Remote Sensing of Environmental Changes in Cold Regions: Methods, Achievements and Challenges

Multi-source machine learning and spaceborne remote sensing data accurately predict three-dimensional soil moisture in an in-service uranium disposal cell

One reason arid and semi-arid environments have been used to store waste is due to low groundwater recharge, presumably limiting the potential for meteoric water to mobilize and transport contaminants into groundwater. The U.S. Department of Energy Office of Legacy Management (LM) is evaluating selected uranium mill tailings disposal cell covers to be managed as evapotranspiration (ET) covers, where vegetation is used to naturally remove water from the cover profile via transpiration, further reducing deep percolation. An important parameter in monitoring the performance of ET covers is soil moisture (SM). If SM is too high, water may drain into tailings material, potentially transporting contaminants into groundwater; if SM is too low, radon flux may increase through the cover. However, monitoring SM via traditional instrumentation is invasive, expensive, and may fail to account for spatial heterogeneity, especially over vegetated disposal cells. Here we investigated the potential for non-invasive SM monitoring using radar remote sensing and other geospatial data to see if this approach could provide a practical, accurate, and spatially comprehensive tool to monitor SM. We used theoretical simulations to analyze the sensitivity of multi-frequency radar backscatter to SM at different depths of a field-scale (3 ha) drainage lysimeter embedded within an in-service LM disposal cell. We then evaluated a shallow and deep form of machine learning (ML) using Google Earth Engine to integrate multi-source observations and estimate the SM profile across six soil layers from depths of 0-2 m. The ML models were trained using in situ SM measurements from 2019 and validated using data from 2014 to 2018 and 2020-2021. Model predictors included backscatter observations from satellite synthetic aperture radar, vegetation, temperature products from optical infrared sensors, and accumulated, gridded rainfall data. The radar simulations confirmed that the lower frequencies (L- and P-band) and smaller incidence angles show better sensitivity to deeper soil layers and an overall larger SM dynamic range relative to the higher frequencies (C- and X-band). The ML models produced accurate SM estimates throughout the soil profile (r values from 0.75 to 0.94; RMSE = 0.003-0.017 cm3/cm3; bias = 0.00 cm3/cm3), with the simpler shallow-learning approach outperforming a selected deep-learning model. The ML models we developed provide an accurate, cost-effective tool for monitoring SM within ET covers that could be applied to other vegetated disposal cell covers, potentially including those with rock-armored covers.

Deep learning estimation of northern hemisphere soil freeze-thaw dynamics using satellite multi-frequency microwave brightness temperature observations

Satellite microwave sensors are well suited for monitoring landscape freeze-thaw (FT) transitions owing to the strong brightness temperature (TB) or backscatter response to changes in liquid water abundance between predominantly frozen and thawed conditions. The FT retrieval is also a sensitive climate indicator with strong biophysical importance. However, retrieval algorithms can have difficulty distinguishing the FT status of soils from that of overlying features such as snow and vegetation, while variable land conditions can also degrade performance. Here, we applied a deep learning model using a multilayer convolutional neural network driven by AMSR2 and SMAP TB records, and trained on surface (~0-5 cm depth) soil temperature FT observations. Soil FT states were classified for the local morning (6 a.m.) and evening (6 p.m.) conditions corresponding to SMAP descending and ascending orbital overpasses, mapped to a 9 km polar grid spanning a five-year (2016-2020) record and Northern Hemisphere domain. Continuous variable estimates of the probability of frozen or thawed conditions were derived using a model cost function optimized against FT observational training data. Model results derived using combined multi-frequency (1.4, 18.7, 36.5 GHz) TBs produced the highest soil FT accuracy over other models derived using only single sensor or single frequency TB inputs. Moreover, SMAP L-band (1.4 GHz) TBs provided enhanced soil FT information and performance gain over model results derived using only AMSR2 TB inputs. The resulting soil FT classification showed favorable and consistent performance against soil FT observations from ERA5 reanalysis (mean percent accuracy, MPA: 92.7%) and in situ weather stations (MPA: 91.0%). The soil FT accuracy was generally consistent between morning and afternoon predictions and across different land covers and seasons. The model also showed better FT accuracy than ERA5 against regional weather station measurements (91.0% vs. 86.1% MPA). However, model confidence was lower in complex terrain where FT spatial heterogeneity was likely beneath the effective model grain size. Our results provide a high level of precision in mapping soil FT dynamics to improve understanding of complex seasonal transitions and their influence on ecological processes and climate feedbacks, with the potential to inform Earth system model predictions.

Characteristics of the Northern Hemisphere cold regions changes from 1901 to 2019

The accurate delineation of the spatial extent of cold regions provides the basis for the study of global environmental change. However, attention has been lacking on the temperature-sensitive spatial changes in the cold regions of the Earth in the context of climate warming. In this study, the mean temperature in the coldest month lower than - 3 °C, no more than 5 months over 10 °C, and an annual mean temperature no higher than 5 °C were selected to define cold regions. Based on the Climate Research Unit land surface air temperature (CRUTEM) of monthly mean surface climate elements, the spatiotemporal distribution and variation characteristics of the Northern Hemisphere (NH) continental cold regions from 1901 to 2019 are analyzed in this study, by adopting time trend and correlation analyses. The results show: (1) In the past 119 years, the cold regions of the NH covered on average about 4.074 × 10⁷ km², accounting for 37.82% of the total land area of the NH. The cold regions can be divided into the Mid-to-High latitude cold regions and the Qinghai-Tibetan Plateau cold regions, with spatial extents of 3.755 × 10⁷ km² and 3.127 × 10⁶ km², respectively. The Mid-to-High latitude cold regions in the NH are mainly distributed in northern North America, most of Iceland, the Alps, northern Eurasia, and the Great Caucasus with a mean southern boundary of 49.48° N. Except for the southwest, the entire region of the Qinghai-Tibetan Plateau, northern Pakistan, and most of Kyrgyzstan are cold regions. (2) In the past 119 years, the rates of change in the spatial extent of the cold regions in the NH, the Mid-to-High latitude, and the Qinghai-Tibetan Plateau were - 0.030 × 10⁷ km²/10 a, - 0.028 × 10⁷ km²/10 a, and - 0.013 × 10⁶ km²/10 a, respectively, showing an extremely significant decreasing trend. In the past 119 years, the mean southern boundary of the Mid-to-High latitude cold regions has been retreating northward at all longitudes. For instance, the mean southern boundary of the Eurasian cold regions moved 1.82° to the north and that of North America moved 0.98° to the north. The main contribution of the study lies in the accurate definition of the cold regions and documentation of the spatial variation of the cold regions in the NH, revealing the response trends of the cold regions to climate warming, and deepening the study of global change from a new perspective.

Classification of Alpine Grasslands in Cold and High Altitudes Based on Multispectral Landsat-8 Images: A Case Study in Sanjiangyuan National Park, China

Land-use–cover change (LUCC)/vegetation cover plays a critical role in Earth system science and is a reflection of human activities and environmental changes. LUCC will affect the structure and function of ecosystems and a series of other terrestrial surface processes, such as energy exchange, water circulation, biogeochemical circulation, and vegetation productivity. Therefore, accurate LUCC mapping and vegetation cover monitoring are the bases for simulating the global carbon and hydrological cycles, studying the interactions of the land surface and climate, and assessing land degradation. Based on field GPS surveys and UAV data, with cloud-free and snow/glacier algorithms and the SVM classifier to train and model alpine grassland, the alpine grassland and LUCC were extracted by using Landsat-8 OLI satellite images in Sanjiangyuan National Park in this paper. The latest datasets of vegetation types with 30 m × 30 m spatial resolution in the three parks were prepared and formed. The classification results show that the SVM classifier could better distinguish the major land-use types, and the overall classification accuracy was very high. However, in the alpine grassland subcategories, the classification accuracies of the four typical grasslands were relatively low, especially between desert steppes and alpine meadows, and desert steppes and alpine steppes. It manifests the limitations of Landsat-8 multispectral remote sensing imageries in finer-resolution grassland classifications of high-altitude alpine mountains. The method can be utilized for other multispectral satellite imageries with the same band matching, such as Landsat 7, Landsat 9, Sentinel-2, etc. The method described in this paper can rapidly and efficiently process annual alpine grassland maps of the source areas of the Yellow River, the Yangtze River, and the Lancang River. It can provide timely and high-spatial-resolution datasets for supporting scientific decisions for the sustainable management of Sanjiangyuan National Park.

Chemiresistive NH3 detection at sub-zero temperatures by polypyrrole- loaded Sn1-xSbxO2 nanocubes

Chemiresistive gas sensors operate mainly at high temperatures, primarily due to the need of energy for surface adsorption-desorption of analytes. As a result, the operating temperature of the chemiresistive sensors could be reduced only to room temperature. Hence, a plethora of sensing requirements at temperatures below ambient have remained outside the scope of chemiresistive materials. In this work, we have developed an antimony-doped SnO₂ nanocube-supported expanded polypyrrole network that could detect low ppm ammonia gas (≤20 ppm) at sub-zero temperatures with high response (∼4), selectivity, and short response and recovery times. The low temperature chemiresistive sensing has been explained in terms of the interplay of an extended conducting network of an in situ deposited polymer, effective transport properties of majority charge carriers and a loosely bound exciton-like electron-hole pair formation and breakage mechanism.

Applying Machine Learning and Time-Series Analysis on Sentinel-1A SAR/InSAR for Characterizing Arctic Tundra Hydro-Ecological Conditions

Synthetic aperture radar (SAR) is a widely used tool for Earth observation activities. It is particularly effective during times of persistent cloud cover, low light conditions, or where in situ measurements are challenging. The intensity measured by a polarimetric SAR has proven effective for characterizing Arctic tundra landscapes due to the unique backscattering signatures associated with different cover types. However, recently, there has been increased interest in exploiting novel interferometric SAR (InSAR) techniques that rely on both the amplitude and absolute phase of a pair of acquisitions to produce coherence measurements, although the simultaneous use of both intensity and interferometric coherence in Arctic tundra image classification has not been widely tested. In this study, a time series of dual-polarimetric (VV, VH) Sentinel-1 SAR/InSAR data collected over one growing season, in addition to a digital elevation model (DEM), was used to characterize an Arctic tundra study site spanning a hydrologically dynamic coastal delta, open tundra, and high topographic relief from mountainous terrain. SAR intensity and coherence patterns based on repeat-pass interferometry were analyzed in terms of ecological structure (i.e., graminoid, or woody) and hydrology (i.e., wet, or dry) using machine learning methods. Six hydro-ecological cover types were delineated using time-series statistical descriptors (i.e., mean, standard deviation, etc.) as model inputs. Model evaluations indicated SAR intensity to have better predictive power than coherence, especially for wet landcover classes due to temporal decorrelation. However, accuracies improved when both intensity and coherence were used, highlighting the complementarity of these two measures. Combining time-series SAR/InSAR data with terrain derivatives resulted in the highest per-class F1 score values, ranging from 0.682 to 0.955. The developed methodology is independent of atmospheric conditions (i.e., cloud cover or sunlight) as it does not rely on optical information, and thus can be regularly updated over forthcoming seasons or annually to support ecosystem monitoring.

The Influence Mechanism of Freeze-Thaw on Soil Erosion: A Review

As an important type of soil erosion, freeze-thaw erosion occurs primarily at high latitude and altitude. The overview on the effect of freeze-thaw on soil erosion was provided. Soil erosion was affected by freeze-thaw processes, as thawing and water erosion reinforce each other. Remote sensing provided an unprecedented approach for characterizing the timing, magnitude, and patterns of large-scale freeze-thaw and soil erosion changes. Furthermore, the essence of soil freeze-thaw was the freeze and thaw of soil moisture in the pores of soil. Freeze-thaw action mainly increased soil erodibility and made it more vulnerable to erosion by destroying soil structure, changing soil water content, bulk density, shear strength and aggregate stability, etc. However, the type and magnitude of changes of soil properties have been related to soil texture, water content, experimental conditions and the degree of exposure to freeze-thaw. The use of indoor and field experiments to further reveal the effect of freeze-thaw on soil erosion would facilitate improved forecasting, as well as prevention of soil erosion during thawing in regions with freeze-thaw cycles.

Using remote sensing to assess peatland resilience by estimating soil surface moisture and drought recovery

Peatland areas provide a range of ecosystem services, including biodiversity, carbon storage, clean water, and flood mitigation, but many areas of peatland in the UK have been degraded through human land use including drainage. Here, we explore whether remote sensing can be used to monitor peatland resilience to drought. We take resilience to mean the rate at which a system recovers from perturbation; here measured literally as a recovery timescale of a soil surface moisture proxy from drought lowering. Our objectives were (1) to assess the reliability of Sentinel-1 Synthetic Aperture Radar (SAR) backscatter as a proxy for water table depth (WTD); (2) to develop a method using SAR to estimate below-ground (hydrological) resilience of peatlands; and (3) to apply the developed method to different sites and consider the links between resilience and land management. Our inferences of WTD from Sentinel-1 SAR data gave results with an average Pearson's correlation of 0.77 when compared to measured WTD values. The 2018 summer drought was used to assess resilience across three different UK peatland areas (Dartmoor, the Peak District, and the Flow Country) by considering the timescale of the soil moisture proxy recovery. Results show clear areas of lower resilience within all three study sites, which often correspond to areas of high drainage and may be particularly vulnerable to increasing drought severity/events under climate change. This method is applicable to monitoring peatland resilience elsewhere over larger scales, and could be used to target restoration work towards the most vulnerable areas.

Are There Differences in the Reaction of the Light-Tolerant Subgenus Pinus spp. Biomass to Climate Change as Compared to Light-Intolerant Genus Picea spp.?

Currently, the problem of the impact of climate change on the productivity of forest ecosystems and their carbon-depositing capacity is far from being solved. Therefore, this paper presents the models for the stand biomass of the two-needled subgenus’ (Pinus spp.) and the genus Picea spp.’s trends along the trans-Eurasian hydrothermal gradients, designed for pure stands in a number of 2110- and 870-sample plots with Pinus and Picea correspondingly. It was found that in the case of an increase in mean winter temperatures by 1 °C, pine and spruce respond by increasing the biomass of most components, and in the case of an increase in the annual sum of precipitation by 100 mm, the total, aboveground, stem and root biomasses of pine and spruce react the same way, but crown biomass reacts in the opposite way. Therefore, all identified trends are species-specific.

Recursive Feature Elimination and Random Forest Classification of Natura 2000 Grasslands in Lowland River Valleys of Poland Based on Airborne Hyperspectral and LiDAR Data Fusion

The use of hyperspectral (HS) and LiDAR acquisitions has a great potential to enhance mapping and monitoring practices of endangered grasslands habitats, beyond conventional botanical field surveys. In this study we assess the potentiality of recursive feature elimination (RFE) in combination with random forest (RF) classification in extracting the main HS and LiDAR features needed to map selected Natura 2000 grasslands along Polish lowland river valleys, in particular alluvial meadows 6440, lowland hay meadows 6510, and xeric and calcareous grasslands 6120. We developed an automated RFE-RF system capable to combine the potentials of both techniques and applied it to multiple acquisitions. Several LiDAR-based products and different spectral indices (SI) were computed and used as input in the system, with the aim of shedding light on the best-to-use features. Results showed a remarkable increase in classification accuracy when LiDAR and SI products are added to the HS dataset, strengthening in particular the importance of employing LiDAR in combination with HS. Using only the 24 optimal features selection generalized over the three study areas, strongly linked to the highly heterogeneous characteristics of the habitats and landscapes investigated, it was possible to achieve rather high classification results (K around 0.7–0.77 and habitats F1 accuracy around 0.8–0.85), indicating that the selected Natura 2000 meadows and dry grasslands habitats can be automatically mapped by airborne HS and LiDAR data. Similar approaches might be considered for future monitoring activities in the context of habitats protection and conservation.

Toward the Detection of Permafrost Using Land-Surface Temperature Mapping

Permafrost is degrading under current warming conditions, disrupting infrastructure, releasing carbon from soils, and altering seasonal water availability. Therefore, it is important to quantitatively map the change in the extent and depth of permafrost. We used satellite images of land-surface temperature to recognize and map the zero curtain, i.e., the isothermal period of ground temperature during seasonal freeze and thaw, as a precursor for delineating permafrost boundaries from remotely sensed thermal-infrared data. The phase transition of moisture in the ground allows the zero curtain to occur when near-surface soil moisture thaws or freezes, and also when ice-rich permafrost thaws or freezes. We propose that mapping the zero curtain is a precursor to mapping permafrost at shallow depths. We used ASTER and a MODIS-Aqua daily afternoon land-surface temperature (LST) timeseries to recognize the zero curtain at the 1-km scale as a “proof of concept.” Our regional mapping of the zero curtain over an area around the 7000 m high volcano Ojos del Salado in Chile suggests that the zero curtain can be mapped over arid regions of the world. It also indicates that surface heterogeneity, snow cover, and cloud cover can hinder the effectiveness of our approach. To be of practical use in many areas, it may be helpful to reduce the topographic and compositional heterogeneity in order to increase the LST accuracy. The necessary finer spatial resolution to reduce these problems is provided by ASTER (90 m).

Inference in Supervised Spectral Classifiers for On-Board Hyperspectral Imaging: An Overview

Machine learning techniques are widely used for pixel-wise classification of hyperspectral images. These methods can achieve high accuracy, but most of them are computationally intensive models. This poses a problem for their implementation in low-power and embedded systems intended for on-board processing, in which energy consumption and model size are as important as accuracy. With a focus on embedded and on-board systems (in which only the inference step is performed after an off-line training process), in this paper we provide a comprehensive overview of the inference properties of the most relevant techniques for hyperspectral image classification. For this purpose, we compare the size of the trained models and the operations required during the inference step (which are directly related to the hardware and energy requirements). Our goal is to search for appropriate trade-offs between on-board implementation (such as model size and energy consumption) and classification accuracy.