, and to identify the physical variables that define its fundamental niche. A subset of least correlated variables from bioclimatic (mean diurnal temperature range, temperature isothermality, precipitation of wettest month, precipitation of driest month and precipitation of warmest quarter, and precipitation of coldest quarter), topographic (elevation), water-related (distance to major water streams), and vegetation-related (NDVI, and vegetation height) data sets were used in the modeling. The accuracy of the final ensemble habitat suitability model was characterized by the receiver operating characteristic curve (ROC) and true skill statistic (TSS). In India, a total of highly suitable habitat for the species was estimated to be 66,743 sq. km, varying between states and biogeographic regions. Final model revealed that predictive suitable habitat of the species was limited primarily to the southern peninsula, including the Western Ghats, the Deccan peninsula, and the semi-arid region, and to a lesser extent, the coastal areas of the Kerala state. At the state level, majority of suitable habitat for the species estimated in Kerala followed by Karnataka, Tamil Nadu, Gujarat, Maharashtra, Rajasthan, and Madhya Pradesh. Climatic factors, particularly temperature and precipitation, have been identified as major determinants of Indian giant flying squirrel's potential habitat. We suggest that several new areas in the modeling that showed suitable habitat for the species necessitate immediate attention to validate species occurrence through field work. Extensive further studies on the species could also provide a detailed insight into the species' dispersion, colonization and the effect of future climate change in the Indian subcontinent.

Towards coupling full-disk and active region-based flare prediction for operational space weather forecasting

Solar flare prediction is a central problem in space weather forecasting and has captivated the attention of a wide spectrum of researchers due to recent advances in both remote sensing as well as machine learning and deep learning approaches. The experimental findings based on both machine and deep learning models reveal significant performance improvements for task specific datasets. Along with building models, the practice of deploying such models to production environments under operational settings is a more complex and often time-consuming process which is often not addressed directly in research settings. We present a set of new heuristic approaches to train and deploy an operational solar flare prediction system for ≥M1.0-class flares with two prediction modes: full-disk and active region-based. In full-disk mode, predictions are performed on full-disk line-of-sight magnetograms using deep learning models whereas in active region-based models, predictions are issued for each active region individually using multivariate time series data instances. The outputs from individual active region forecasts and full-disk predictors are combined to a final full-disk prediction result with a meta-model. We utilized an equal weighted average ensemble of two base learners’ flare probabilities as our baseline meta learner and improved the capabilities of our two base learners by training a logistic regression model. The major findings of this study are: 1) We successfully coupled two heterogeneous flare prediction models trained with different datasets and model architecture to predict a full-disk flare probability for next 24 h, 2) Our proposed ensembling model, i.e., logistic regression, improves on the predictive performance of two base learners and the baseline meta learner measured in terms of two widely used metrics True Skill Statistic (TSS) and Heidke Skill Score (HSS), and 3) Our result analysis suggests that the logistic regression-based ensemble (Meta-FP) improves on the full-disk model (base learner) by ∼9% in terms TSS and ∼10% in terms of HSS. Similarly, it improves on the AR-based model (base learner) by ∼17% and ∼20% in terms of TSS and HSS respectively. Finally, when compared to the baseline meta model, it improves on TSS by ∼10% and HSS by ∼15%.

Brown Bear Food-Probability Models in West-European Russia: On the Way to the Real Resource Selection Function

Most habitat suitability models and resource selection functions (RSFs) use indirect variables and habitat surrogates. However, it is known that in order to adequately reflect the habitat requirements of a species, it is necessary to use proximal resource variables. Direct predictors should be used to construct a real RSF that reflects the real influence of main resources on species habitat use. In this work, we model the spatial distribution of the main food resources of brown bear Ursus arctos L. within the natural and human-modified landscapes of the Central Forest State Nature Reserve (CFNR) for further RSF construction. Food-probability models were built for Apiaceae spp. (Angelica sylvestris L., Aegopodium podagraria L., Chaerophyllum aromaticum L.), Populus tremula L., Vaccinium myrtillus L., V. microcarpum (Turcz. ex Rupr.) Schmalh., V. oxycoccos L., Corylus avellana L., Sorbus aucuparia L., Malus domestica Borkh., anthills, xylobiont insects, social wasps and Alces alces L. using the MaxEnt algorithm. For model evaluation, we used spatial block cross-validation and held apart fully independent data. The true skill statistic (TSS) estimates ranged from 0.34 to 0.95. Distribution of Apiaceae forbs was associated with areas having rich phytomass and moist conditions on southeastern slopes. Populus tremula preferred areas with phytomass abundance on elevated sites. Vaccinium myrtillus was confined to wet boreal spruce forests. V. microcarpum and V. oxycoccos were associated with raised bogs in depressions of the terrain. Corylus avellana and Sorbus aucuparia preferred mixed forests on elevated sites. Distribution of Malus domestica was associated with meadows with dry soils in places of abandoned cultural landscapes. Anthills were common on the dry soils of meadows, and the periphery of forest areas with high illumination and low percent cover of tree canopy. Moose preferred riverine flood meadows rich in herbaceous vegetation and sparse mixed forests in spring and early summer. The territory of the human-modified CFNR buffer zone was shown to contain a higher variety of food resources than the strictly protected CFNR core area.

Predicting the potential geographical distribution of Ageratina adenophora in China using equilibrium occurrence data and ensemble model

Invasive alien plants (IAPs) pose a significant threat to the ecological environment and agricultural production in China. Ageratina adenophora is one of the most aggressive IAPs in China and poses serious ecological and socioeconomic threats. Estimating the distribution pattern of A. adenophora in China can provide baseline data for preventing damage by this weed. In the present study, based on the equilibrium occurrence data of A. adenophora in China and related environmental variables, we used an ensemble model to predict the distribution pattern of A. adenophora in China under climate change. Our findings indicated that true skill statistic (TSS), area under the receiver operating characteristic (ROC) curve (AUC), and Cohen’s Kappa (KAPPA) values for the ensemble model were 0.925, 0.993, and 0.936, respectively. The prediction results of the ensemble model were more accurate than those of the single models. Temperature variables had a significant impact on the potential geographical distribution (PGD) of A. adenophora in China. The total, high, and moderate suitability habitat areas of A. adenophora in China were 153.82 × 104, 92.13 × 104, and 21.04 × 104 km2, respectively, accounting for 16.02, 9.60, and 2.19% of the Chinese mainland area, respectively. The PGD of A. adenophora in China under the current climate is mainly located in southwestern and southeastern China, which are located in the tropical and subtropical zone. The high-suitability habitat areas of A. adenophora decreased under the future climate scenarios, mainly by changing to moderately suitable habitats in Southwest China. The geographical distribution of A. adenophora in southwestern China is currently saturated and will spread to southeastern China under climate change in the future. More attention should be paid to early warning and monitoring of A. adenophora in southeastern China to prevent its further spread.

Probabilistic Prediction of Dst Storms One‐Day‐Ahead Using Full‐Disk SoHO Images

AbstractWe present a new model for the probability that the disturbance storm time (Dst) index exceeds −100 nT, with a lead time between 1 and 3 days. Dst provides essential information about the strength of the ring current around the Earth caused by the protons and electrons from the solar wind, and it is routinely used as a proxy for geomagnetic storms. The model is developed using an ensemble of Convolutional Neural Networks that are trained using Solar and Heliospheric Observatory (SoHO) images (Michelson Doppler Imager, Extreme ultraviolet Imaging Telescope, and Large Angle and Spectrometric Coronagraph). The relationship between the SoHO images and the solar wind has been investigated by many researchers, but these studies have not explicitly considered using SoHO images to predict the Dst index. This work presents a novel methodology to train the individual models and to learn the optimal ensemble weights iteratively, by using a customized class‐balanced mean square error (CB‐MSE) loss function tied to a least‐squares based ensemble. The proposed model can predict the probability that Dst &lt; −100 nT 24 hr ahead with a True Skill Statistic (TSS) of 0.62 and Matthews Correlation Coefficient (MCC) of 0.37. The weighted TSS and MCC is 0.68 and 0.47, respectively. An additional validation during non‐Earth‐directed CME periods is also conducted which yields a good TSS and MCC score.

Predicting habitat suitability and niche dynamics of Dactylorhiza hatagirea and Rheum webbianum in the Himalaya under projected climate change

In the era of anthropocene, global warming tends to alter the distribution range of the plant species. Highly fragile to such changes are the species that are endemic, inhabit higher elevations and show narrow distribution ranges. Predicting and plotting the appropriate suitable habitats and keeping knowledge of how climate change will affect future distribution become imperative for designing effective conservation strategies. In the current study we have used BIOMOD ensemble forecasting to study the current and predict the future potential distribution of Dactylorhiza hatagirea and Rheum webbianum and describe their niche dynamics in Himalayan biodiversity hotspots under climate change scenarios using ecospat R package. Results reveal sufficient internal evaluation metrics with area under curve (AUC) and true skill statistic (TSS) values greater than 0.8 i.e. 0.93 and 0.98 and 0.82 and 0.90 for D. hatageria and R. webbianum respectively, which signifies robustness of the model. Among different bioclimatic variables, bio_1, bio_3, bio_8, bio_14 and bio_15 were the most influential, showing greater impact on the potential distribution of these plant species. Range change analysis showed that both the studied species will show significant contraction of their suitable habitats under future climatic scenarios. Representative Concentration Pathway (RCP) 8.5 for the year 2070, indicate that the suitable habitats could be reduced by about 51.41% and 70.57% for D. hatagirea and R. webbianum respectively. The results of the niche comparisons between the current and future climatic scenarios showed moderate level of niche overlap for all the pairs with D. hatageria showing 61% overlap for current vs. RCP4.5 2050 and R. webbianum reflects 68% overlap for current vs. RCP4.5 2050. Furthermore, the PCA analysis revealed that climatic conditions for both the species vary significantly between current and future scenarios. The similarity and equivalence test showed that the niche between present and future climate change scenarios is comparable but not identical. From the current study we concluded that the influence of climate change on the habitat distribution of these plant species in the Himalayan biodiversity hotspots can be considered very severe. Drastic reduction in overall habitat suitability poses a high risk of species extinction and thereby threatens to alter the functions and services of these fragile ecosystems. Present results can be used by conservationists for mitigating the biodiversity decline and exploring undocumented populations on one hand and by policymakers in implementing the policy of conservation of species by launching species recovery programmes in future on the other. The outcomes of this study can contribute substantially to understand the consequences of climate change in the Himalayan biodiversity hotspots.