Conference Agenda

In this paper, continuously altimetric satellite sea surface height measurements from Cryo-TEMPO between 2011 and 2020 are used to illustrate that the NPO plays a significant role in connecting the Beaufort Sea level to the Pacific Ocean. It is found that summertime NPO has a significant negative connection with sea surface heights in the Beaufort Sea. A negative NPO phase tends to be associated to an intensified Beaufort High paired with anomalous anticyclonic circulations over the Arctic, contributing to positive SSH anomalies locally because of increasing more freshwater entering the Beaufort Sea from the Chukchi Sean through Bering Strait. CESM2-LE is used to examine the connection between North Pacific teleconnection and the Beaufort Sea level change for longer time spans. It is suggested that the remarkable relationship between SSH in the Beaufort Sea and NPO is reproduced during 2011–2020, 2000–2020 and 1990–2020. In addition, the pre-winter SST may be a predictor for SSH in the Beaufort Sea. These findings highlight that the impacts of the teleconnection and SST anomalies in North Pacific on the Arctic sea level are of great importance and need to be taken into consideration when evaluating future climate predictions and projections.

The increasing influence of the Atlantic Water (AW) in the Arctic, known as “Atlantification”, has been an important topic of scientific interest for several years. Recent studies reiterated the need to have a better understanding of AW transformation in Nordic Seas (NS) to understand and predict the ocean’s role in ongoing and future Arctic climate change (Asbjørnsen et al., 2020). A “missing puzzle” yet to be studied in detail is the role of mesoscale eddies on the Atlantification. Eddies generated from instabilities of the mean-flow (Stammer and Wunsch, 1999) are ubiquitous features in the NS (e.g., Raj et al., 2016) whereby mean kinetic energy is transformed to eddy kinetic energy with subsequent reduction in the mean northward flow of the AW. Eddies can also capture and trap heat and salt from the mean AW flow (Bolenenko et al., 2020), thereby cooling the AW poleward heat transport (Isachsen et al., 2012). In regards to the Atlantification in the Arctic Ocean, the question is therefore related to the occurrences of eddies in the NS over the last decades; has the number of eddies changed or is it stable?

In this poster we present the results of two distinct analysis of 11 years (2011-2022) of satellite sensed data (interpolated to 25 km spatial resolution at monthly to seasonal timescales) combined with mesoscale eddy tracking to advance the insight of mesoscale eddy activity and upper ocean circulation in the NS. First, the Multiparameter Eddy Significance Index (MESI) proposed by Roman-Stork et al.(2023) is estimated. The index combines sea level anomaly, sea surface temperature and salinity fields, chlorophyll distribution and eddy kinetic energy for all the eddies in the NS. Second, climatologies of the singularity exponents for the satellite-based sea surface temperature and salinity values are provided. The singularity exponent is expected to reveal mixture of horizontal transport and dispersion processes of the upper ocean circulation with particular focus on the impact of mesoscale eddies.

In this presentation we will highlight the findings and results in relation to: (i) observed changes in the annual number of eddies in the NS from altimetry; (ii) assessment of the number of eddies based on the MESI approach; and (iii) consistency between the climatology of singularity exponents and MESI. The relevance of the results, in turn, will also be discussed in relation to the Atlantification of the Arctic Ocean.

High-resolution soil moisture data are essential for studying the complex interactions between the water, energy, and carbon cycles from local to global scales. For agricultural and hydrological applications, a 1 km global-scale soil moisture product is of great interest to the community. In this study, we propose a new dual-polarization algorithm (DPA) for soil moisture retrieval using C-band synthetic aperture radar (SAR) observations. Based on this algorithm, a Sentinel-1-based global-scale soil moisture dataset with a spatial resolution of 1 km (S1-DPA) was generated. Specifically, using optical data as a proxy of vegetation water content, a semi-empirical forward model from soil moisture to backscattering was constructed and calibrated based on the relationship between Sentinel-1 SAR backscatter and SMAP (Soil Moisture Active and Passive) soil moisture product under different vegetation and soil texture conditions. With the calibrated forward model, soil moisture was estimated using the backscatter coefficients on VV and VH polarizations observed by Sentinel-1 C-band SAR in ascending and descending orbits. The S1-DPA soil moisture data product has the same temporal resolution as Sentinel-1, of 3-6 days for Europe and 6-12 days for other regions. It covers the global land surface and spans the period from 2016 to 2020, utilizing both daily ascending and descending data. The S1-DPA product was validated using ground measurements from the International Soil Moisture Network (ISMN). The results show that the S1-DPA product captures the spatial and temporal characteristics of in-situ soil moisture reasonably, with an overall median Pearson correlation of 0.372, bias of -0.003 m³/m³, RMSD (root mean squared difference with respect to in-situ measurements) of 0.105 m³/m³, and ubRMSD (unbiased root mean squared difference) of 0.076 m³/m³. The generated global 1 km soil moisture product has the potential to promote the application of high-resolution soil moisture data in the fields of hydrology, ecology, and meteorology.

The Tibetan Plateau is a climate change sensitive and ecologically fragile area. Global climate change is prone to have an higher impact on the region's local climate than other regions, and therefore have a potential that impose a significant influence on local ecological environment. Therefore, this study aims at evaluating the climate change in the Tibetan Plateau and its impact on vegetation in the plateau in the past up to the end of 21th century in the future. First of all, this study uses CN05.1 meteorological data to first conduct trend analysis, mutation analysis, and periodic analysis on precipitation and temperature in the Tibetan Plateau region over the past 40 years (1979-2017). Then, combined with 11 GCM model data and CN05.1 data from the CMIP6, the ability of a single climate model, a full model set (MME) and a better model set (BMME) to simulate precipitation and temperature in the Tibetan Plateau was evaluated using Taylor chart, interannual variability assessment index and rank scoring method (RS method). A set of optimal models with good simulation capabilities of precipitation and temperature under three future climate scenarios (SSP126, SSP245, and SSP585) was selected, and the Delta method was used for bias correction. After that, by utilizing CN05.1 data, GIMMS NDVI data, and 1:1000000 vegetation distribution map data in China, methods such as linear regression analysis, Sen's slope, Hurst index, partial correlation coefficient, and residual analysis were applied to explore the dynamic changes of existing vegetation in the Tibetan Plateau and its response to climate factors. Finally, based on the corrected CMIP6 climate model data, the CSCS model and land use transfer matrix were used to analyze the potential vegetation distribution and changes in the Tibetan Plateau under three different climate change scenarios in the early 21st century (2021-2040), mid 21st century (2041-2060), and late 21st century (2081-2100).

Precipitation is an important component of water circulation and an essential input for various hydrological models. A high quality, high spatial resolution, and long-term precipitation dataset would benefit hydrological investigations, particularly for regions having insufficient precipitation records. The upper Huaihe River Basin (UHRB) was selected as the research location in this study, and the accuracies of three precipitation products (PPs: a high-resolution daily gridded precipitation dataset for China (HRLT), Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR), and the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center Global (CPC) precipitation dataset) were assessed at multiple spatio-temporal scales comparing with the gauge precipitation (GP) for 2000–2019. Subsequently, the applicability of the three PPs on streamflow (Q) and sediment yield (SY) simulations, as well as the impact on parameter sensitivity, were compared using the Soil and Water Assessment Tool (SWAT) model. The results showed that the accuracy of the three PPs were ranked as CPC > HRLT > PERSIANN-CDR on the watershed average scale, HRLT would underestimate the extreme precipitation; and PERSIANN-CDR would overestimate the annual precipitation. On the grid-to-point scale, PERSIANN-CDR was found to be the most stable with high accuracy, followed by CPC and HRLT on all temporal scales. The ability of these PPs to detect rainfall events was ranked as CPC > HELT > PERSIANN-CDR. The sensitivity of the Q parameters changed with the variation in the precipitation input. The sensitive parameters for GP were distributed on average for almost all processes, while the sensitive parameters for PPs mainly controlled the groundwater and evapotranspiration processes. Among all the PPs, the performance of CPC in the Q and SY simulations was found to be the best, followed by HRLT and PERSIANN-CDR, and all the PPs could simulate SY better than Q in spatial distribution. HRLT has the potential to be used in long-term hydrological simulations in ungauged or small watersheds based on its high spatial resolution compared to other products.

Accurate calculation of evapotranspiration at the basin scale can provide the information for dynamic analysis of the hydrological cycle within the basin. In this study, The Qinghai-Tibet Plateau (TP) , consisting of 12 watersheds, was used as the study area. The process of realization of the medium-scale evapotranspiration calculation by the REOF-3T model can be generalized as follows. Each watershed was divided into several subregions based on the analysis results of the rotated empirical orthogonal function (REOF) method for 10a downward shortwave radiation. The modified 3T model was used to calculate the evaporation in the subregions, thus realizing the distributed calculation of the 3T model. To validate the accuracy of the model, site observations and other remote sensing products were compared to the calculated ET series. The results showed that the REOF-3T model has a significant correlation with the average ET in 8 days of six eddy covariance flux stations over the TP. The Pearson’s correlation coefficient (R) of EC observed sites ranged from 0.6 to 0.78 (P<0.01), the root-mean square error (RMSE) ranged from 1.006 mm/d to 1.408 mm/d. The estimated ET (REOF-3T model) also displayed a good consistency with the observed ET (water evaporation) in 93 meteorological stations during 2008 – 2018. More than 93% of sites have R-values over 0.6. The average annual R in 93 stations exceeded 0.9, except for 2008, 2016, and 2018. There is an increasing trend of ET in the southwestern of TP, especially in the upper Yangtze River basin. While the north and northwest are on a downward trend.

Although global land-atmosphere energy and carbon fluxes is a key driver of Earth’s climate system, global continuous high resolution fluxes datasets are still limited. In this study, we used the STEMMUS-SCOPE simulations at 170 FLUXNET sites as the training dataset, enabling the physics-informed Machine Learning (PIML) to generate a global, long-term, spatially continuous high resolution dataset of fluxes. STEMMUS-SCOPE model is a process-based model simulating water, carbon, and energy fluxes, along with predicting leaf to canopy photosynthesis, reflectance and fluorescence spectra, as well as subsoil moisture and temperature dynamics. Results show that PIML can estimate fluxes with Pearson Correlation Coefficient score (r score) 0.99 for latent heat (LE), and 0.99 for sensible heat (H), and the root mean square error (RMSE score) are 12.89 W/m² and 18.6 W/m² respectively. It can also predict net radiation (Rn) with r score 0.99 and RMSE 7.54 W/m², and root zone soil moisture (RZSM) with r score 0.99 and RMSE 0.0045 cm³/cm³. With solar induced chlorophyll fluorescence (SIF), the r score is 0.99 and RMSE lower than 0.03 W/m²/μm/sr. Incoming shortwave radiation, surface soil moisture, and air temperature are the main predictor variables that determine the prediction performance, followed by incoming longwave radiation and wind speed etc.

Soil moisture is an essential variable in the hydrological cycle and exhibits a strong connection to weather and climate change. The comprehensive understanding of the physical mechanism underlying brightness temperature enables more accurate estimation of soil moisture. The integration of process-based understanding into machine learning models has the potential to leverage the advantages of both methods. This research aims to develop an emulator using machine learning algorithms to conduct a forward simulation of ELBARA-III brightness temperature at L-band. A combination of meteorological data, in-situ soil moisture and soil temperature data and vegetation parameters was used for training. A total of four years’ data, encompassing various combinations, is employed for training purposes, resulting in the construction of 64 models each for horizontal and vertical polarizations. The best-performing model exhibits a correlation coefficient of R = 0.995 for horizontal polarization and R = 0.998 for vertical polarization. Notably, there was a significant enhancement in performance after incorporating the observed data for model training. The primary objective of this research is to investigate the underlying physical mechanisms involved in the emission process and explore the potential of employing machine learning algorithms for simulating microwave signals across extensive spatial and temporal domains. These findings suggest that while random forest regression and support vector regression can capture the general variation trend observed in brightness temperature, some challenges remain. During specific time periods, such as the transition season of October and November, the models' predictions appear smoother and fail to fully capture all signal fluctuations.