Glaciers playa vital role in providing water resources for drinking, agriculture, and hydro-electricity in many mountainous regions. As global warming progresses, accurately reconstructing long-term glacier mass changes and comprehending their intricate dynamic relationships with environmental variables are imperative for sustaining livelihoods in these regions. This paper presents the use of eXplainable Machine Learning (XML) models with GRACE and GRACE-FO data to reconstruct long-term monthly glacier mass changes in the Upper Yukon Watershed (UYW), Canada. We utilized the H2O-AutoML regression tools to identify the best performing Machine Learning (ML) model for filling missing data and predicting glacier mass changes from hydroclimatic data. The most accurate predictive model in this study, the Gradient Boosting Machine, coupled with explanatory methods based on SHapley Additive eXplanation (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) analyses, led to automated XML models. The XML unveiled and ranked key predictors of glacier mass changes in the UYW, indicating a decrease since 2014. Analysis showed decreases in snow water equivalent, soil moisture storage, and albedo, along with increases in rainfall flux and air temperature were the main drivers of glacier mass loss. A probabilistic analysis hinging on these drivers suggested that the influence of the key hydrological features is more critical than the key meteorological features. Examination of climatic oscillations showed that high positive anomalies in sea surface temperature are correlated with rapid depletion in glacier mass and soil moisture, as identified by XML. Integrating H2OAutoML with SHAP and LIME not only achieved high prediction accuracy but also enhanced the explainability of the underlying hydroclimatic processes of glacier mass change reconstruction from GRACE and GRACE-FO data in the UYW. This automated XML framework is applicable globally, contingent upon sufficient high-quality data for model training and validation.

As a key component of the cryosphere, permafrost is sensitive to climate change, but mapping permafrost, especially in the Tibetan Plateau, has been challenging due to the heterogeneous mountainous landscape and limited representativeness of ground observations. Using 155 compiled ground observations and more than 20,000 rock glacier records, we developed a machine learning model to map the distribution of permafrost and produce an improved permafrost zonation index (PZI) map. The model was applied by incorporating several control variables, including terrain (elevation and relief), soil (bulk density, clay, coarse fragments, sand, and silt), and temperature (MAAT, FDD, and TDDT) to estimate the PZI at a 1-km resolution in the southern Tibetan Plateau. Excluding glaciers and lakes, the area of permafrost estimated by the new map is approximately 103.5 x 103 km2, accounting for 47.8% of the total area of the region. The result was assessed with various datasets and compared with existing permafrost maps and achieved higher accuracy compared with previous studies. The overall classification accuracy was 96.1% in high plain areas and 84.4% in mountain areas. The results demonstrated the substantial potential for improving mapping permafrost and understanding the periglacial environment with rock glacier inventories and machine learning, especially in complex terrain and climate.

Glacial responses to climate change exhibit considerable heterogeneity. Although global glaciers are generally thinning and retreat, glaciers in the Karakoram region are distinct in their surging or advancing, exhibiting nearly zero or positive mass balance-a phenomenon known as the Karakoram Anomaly. This anomaly has sparked significant scientific interest, prompting extensive research into glacier anomalies. However, the dynamics of the Karakoram anomaly, particularly its evolution and persistence, remain insufficiently explored. In this study, we employed Landsat reflectance data and Moderate Resolution Imaging Spectroradiometer (MODIS) MCD43A3 albedo products to developed high-resolution albedo retrieval models using two machine learning (ML) regressions--random forest regression (RFR) and back-propagation neural network regression (BPNNR). The optimal BPNNR model (Pearson correlation coefficient [r] = 0.77-0.97, unbiased root mean squared error [ubRMSE] = 0.056-0.077, RMSE = 0.055-0.168, Bias = -0.149 similar to -0.001) was implemented on the Google Earth Engine cloud-based platform to estimate summer albedo at a 30-m resolution for the Karakoram region from 1990 to 2021. Validation against in-situ albedo measurements on three glaciers (Batura, Mulungutti and Yala Glacier) demonstrated that the model achieved an average ubRMSE of 0.069 (p < 0.001), with RMSE and ubRMSE improvements of 0.027 compared to MODIS albedo products. The high-resolution data was then used to identify firn/snow extents using a 0.37 threshold, facilitating the extraction of long-term firn-line altitudes (FLA) to indicate the glacier dynamics. Our findings revealed that a consistent decline in summer albedo across the Karakoram over the past three decades, signifying a darkening of glacier surfaces that increased solar radiation absorption and intensified melting. The reduction in albedo showed spatial heterogeneity, with slower reductions in the western and central Karakoram (-0.0005-0.0005 yr(-1)) compared to the eastern Karakoram (-0.006 similar to -0.01 yr(-1)). Notably, surge- or advance-type glaciers, avalanche-fed glaciers and debris-covered glaciers exhibited slower albedo reduction rates, which decreased further with increasing glacier size. Additionally, albedo reduction accelerated with altitude, peaking near the equilibrium-line altitude. Fluctuations in the albedo-derived FLAs suggest a transition in the dynamics of Karakoram glaciers from anomalous behavior to retreat. Most glaciers exhibited anomalous behavior from 1995 to 2010, peaking in 2003, but they have shown signs of retreat since the 2010s, marking the end of the Karakoram anomaly. These insights deepen our understanding of the Karakoram anomaly and provide a theoretical basis for assessing the effect of glacier anomaly to retreat dynamics on the water resources and adaptation strategies for the Indus and Tarim Rivers.

Soil freeze-thaw cycles play a critical role in ecosystem, hydrological and biogeochemical processes, and climate. The Tibetan Plateau (TP) has the largest area of frozen soil that undergoes freeze-thaw cycles in the low-mid latitudes. Evidence suggests ongoing changes in seasonal freeze-thaw cycles during the past several decades on the TP. However, the status of diurnal freeze-thaw cycles (DFTC) of shallow soil and their response to climate change largely remain unknown. In this study, using in-situ observations, the latest reanalysis, machine learning, and physics-based modeling, we conducted a comprehensive assessment of the spatiotemporal variations of DFTC and their response to climate change in the upper Brahmaputra (UB) basin. About 24 +/- 8% of the basin is subjected to DFTC with a mean frequency of 87 +/- 55 days during 1980-2018. The area and frequency of DFTC show small long-term changes during 1980-2018. Air temperature impacts on the frequency of DFTC changes center mainly around the freezing point (0 degrees C). The spatial variations in the response of DFTC to air temperature can primarily be explained by three factors: precipitation (30.4%), snow depth (22.6%) and seasonal warming/cooling rates (14.9%). Both rainfall and snow events reduce diurnal fluctuations of soil temperature, subsequently reducing DFTC frequency, primarily by decreasing daytime temperature through evaporation-cooling and albedo-cooling effects, respectively. These results provide an in-depth understanding of diurnal soil freeze-thaw status and its response to climate change. Freeze-thaw transitions of terrestrial landscapes are a common phenomenon in cold regions. The seasonal and diurnal freeze-thaw cycles (DFTC) of shallow soil exhibit substantial differences in response to climate. Understanding of the spatiotemporal patterns of DFTC and their response to climate change remains limited over the Tibetan Plateau (TP), which is characterized by the largest areas of freeze-thaw terrain in the mid- and low-latitudes of the world. We found the frequency and area of DFTC show a slight increase trend in a significantly warming climate in upper Brahmaputra (UB) basin, the largest river basin of the TP. The variation of DFTC depends on climatic conditions, with soils near the freezing point (0 degrees C) being more susceptible to changes in DFTC. Precipitation, snow depth and seasonal warming/cooling rates are the top three factors influencing the response of DFTC to air temperature changes. Snowfall plays a more important role in the temporal variability of DFTC frequency than rainfall. The number of diurnal freeze-thaw cycles (DFTC) in shallow soil increase slightly during the period 1980-2018 in the upper Brahmaputra (UB) basin Air temperature effects on the changes in DFTC frequency center on the freezing point Snowfall plays a more important role in the temporal variability of DFTC than rainfall

Quantifying seasonal deformation is essential for accurately determining the thickness of the active layer and the distribution of water content within it, providing insights into the freeze-thaw dynamics of permafrost environments and their sensitivity to climate change. Due to the limited hydraulic conductivity of the underlying permafrost, the freeze-thaw processes are largely confined to the active layer, allowing for predictable seasonal deformations. This study employed Independent Component Analysis to isolate large-scale seasonal deformation from Interferometric Synthetic Aperture Radar (InSAR) measurements taken from 2016 to 2020 in the Yangtze River Source Region (YRSR) of the Qinghai-Tibet Plateau (QTP), covering 18,500 km2. We developed dedicated machine learning (ML) models that integrate these InSAR-derived measurements with various environmental proxies. By applying these models to the YRSR, we generated a comprehensive, full-coverage deformation map for permafrost terrains, achieving an R2 value of 0.91 and an Root Mean Squared Error of approximately 0.5 cm, thus confirming the model's strong predictability of seasonal deformation in permafrost regions. Deformation magnitude varied from less than 1 cm to over 10 cm. Our analysis suggests that terrain attributes, influenced by climate and soil conditions, are the primary factors driving these deformations. This research provides valuable insights into quantifying permafrost-related seasonal deformation across expansive and rural landscapes. It also aids in assessing subsurface hydrological processes and the resilience and vulnerability of permafrost. The developed ML algorithm, with access to precise environmental data, is capable of forecasting seasonal deformations across the entire QTP and potentially throughout the Arctic. Seasonal ground deformation, including both subsidence and uplift, is common in areas with a layer of ground that freezes and thaws seasonally, underlain by permafrost-a type of ground that remains at or below 0 degrees C for at least 2 years. These deformations are crucial indicators of changes in water content and thickness of this layer, offering insights into the freeze-thaw dynamics of cold environments and their sensitivity to climate change. However, accurately mapping ground deformation over large areas has been challenging. In this study, we developed machine learning (ML) models that use radar remote sensing data, statistical methods, and a set of environmental variables to predict these seasonal ground movements. Our models can accurately forecast seasonal deformation using readily available environmental data. We find that slope of the terrain is the main factor influencing seasonal deformation, with climate and soil conditions also playing significant roles. This research offers new ways to measure and understand ground deformation in remote permafrost regions and demonstrates how ML can be used to predict such deformations on a continental or even global scale large. Our findings provide valuable insights for environmental scientists and could help inform strategies for managing these regions under changing climatic conditions. Our results underscore the predictability of seasonal deformation with high accuracy in permafrost terrains Machine learning models predict full-coverage seasonal deformation with high accuracy (R2 = 0.91, Root Mean Squared Error [RMSE] = 0.5 cm) Seasonal deformation is primarily determined by terrain slope and regulated by climate and soil conditions

Soil organic carbon (SOC) plays a vital role in the global carbon cycle and soil quality assessment. The Qinghai-Tibet Plateau is one of the largest plateaus in the world. Therefore, in this region, SOC density and the spatial distribution of SOC are highly sensitive to climate change and human intervention. Given the insufficient understanding of the spatial distribution of SOC density in the Qinghai-Tibet Plateau, this study utilized machine learning (ML) algorithms to estimate the density and distribution pattern of SOC density in the region. In this study, we first collected multisource data, such as optical remote sensing data, synthetic aperture radar) (SAR) data, and other environmental variables, including socioeconomic factors, topographic factors, climate factors, and soil properties. Then, we used ML algorithms, namely random forest (RF), extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM), to estimate the topsoil SOC density and spatial distribution patterns of SOC density. We also aimed to investigate any driving factors. The results are as follows: (1) The average SOC density is 5.30 kg/m(2). (2) Among the three ML algorithms used, LightGBM showed the highest validation accuracy (R-2 = 0.7537, RMSE = 2.4928 kgC/m(2), MAE = 1.7195). (3) The normalized difference vegetation index (NDVI), valley depth (VD), and temperature are crucial in predicting the spatial distribution of topsoil SOC density. Feature importance analyses conducted using the three ML models all showed these factors to be among the top three in importance, with contribution rates of 14.08%, 12.29%, and 14.06%; 17.32%, 20.73%, and 24.62%; and 16.72%, 11.96%, and 20.03%. (4) Spatially, the southeastern part of the Qinghai-Tibet Plateau has the highest topsoil SOC density, with recorded values ranging from 8.41 kg/m(2) to 13.2 kg/m(2), while the northwestern part has the lowest density, with recorded values ranging from 0.85 kg/m(2) to 2.88 kg/m(2). Different land cover types showed varying SOC density values, with forests and grasslands having higher SOC densities compared to urban and bare land areas. The findings of this study provide a scientific basis for future soil resource management and improved carbon sequestration accounting in the Qinghai-Tibet Plateau.

Extensive, detailed information on the spatial distribution of active layer thickness (ALT) in northern Alaska and how it evolves over time could greatly aid efforts to assess the effects of climate change on the region and also help to quantify greenhouse gas emissions generated due to permafrost thaw. For this reason, we have been developing high-resolution maps of ALT throughout northern Alaska. The maps are produced by upscaling from high-resolution swaths of estimated ALT retrieved from airborne P-band synthetic aperture radar (SAR) images collected for three different years. The upscaling was accomplished by using hundreds of thousands of randomly selected samples from the SAR-derived swaths of ALT to train a machine learning regression algorithm supported by numerous spatial data layers. In order to validate the maps, thousands of randomly selected samples of SAR-derived ALT were excluded from the training in order to serve as validation pixels; error performance calculations relative to these samples yielded root-mean-square errors (RMSEs) of 7.5-9.1 cm, with bias errors of magnitude under 0.1 cm. The maps were also compared to ALT measurements collected at a number of in situ test sites; error performance relative to the site measurements yielded RMSEs of approximately 11-12 cm and bias of 2.7-6.5 cm. These data are being used to investigate regional patterns and underlying physical controls affecting permafrost degradation in the tundra biome.

Permafrost degradation on the Tibetan Plateau (TP) will significantly affect local water cycle processes, downstream water ecology, and water security. In this study, we evaluate the long-term interannual dynamics of permafrost distribution and active layer thickness (ALT) on the TP based on historical data from Climatic Research Unit gridded Time Series (CRU TS) downscaling and projected data under four shared socio-economic pathways (SSPs) in Scenario Model Intercomparison Project (ScenarioMIP) of the Coupled Model Intercomparison Project Phase 6 (CMIP 6). To achieve this, we employ a data-driven scheme at 1 km resolution for both historical and future periods (1901-2100) that compares the performance of four machine learning algorithms to select the optimal algorithm for permafrost distribution and ALT simulations. Our results indicate that the permafrost on the TP has been undergoing degradation in both historical and future periods, with a decrease in permafrost area and an increase in ALT. The changing rates of permafrost area and regionally averaged ALT during the historical period (1901-2020) are -1.05 x 104 km2 decade-1 and 0.012 m decade-1, while an accelerated degradation is observed after the 1970 s (with changing rates of permafrost area and regionally average ALT of -3.62 x 104 km2 decade-1 and 0.055 m decade-1). Our results also suggested that permafrost degradation on the TP will continue in the future under the four SSP scenarios. The individual global climate models (GCMs) exhibit a consistent degradation trend but great uncertainty in degradation speed. The ensemble mean of simulations across 15 selected GCMs showed that the degradation percentage of permafrost area on the TP under scenarios SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 was 26.0 +/- 6.8 %, 50.4 +/- 5.6 %, 79.2 +/- 4.5 %, and 89.0 +/- 4.0 % by 2100, and the regionally average ALT increased by 0.301 +/- 0.112 m, 0.628 +/- 0.113 m, 1.204 +/- 0.119 m, and 1.486 +/- 0.125 m, respectively. We also analyze permafrost stability and elevationdependent changes of ALT on the TP. The permafrost stability increases with elevation and latitude, and ALT changes more intensely with increasing elevation. This study will provide valuable data for hydrological and ecological studies related to permafrost on the TP.

The aerosol particles present in the atmospheric region mainly affect the climate radiative forcing directly by scattering & absorbing the sunlight. Also, it indirectly influences the formation of clouds, precipitation and acts as a considerable uncertainty in assessing Earth's radiation budget. Determination of aerosol type is significant in characterizing the aerosol role in the atmospheric processes, feedback, and climate models. This paper proposes two aerosol classification models, one based on the source and another based on the composition, to classify the aerosols using aerosol optical properties. The source based aerosol classification method helps to identify the sources which cause pollution in a particular region. Based on the results, proper control measures can be taken to reduce pollution. The composition based aerosol classification helps to identify the nature of aerosol types, such as absorbing or non-absorbing. This classification helps to study the climate of the Kanpur region. The aerosol data is taken from AERONET (AErosol RObotic NETwork) for the period 2002-2018 for the Kanpur region. The composition based aerosol classification model uses Single Scattering Albedo (SSA), Angstrom Exponent (AE), and Fine Mode Fraction (FMF) parameters to categorize aerosols based on their composition. The source based aerosol classification model classifies the aerosols based on values of AE and Aerosol Optical Depth (AOD) and describes the source of the aerosol particles. Knowledge of aerosol sources and compositions helps execute policies or controls to reduce aerosol concentrations. Machine learning algorithms, Nai center dot ve Bayes, K Nearest Neighbor, Decision Tree, Support Vector Machine, and Random Forest are used to validate classification schemes. The performance analysis of machine learning algorithms is compared using ten different metrics, and the results are also compared with the existing aerosol classification models. The results of the classification show that the source based aerosols of the desert and arid background and the composition based aerosols of types, Mixture Absorbing, Coarse absorbing (Dust), and Black Carbon are dominant over the Kanpur region during the study period considered. The Number of non -absorbing (scattering) type aerosols are least in the study region considered during the study period at all the seasons. It is found that the Random Forest and Decision Tree models outperform the other machine learning models considered and the existing classification models in terms of accuracy (99.55 %) and other performance metrics considered.(c) 2023 COSPAR. Published by Elsevier B.V. All rights reserved.

Large stocks of soil organic carbon (SOC) accumulated in the Northern Hemisphere permafrost regions may be vulnerable to climatic warming, but global estimates of SOC distribution and magnitude in permafrost regions still have large uncertainties. Based on multiple high-resolution environmental variables and a compiled soil sample dataset (>3000 soil profiles), we used machine-learning methods to estimate the size and spatial distribution of SOC for the top 3 m soils in the Northern Hemisphere permafrost regions. We also identified key environmental predictors of SOC. The results showed that the SOC storage for the top 3 m soil was 1079 +/- 174 Pg C across the Northern Hemisphere permafrost regions (20.8 x 10(6) km(2)), including 1057 +/- 167 Pg C in the northern permafrost regions and 22 +/- 7 Pg C in the Third Pole permafrost regions. The mean annual air temperature and NDVI are the main controlling factors for the spatial distribution of SOC stocks in the northern and the Third Pole permafrost regions. Our estimations were more accurate than the existing global SOC stock maps. The results improve our understanding of the regional and global permafrost carbon cycle and their feedback to the climate system.