No abstract

No abstract

The usage of radiosonde humidity data in climate studies and atmospheric reanalysis has been greatly hampered by the inhomogeneity issue mainly caused by radiosonde sensor changes. In this work, high‐quality precipitable water (PW) products derived from a national Global Positioning System (GPS) network in China covering the period from 1999 to 2015 were used to quantify errors in radiosonde‐derived PW products for different radiosonde types. Correlations between PW biases and factors including the station location, mean PW, weighted mean temperature (Tm), and solar elevation angle were carefully analyzed. Biases in PW products derived from the mechanical radiosondes (GZZ2, used at most stations before 2001) show strong correlations with the station elevation and Tm, and a PW correction model was then developed. For GTS1 and GTS1‐1 (widely used in current operational system), due to the unobvious correlations between PW biases and these factors, corrections were estimated as mean values of PW biases at all collocated stations. For GTS1‐2, GTS2(U)‐1, and TD2‐A (used at a few stations), subject to insufficient GPS‐radiosonde collocated stations, they are left uncorrected. The corrected radiosonde‐derived PW products (referred as Corr) were compared with the uncorrected products (referred as Raw) as well as PW products derived from the radiosonde data homogenized using the method proposed by Dai et al. (2011; referred as Dai, https://doi.org/10.1175/2010JCLI3816.1). Corr products show better agreements with GPS‐derived PW than Raw and Dai, and artificial significant decreasing PW trends in the recent two decades in Raw products were greatly reduced after applying the proposed corrections.

No abstract

Few-Shot Relation Extraction (FSRE), a subtask of Relation Extraction (RE) that utilizes limited training instances, appeals to more researchers in Natural Language Processing (NLP) due to its capability to extract textual information in extremely low-resource scenarios. The primary methodologies employed for FSRE have been fine-tuning or prompt tuning techniques based on Pre-trained Language Models (PLMs). Recently, the emergence of Large Language Models (LLMs) has prompted numerous researchers to explore FSRE through In-Context Learning (ICL). However, there are substantial limitations associated with methods based on either traditional RE models or LLMs. Traditional RE models are hampered by a lack of necessary prior knowledge, while LLMs fall short in their task-specific capabilities for RE. To address these shortcomings, we propose a Dual-System Augmented Relation Extractor (DSARE), which synergistically combines traditional RE models with LLMs. Specifically, DSARE innovatively injects the prior knowledge of LLMs into traditional RE models, and conversely enhances LLMs' task-specific aptitude for RE through relation extraction augmentation. Moreover, an Integrated Prediction module is employed to jointly consider these two respective predictions and derive the final results. Extensive experiments demonstrate the efficacy of our proposed method.