Conference Agenda

This project leverages the ESA and NRSCC opportunity to access satellite observations of Earth’s surface to assess precipitation and temperature biases in climate reanalyses. We are applying the land surface models Tethys & Chloris (T&C), validated by independent observations, to deepen our understanding of the cryosphere and water cycle of key water towers in High Mountain Asia (HMA).

Our main aim is to use Earth Observation data to constrain glacio- and eco-hydrological processes, in order to quantify the interplay of blue, green, and white water fluxes in glacierized catchments across High Mountain Asia. The focus of remote sensing data analysis is the observation and understanding of the relation between climate forcing through the surface energy balance of snow and ice and the cryosphere dynamics in terms of both mass and dynamics of snowpack and glaciers.

Extended time series of glacier observations and integration with modelling have been created. Specifically, the following topics were addressed: a) Improvement to catchment and glacier snowline retrieval algorithm and testing at several sites across HMA; b) Advanced data compilation of stereo satellite data across most sites spanning 1970s-present, and development of methods to derive multidecadal mass balance, applied at the Sangvor site; c) Method development for glacier meteorology inference and evaluation at individual and subregional glaciers; d) Application of Tethys-Chloris to understand historic snow, glacier, and hydrological changes at the Kyzylsu site in the Pamir region.

In parallel, multisource satellite data have been used to improve data products and models to describe hydro-meteorological processes in high elevation catchments: a) vapour flux components and drivers; b) near-real-time retrieval of precipitation rate; c) improved retrieval of shortwave solar irradiance; d) quality of observations for glacier flow retrievals; e) lake ice evolution monitoring.

Catchment and glacier snowline datasets have proven effective to inform climatic information and to constrain bias corrections for meteorological forcing. The acquisitions from Sentinel-2 and the Landsat family have provided high-resolution retrievals at increasing frequency and accuracy, In parallel, we have compiled multitemporal stereo satellite data for 8 sites in the Pamir region, spanning the 1970’s to present. The sites have each been imaged by 7-20 high-quality stereo acquisitions from KH-9 HEXAGON, CORONA, SPOT5, ALOS, SPOT6/7, and Pleaides satellites, providing a rich historical archive of glacier changes for our derivation of multitemporal mass balance in the region of the Karakoram Anomaly. A Bayesian inversion framework has been developed to infer air temperature and precipitation and to reconstruct seasonal glacier mass balance by combining physical modeling with remote sensing data, allowing us to extend the physical modelling approach to domains with few or no direct measurements. We have applied the T&C land-surface model at three study sites (Parlung No.4, Kyzylsu, and Trambau-Trakarding), where the availability of ground observations and knowledge from past studies helped us with the model setup and evaluation. These efforts pave the way for more extensive model deployments to infer the balance of white, blue, and green water fluxes at hyper resolution.

We identified the dominant drivers of vapour flux (ET) trends and their causal pathways using partial correlation analysis and partial least squares structural equation modeling (PLS-SEM) method based on a high-resolution ET dataset at 1-km resolution from 2000 to 2023. We have developed and demonstrated a novel Dynamic Two-Step Real-Time Precipitation Retrieval Algorithm (D-PRA) based on geostationary satellite observations. Results show that compared to the global precipitation product GSMaP_NOW, D-PRA achieves a nearly fivefold increase in Probability of Detection (POD) and a 25.5% reduction in Root Mean Square Error (RMSE). We have developed and applied a terrain-aware orthorectification method to correct the geolocation displacement in the Himawari-8 Level-2 SWR product. Compared to the terrain-corrected MODIS MCD18A1 product (used as a spatially reliable reference), the correction increased the spatial correlation coefficient by 0.13 and reduced the root-mean-square difference by 30.8 W m⁻² over the Tibetan Plateau. Glacier surface flow velocity has been derived from Sentinel-2 Multispectral Instrument (MSI) imagery, utilizing the Co-registration of Optically Sensed Images and Correlation (COSI-Corr) method. The quality of the time-series dataset of glacier flow velocity has been significantly improved through optimizing sampling time and image coherence. Using MODIS observations, we developed a global, high-spatiotemporal-resolution lake ice dataset for 32,800 lakes spanning 2002–2024, providing daily ice coverage, annual ice-cover status, key phenology metrics, and the probability of complete ice cover.

This Dragon 6 project aims to advance understanding of cryospheric change under global warming by integrating multi-source satellite observations with process-based modelling across the Greenland Ice Sheet and the Pan Third Pole region. By linking observation-driven analysis with physical modelling, the resulting framework improves quantitative understanding and enhances predictive capability of ice dynamics and permafrost deformation.

Methodologically, advanced techniques based on Sentinel-1 synthetic aperture radar (SAR) and multi-sensor datasets are developed for key cryospheric processes. For glacier surface melting detection, dual-polarized SAR backscatter and the polarimetric decomposition features are combined with ensemble learning methods to detect the melt conditions, achieving an overall accuracy of ~80%. The fusion of Sentinel-1 SAR and Sentinel-2 optical velocity products significantly improves summer velocity coverage (by 18–57%) and reduces underestimation of near-terminus mass discharge by up to 20%. In addition, a novel multiple aperture coherence (MAC) approach is proposed for mapping landfast sea ice by exploiting sub-aperture phase stability, enabling robust discrimination between stable landfast ice and dynamic sea ice under challenging conditions such as strong winds and rough sea states, while overcoming limitations of repeat-pass InSAR and/or only backscatter coefficient.

Novel methods for retrieving key glacier physical parameters are developed. To address the difficulty of traditional ground-penetrating radar techniques in detecting the ice-bed interface of water-rich glacier, a stochastic dispersive medium model based on finite-difference time-domain (FDTD) simulations is constructed, together with a particle swarm optimization (PSO)-based corrected trimmed mean CFAR (CTM-CFAR) detector. For Bayi Glacier and 23k glacier on the Tibetan Plateau, the PSO-CTM-CFAR method reduced the ice thickness measurement error by approximately 30% compared with traditional methods. A novel method combining remote sensing albedo and differencing Digital Elevation Models (DEM) to estimate glacier annual mass balance is also developed, achieving a high accuracy (RMSE ~495 mm w.e., an ~11% improvement over existing methods) and demonstrating strong potential for large-scale applications.

For glacier systems, the project investigates two complementary domains: the Greenland Ice Sheet and mountain glaciers in the Pan Third Pole region. For Greenland, a high-resolution and internally consistent dataset of ice front positions (Greenland Terminus Position Dataset, GrTPD) is developed, comprising 19,171 delineations for 465 glaciers from 2002 to 2021 with seasonal coverage. The dataset shows high geometric fidelity (mean average minimum distance of 86 m relative to alternative products like TermPicks) and provides a benchmark for algorithm validation and large-scale analysis of terminus variability. Combined with multi-source velocity and geometric observations, it enables improved characterization of outlet glacier dynamics and associated mass transport. For mountain glaciers in the Pan Third Pole region, the project investigates glacier–proglacial lake interactions and their influence on glacier dynamics. Results show that, compared with land-terminating glaciers, lake-terminating glaciers exhibit, on average, ~32% higher surface velocities and ~41% greater thinning rates. These effects display pronounced spatial heterogeneity and temporal evolution, particularly in the eastern Himalayas, central Himalayas, and the Nyainqentanglha Mountains, highlighting the important role of proglacial lake processes in mountain glacier retreat.

For permafrost systems, the project independently investigates deformation processes across the Qinghai–Tibet Plateau. Based on Sentinel-1 SAR data and time-series InSAR analysis, long-term subsidence and seasonal deformation patterns are characterized at regional scale. Results show that the mean long-term deformation rate in permafrost regions is −2.46 mm yr⁻¹, significantly higher than in seasonally frozen ground (−0.54 mm yr⁻¹). Spatial variability is jointly controlled by permafrost thermal stability, solar radiation, and precipitation. Areas with pronounced seasonal deformation are mainly concentrated in low-relief regions and around lakes and rivers, reflecting the important role of active-layer hydrology.

Building on the observation-based datasets, high-temporal-resolution constraints are incorporated into physical modelling. Ice-flow simulations using the Ice-sheet and Sea-level System Model (ISSM), constrained by sub-monthly terminus positions, show that frontal retreat explains more than 76% of velocity variability in Greenland’s fastest glacier, Jakobshavn Isbræ. Remaining model–observation misfit extending tens of kilometres inland is strongly linked to variations in height above flotation and associated effective pressure changes. By incorporating these processes through parameterized basal shear stress adjustments, velocity misfits are reduced by over 90%, highlighting the importance of coupling terminus dynamics with basal processes.

Young scientist contributions include the development of SAR-based algorithms, multi-source data fusion, retrieval of key glacier parameters, and integration of observation-derived constraints into process-based modelling. By advancing Earth observation and its integration with physical modelling, the project aims to facilitate a transition from observation-driven analysis to process-based understanding, providing new insights into cryosphere dynamics and supporting improved projections of climate-driven changes in polar and high-altitude environments.