Abstract Efficient purification of polymer-grade ethylene (C2H4) from methanol-to-olefins (MTO) products remains challenging because of the similar physicochemical properties of light hydrocarbons. Herein, we report a fluorinated metal–organic framework, CF3-Ni, featuring uniformly distributed trifluoromethyl (−CF3) groups that construct a fluorine-rich polar pore environment for selective hydrocarbon recognition. CF3-Ni exhibits preferential adsorption toward C3 hydrocarbons, with an affinity sequence of C3H6 > C3H8 > C2H4. At 298 K and 1 bar, it shows a C3H6/C2H4 IAST selectivity of 9.3 and a moderate C3H6 adsorption enthalpy of 38.4 kJ mol–1, enabling strong yet reversible host–guest interactions. Breakthrough experiments demonstrate efficient separation of C3H6/C2H4 mixtures over a wide range of feed compositions, affording polymer-grade C2H4 (>99.95%) with a maximum productivity of 7.4 mol kg–1 and simultaneous recovery of high-purity C3H6. Notably, one-step purification of polymer-grade C2H4 is achieved from a ternary C3H6/C3H8/C2H4 mixture, even under humid and elevated-pressure conditions. DFT calculations and in situ FT-IR spectroscopy reveal that C–H···F interactions contribute to the preferential binding of C3H6. This work demonstrates fluorinated pore chemistry as an effective strategy for practical olefin purification.

Accurate global digital elevation models (GDEMs) are essential for various geoscience applications. However, the accuracy of GDEMs in vegetated mountainous regions is relatively low due to substantial topographic relief and the penetration limitations of data acquisition techniques. The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) acquires high-precision and high-density elevation measurements along its ground tracks on a global scale, offering reliable reference data for GDEM corrections. However, previous GDEM corrections using ICESat-2 altimetry data primarily focused on pixel-bypixel vertical elevation corrections, often neglecting interpixel neighboring structure information, which is crucial for terrain modeling and analysis, particularly in areas with high relief. Therefore, this study incorporates not only ICESat-2 precise elevation observations and but also its along-track neighboring structure knowledge into a convolutional neural network-transformer hybrid model, termed NSCT, to correct GDEMs in a global scale. The 1 arc-second Copernicus digital elevation model (DEM) was selected as the target DEM for correction due to its demonstrated superior accuracy. The proposed NSCT model, trained on a diverse range of globally distributed areas with varying topography and vegetation, was evaluated using ICESat-2, global control points, and high-resolution DEMs (HRDEMs). Its performance was compared against eight currently most used DEM correction models and publicly available corrected GDEM products, including FABDEM, FathomDEM, and GEDTM30. Correction results demonstrated that the NSCT model generally improved Copernicus DEM accuracy by 66.34%, and outperformed existing correction models in both vertical elevation and neighboring structure assessment across diverse topographic and vegetation conditions. Furthermore, validation using ICESat-2 data and HRDEMs outside the training area, as well with its application to SRTM and AW3D30 GDEM, demonstrated that the NSCT model exhibited superior transferability capability and consistently outstanding performance. This study is the first to integrate the precise elevation observations of ICESat-2 with along-track neighboring structure knowledge into GDEM corrections, offering valuable insights for future research in terrain modeling and analysis. The relevant code and test data are publicly available at: https://doi.org /10.6084/m9.figshare.29380553

For inland rivers with complex topographies, however, conventional nadir altimeters often fail to meet accuracy requirements and exhibit substantial spatial sampling limitations. Fortunately, the Surface Water and Ocean Topography (SWOT) Satellite, with its wide swath, high spatial resolution, and improved vertical precision, offers new opportunities for the dynamic monitoring of rivers and small water bodies. However, the short duration since SWOT data became available means that existing validations of SWOT over rivers are often limited by insufficient samples, coarse temporal alignment with in-situ data, and a lack of physical explanation for accuracy variations across diverse basins. Here, we used in-situ water levels from 52 monitoring stations located within the Yangtze River Basin and river width data extracted from Sentinel-2 imagery to evaluate the accuracy of the SWOT Water Surface Elevation (WSE) and river width (RW). The results showed that although the SWOT WSE achieved moderate accuracy across all 52 monitoring stations (MAE = 0.42 m), the 28 waterway survey stations located along the Yangtze River main stem exhibited higher accuracy (MAE = 0.36 m) than the 15 monitoring stations on natural tributaries (MAE = 0.50 m) among the 43 natural river-channel monitoring stations. In addition, the nine reservoir stations were reported separately because they were governed by fundamentally different physical error mechanisms. Regrettably, the SWOT RW showed relatively low accuracy with a mean normalized MAE/m of 0.32 (where MAE/m means MAE/RW), with a spatial pattern similar to WSE, with higher accuracy along the gently sloping and wide-channel sections of the Yangtze River main stem where channel gradients are gentler and river widths are larger (MAE/m = 0.21), but its accuracy decreased in the Jinsha River and other tributaries (MAE/m = 0.50). On this basis, we investigated the physical processes through which geomorphological conditions regulated the accuracy of SWOT WSE, focusing on two key aspects: observation geometry (e.g. incidence angle, the angle between the river channel and the SWOT track, and interferometric baseline stability) and phase noise (e.g. RW, sandbars, river channel meandering, steep riverbanks, and river channel slope). The analysis demonstrated that the accuracy of SWOT in complex basins is determined by the degree of alignment between the dynamic observation geometry and static geomorphic characteristics. In summary, the SWOT provides a high-accuracy WSE but a moderate RW dataset and enables the dynamic monitoring of inland rivers and small water bodies globally in near-real-time.

Accurate terrain representation is critical for applications in urban, environmental, and geospatial sciences, but current global digital elevation models (GDEMs) can be regarded as digital surface models exhibiting height biases caused by vegetation and buildings. These biases are particularly pronounced and challenging to remove in vegetated mountainous regions because of multiple bias sources, rugged terrain, limitations of correction models, and spatial heterogeneity outside training regions. To address this challenge, we first collected 120 globally distributed 1 degrees & times; 1 degrees high-resolution bare-earth digital terrain model (DTM) tiles from mountainous regions for model development. An explainable automated machine learning-Shapley additive explanations (AutoML-SHAP) framework was then introduced to evaluate 35 commonly used prediction features, and 15 optimized features that effectively captured the bias sources were identified. A hybrid convolutional neural network (CNN)-Transformer model was subsequently developed to correct GDEM biases across mountainous regions for global bare-earth DTM (GDTM) generation. For further accuracy improvement beyond training regions, a fusion model was proposed to integrate the corrected results with advantageous corrections of multisource GDTMs by using globally distributed Ice, Cloud, and Land Elevation Satellite-2 altimetry. The 1-arcsecond Copernicus DEM was selected as the target because of its demonstrated superior accuracy. Experimental results showed that (1) the optimized features substantially outperformed those selected by traditional feature selection methods and used in previous studies. (2) The proposed correction model achieved average relative improvements of 43.13%-76.86% in vertical accuracy over Copernicus DEM, and surpassed eight correction models and three GDTM products in terms of vertical accuracy and spatial consistency across diverse validation datasets. (3) Furthermore, the fusion model consistently improved correction accuracy in regions outside the training areas. This study integrated an explainable model, advanced deep learning techniques, fusion strategies, and multisource elevation datasets to generate high-accuracy bare-earth GDTMs in mountainous regions worldwide, offering valuable insights for terrain modeling and geospatial analysis. The code and dataset are available at link: https://doi.org/10.6084/m9.figshare.30186856.

Global digital elevation models (GDEMs) are critical in the measurement and analysis of Earth's surface, and should be evaluated prior to use. However, existing GDEM evaluations mainly use global statistical metrics to evaluate vertical elevation (VE) differences with reference data, ignoring the relationship between a centre pixel and its neighbouring pixels, which is defined as the GDEM's neighbouring structure (NS). Along track ATL03 points allow evaluation of the along track NS (ATNS). This study comprehensively accesses the VE and ATNS accuracy of 1 arc-second GDEMs, including Copernicus, NASA, AW3D30 and ASTER DEM, using for the first time ICESat-2 ATL03 along track points throughout the Tibetan Plateau, where the rugged terrains and various features make it difficult to maintain its NS's accuracy. This study first introduces continuous and discrete ATNS metrics, then evaluates their effectiveness by analysing their relationships with errors in GDEM terrain derivatives. Finally, the better-performing metric is used for evaluations across various terrain parameters, landforms and land covers. The proposed framework achieved DEM evaluations from pixel-by-pixel statistical analysis of elevation differences to local ATNS assessment. Evaluation results demonstrate that the ATNS errors of the GDEMs are linearly correlated with the RMSE of the vertical errors. Overall, the errors of the VE RMSE and ATNS are ranked as Copernicus< AW3D30 < NASA< ASTER. However, evaluations conducted in the Andes and Alps reveal regional variations in these rankings. This study endeavours to introduce a new framework for large-scale GDEM evaluations, and the conclusions are beneficial for GDEM selection in further applications.

Diffusion models have demonstrated significant potential for generating high-quality images, audio, and videos. However, their iterative inference process entails substantial computational costs, limiting practical applications. Recently, researchers have introduced accelerated sampling methods that enable diffusion models to generate samples with far fewer timesteps than those used during training. Nonetheless, as the number of sampling steps decreases, the prediction errors significantly degrade the quality of generated outputs. Additionally, the exposure bias in diffusion models further amplifies these errors. To address these challenges, we leverage a manifold hypothesis to explore the exposure bias problem in depth. Based on this geometric perspective, we propose a manifold constraint that effectively reduces exposure bias during accelerated sampling of diffusion models. Notably, our method involves no additional training and requires only minimal hyperparameter tuning. Extensive experiments demonstrate the effectiveness of our approach, achieving a FID score of 15.60 with 10-step SDXL on MS-COCO, surpassing the baseline by a reduction of 2.57 in FID.

Digital terrain modelling and analysis play critical roles in geoscientific research, which underpins the understanding of Earth's surface dynamics and processes. Traditional digital elevation models (DEMs) capture the geometric characteristics of the Earth's surface at a specific time, providing limited insights into the underlying geomorphological processes. To address these limitations, this study proposes the concept of value-added digital terrain, which extends elevation-only DEMs by integrating multidimensional information such as temporal dynamics, spatial relationships, and geomorphological attributes. The proposed framework encompasses quality enhancement methods, multidimensional information integration, and process-based dynamic modelling. Quality enhancement is achieved through a comprehensive framework involving positional accuracy improvement, terrain feature enhancement, spatial relationship optimization, and regional morphology reconstruction. Enrichment with multidimensional information, including temporal and spatial dimension extensions, object and morphological feature integrations, and attribute augmentation, allows for a deeper exploration of terrain processes and mechanisms. Finally, the paper highlights the significance of transitioning from morphological representation to process-oriented analysis, emphasizing the necessity for future digital terrain analysis to evolve towards dynamic, multidimensional, and globally applicable frameworks.

Lake water storage variations on the Tibetan Plateau (TP) serve as crucial indicators of regional hydrological dynamics and climate changes, providing more comprehensive insights than discrete measurements of lake area or water level alone. While accurate bathymetric data is fundamental for quantifying lake water storage, conventional bathymetric surveys are often constrained by logistical challenges and high operational costs in the remote region like the TP. The high altitude and minimal human activity on the TP result in exceptional lake water clarity, allowing laser altimetry to penetrate water depths of several tens of metres. In this study, we used data from Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) laser altimetry data collected from 2019 to 2023 to map five shallow, elongated lakes on the TP. First, we applied the DBSCAN denoising algorithm to eliminate anomalous photons and then fitted polynomial functions to the lakebed elevation profiles for individual tracks. Subsequently, we merged the profiles from all valid tracks within each lake area to derive comprehensive lakebed topography and depth estimates. Comparative analysis with depth measurements from previous studies revealed strong agreement in both absolute depths and spatial patterns of bottom topography. Our results showed that the water depths of the five studied lakes range from 0 to 47 m, with Puma Yumco identified as the deepest (maximum depth of 47 m) and Pelrap Tso as the shallowest (maximum depth of 26 m. The shoreline of Puma Yumco exhibited steeper topography compared to the other four lakes. This study demonstrated the capability of ICESat-2 laser altimetry as a cost-effective and reliable tool for lake bathymetry estimation on the TP. The approach presented in this study holds promise for broader applications in other regions with optically clear water bodies, thereby contributing to improve monitoring of lake dynamics and understanding of regional water storage dynamics and climate change impacts.

Compared to lake area and water level, lake storage capacity more intuitively reflects regional climate changes. In this study, we first derived lakebed elevation profiles for individual ICESat-2 tracks based on the underwater stratification of laser photons, then integrating all valid elevation tracks within the water body to interpolate the bathymetry. On this basis, we calculated the capacity and its time series directly, with the aid of lake boundaries and water levels obtained from optical imagery and CryoSat-2 data. Next, we also applied an empirical formula to estimate the water volume changes of Bangdag Co by combining the area and water levels from 2010 to 2023. Finally, we compared the results of Bangdag Co’s water volume changes obtained from the two different methods and conducted a detailed analysis of their performance and regional applicability. The bathymetric map of Bangdag Co reveals a distinct spatial pattern, with the northeastern part significantly deeper (with a maximum depth of 35.27 m) and the southwestern part shallower. The average depth of the lake is 13.99 m. We further estimated that the lake storage capacity in November 2023 was 2.95 km3. Water volume changes estimated using the empirical formula were highly consistent with those derived from the lake storage capacity time series (from 2010 to 2023, the lake storage capacity increased by 1.04 km3). Our comparison revealed that the empirical formula method reflects only changes in water volume. In contrast, while our method can accurately estimate lake storage capacity, it is constrained to shallow, clear, and elongated east-west lakes (e.g., Ayakkum Lake). In summary, the ICESat-2 laser altimetry data, which do not rely on measured water depths, offer an essential complement to underwater topography detection and provide a novel perspective on lake volume estimation research.

Glaciers, especially the small/local ones, are rapidly melting and disappearing due to their heightened sensitivity to climate change. A holistic understanding of the key criteria and fundamental challenges in developing materials for local glacier conservation is urgently needed, coupled with a call for interdisciplinary collaboration to effectively address the pressing issue of local glacier retreat.

This study aimed to quantify the impact of sponge city facilities on both runoff reduction and carbon emission mitigation, providing valuable insights for sustainable urban development. Using the Storm Water Management Model (SWMM) 5.2 in conjunction with carbon emission factor calculations, we comparatively evaluated the annual runoff reduction and carbon emission abatement potential of traditional drainage systems versus those incorporating sponge city facilities. Our results showed that the implementation of sponge city facilities resulted in a substantial decrease in runoff volume (100,840 m3), and a corresponding reduction in carbon emissions (7,089.85 kg CO2 eq) compared to the pre-renovation conditions. Additionally, this work assessed five sponge city facilities: green roofs, permeable pavements, sunken green spaces, rain gardens, and overflow storage ponds. Among these, overflow storage ponds demonstrated the highest efficiency in both runoff reduction (35,879 m3) and carbon emission mitigation (2,522.57 kg CO2 eq). Rain gardens showed the second-best performance, while sunken green spaces had the least impact. Our study provides a novel technical framework for quantifying and evaluating carbon emissions in urban drainage systems. Our findings offer reliable data support for urban planners and policymakers, contributing to evidence-based decision-making in the design and implementation of sponge city projects.

As a type of terrain break, the gully shoulder line is of great significance to locate the scope of surface failure and soil erosion. Based on the knowledge of the terrain structure of loess gullies, this paper modeled sight lines to calculate the three-dimensional maximum elevation angle (3D MEA). Then, a critical MEA value of 35(degrees) was determined via human vision to extract negative terrain and shoulder lines considering unmanned aerial vehicle (UAV)- and light detection and ranging (lidar)-based digital elevation models (DEMs). In addition to the DEM resolution, the uncertainties caused by the direction number, look distance and skip radius of computer sight were further examined. It was suggested that the process should involve observing at least 12 directions to detect complete information on terrain breaks. The simulation process required a relatively large look distance to analyze global terrain changes because a small distance could limit sight within the local window. The skip radius could result in the process ignoring nearby abrupt changes in terrain, thus causing negative terrain to be misclassified as positive terrain near shoulder lines. The concept of wall morphology was introduced to attribute the critical value to the repose angle of loess gullies. Eventually, the MEA was compared with many other parameters. It could be concluded that the slope is limited to the local scale, while the MEA could combine local and global scales using a large look distance. The hillshade is limited to one viewing direction, but the 3D MEA is a comprehensive layered index from multiple directions. This study is a novel exploration of interdisciplinary hydraulic engineering and information technology. This approach reduces terrain breaks from three-dimensional space into a one-dimensional index and simplifies shoulder lines as a critical 3D MEA value related to loess material properties. This research could be adopted for soil and water conservation.

The ice phenology of alpine lakes on the Tibetan Plateau（TP） is a rapid and direct responder to climate changes, and the variations in lake ice exhibit high temporal frequency characteristics. MODIS and passive microwave data are widely used to monitor lake ice changes with high temporal resolution. However, the low spatial resolutions make it difficult to effectively quantify the freeze-melt dynamics of lakes. This work used Sentinel-1 synthetic aperture radar（SAR） data to derive highresolution ice maps（about 6 days）, then with the aid of Sentinel-2 optical images to quantify freeze-melt processes in three typical lakes on the TP（e.g. Selin Co, Ayakekumu Lake, and Nam Co）. The results showed that three lakes had an average annual ice period of 125-157 days and a complete ice cover period of 72-115 days, from 2018 to 2022. They exhibit different ice phenology patterns. Nam Co is characterized by repeated episodes of freezing,melting, and refreezing, resulting in a prolonged freeze-up period. Meanwhile, the break-up period of Nam Co lasts for a longer duration（about 19 days）, and the break-up exhibits a smooth process. Similarly, Ayakekumu Lake showed more significant inter-annual fluctuations in the freeze-up period, with deviations of up to 28 days observed among different years. Compared to the other two lakes, Selin Co experienced a relatively short freeze-up and break-up period. In short, Sentinel-1 SAR data can effectively monitor the weekly and seasonal variations in lake ice on the TP. Particularly, this data facilitates quantification of the freeze-melt dynamics.

Remote sensing is an effective means for lake water level monitoring on the Tibetan Plateau (TP). The purpose of this study is to estimate water levels of lakes on the TP using the Global Ecosystem Dynamics Investigation (GEDI) and Cloud and Land Elevation Satellite-2 (ICESat-2), evaluate the performance of ICESat-2 and GEDI in estimating water levels, and analyze the differences of water level obtained by the two altimeters. The results showed that the average coefficient of determination (R 2 ) values between the estimated water levels (GEDI and ICESat-2) and the datasets (DAHITI and Hydroweb) were greater than 0.80, respectively. The water level of DAHITI and Hydroweb are mainly from radar nadir altimeters. The average root mean square error (RMSE) between GEDI and DAHITI was 0.54 m, between GEDI and Hydroweb was 0.38 m for Qinghai Lake. The average RMSE of Qinghai Lake between ICESat-2 and DAHITI was 0.50 m, and between ICESat-2 and Hydroweb was 0.28 m. The comparison results showed that the accuracy of GEDI seems to be slightly lower than that of ICESate-2. The main impact indicators of the difference between the GEDI and ICESat-2 in lake level estimations were the viewing angles (VAs), solar elevation, air temperature, and wind. From 2019 to 2021, GEDI covered 770 more lakes than ICESat-2, and the lake level fluctuation mainly occurred in the Inner Plateau and Yangtze basins. The GEDI can effectively estimate lake levels, which provides more water levels for lakes and lays a foundation for future research on the TP.

基于GF-2卫星影像数据、路网数据、POI数据,应用缓冲区分析、最短路径分析等GIS空间分析技术对凤阳县在宜居城市维度中的典型指标城市综合医院覆盖率和公交站点覆盖率进行研究.结果表明,凤阳县城市综合医院覆盖率为91.98％,评价结果为良,布局合理,可进一步优化;公交站点覆盖率为80.11％,评价结果为中,虽基本满足居民公交可达需求,但覆盖能力不足.23项宜居城市维度指标评价结果为6项优、9项良、4项中、4项差,总体优良比例为65.22％,表现良好,公共服务设施的平衡性和共享水平有待提高.GIS可视化特征能更直观地识别城市发展中的问题,提高城市宜居性评价精度.

Point clouds are widely used in Earth surface research but usually exhibit gaps of missing data. Previous point cloud restoration methods used in terrain modelling have not fully considered complex terrain characteristics, which can be summarised as the controlling role of topographic features in shaping terrain surfaces and the inherent similarities observed among these surfaces. This work introduces a novel method that integrates Topographic Features and Patch Matching (TFPM) into point cloud restoration processes for terrain modelling. The method mainly contains three steps. First, identifying gap boundary points. Second, topographic feature points are extracted and subsequently interpolated into the identified gaps. Third, searching other parts of the raw point cloud for patches resembling the gaps, and the identified patches are used as templates to restore the point cloud. The proposed method is benchmarked against three state-of-the-art point cloud restoration methods. The experimental results demonstrate that the TFPM method consistently exhibits superior accuracy in terrain modelling and analysis, as evidenced by low values of the root mean square error, average elevation difference, and average slope difference. This work endeavours to incorporate topographic features into point cloud restoration processes and can benefit future research related to terrain modelling and analysis.

Mountain glaciers and alpine lakes are significant indicators of regional climate change. The temperature rise has resulted in the retreat of glaciers, further influencing the ecological cycle and the water balance. The present work mainly focuses on recent glacier and lake elevation changes in the western Kunlun Mountains. The CryoSat-2 satellite altimetry with relatively dense ground track spacing can measure more elevation samples for small glaciers and lakes surrounded by complex topography. The CryoSat-2 data with various algorithms were employed to monitor the water levels of the three typical lakes and the surface elevation changes of their supplying glaciers. The glacier surface elevations showed a slightly increasing trend and remarkable spatial heterogeneity from 2010 to 2021. Meanwhile, the Bangda Lake and Aksayqin Lake displayed clear and strong increasing trends over the observation period with an annual change rate of +0.65 and +0.20 m / yr, respectively. Surprisingly, Gozha Lake exhibited a decreasing trend with the rate of -0.04 m / yr. Regional climate warming led to more melt-water from the glaciers and permafrost flowing into the three typical lakes. The increase in precipitation played a leading role in lake expansion over the western Kunlun Mountains. The present work demonstrated that the satellite altimetry data provide important supplementary data to help us perform the cryosphere research in the data-scarce northern Tibetan Plateau depopulated zone.

Terrain models are widely used to depict the shape of the Earth's surface. With the development of photogrammetric methods, point cloud data have become one of the most popular data sources for terrain modelling. However, the obtained point clouds are of high density, which often increases redundancy rather than improving accuracy. Therefore, point cloud simplification should be a core component of terrain modelling. This paper proposes a point cloud simplification method by integrating topographic knowledge into terrain modelling (TKPCS). The method contains two steps: (1) topographic knowledge recognition and construction and (2) point cloud simplification using this topographic knowledge for terrain modelling. The proposed approach is benchmarked against improved versions of existing methods to validate its capability and accuracy in digital elevation model construction and terrain derivative extraction. The results show that the simplified points of the TKPCS method can generate finer resolution terrain models with higher accuracy and greater information entropy. The good performance of the TKPCS method is also stable at different scales. This work endeavours to transform perceptive topographic knowledge into a process of point cloud simplification and can benefit future research related to terrain modelling.

针对以传统统计数据为基础的城市体检工作量大、更新周期慢、数据获取难的问题,提出将遥感大数据应用到绿色城市维度的城市体检指标计算中,对GF-2卫星全色波段光学影像采用多规则的面向对象分类法提取城市绿地等地物,来解决绿色城市维度具体指标的计算问题.结果表明,以凤阳县绿色城市维度典型指标为例计算后得出的蓝绿空间占比为24.64％,评价结果为差,人均公园绿地面积为12.90 m2/人,评价结果为中,其余20项指标中3项优、5项良、4项中、8项差,总体优良比例为36.36％,表现较差.该方法能实现体检过程的精细化操作,使得体检工作更加高效、结果更加准确.

Precipitation is a major component of the water cycle. Accurate and reliable estimation of precipitation is essential for various applications. Generally, there are three main types of precipitation products: satellite based, reanalysis, and ground measurements from rain gauge stations. Each type has its advantages and disadvantages. Recent efforts have been made to develop various merging methods to improve precipitation estimates by combining multiple precipitation products. This study evaluated for the first time the performance of the random forest-based merging procedure (RF-MEP) method in enhancing the accuracy of daily precipitation estimates in Chongqing city, China with a complex terrain and sparse observational data. The RF-MEP method was used to merge three widely used gridded precipitation products (CHIRPS, ERA5-Land, and GPM IMERG) with ground measurements from a limited number of rain gauge stations to produce the merged precipitation dataset. Eight stations (approximately 70% of the available stations) were used to train the RF-MEP approach, while four stations (30%) were used for independent testing. Various statistical metrics were employed to assess the performance of the merged precipitation dataset and the three existing precipitation products against the ground measurements. Our results demonstrated that the RF-MEP approach significantly enhances the accuracy of daily precipitation estimates, surpassing the performance of the individual precipitation products and two other merging methods (the simple linear regression model and the simple averaging). Among the three existing products, ERA5-Land exhibited the best performance in capturing daily precipitation, followed by GPM IMERG, while CHIRPS performed the worst. Regarding precipitation intensity, all three existing products and the RF-MEP merged dataset performed well in capturing light precipitation events with an intensity of less than 1 mm/day, which accounts for the majority (more than 70%) of occurrences. However, all datasets showed rather poor capability in capturing precipitation events beyond 1 mm/day, with the worst performance observed for extreme heavy precipitation events exceeding 50 mm/day. The RF-MEP approach significantly improves the detection ability for all precipitation intensities, except for the most extreme intensity (>50 mm/day), where only marginal improvement is observed. Analysis of the spatial pattern of precipitation estimates and the temporal bias of daily precipitation estimates further confirms the superior performance of the RF-MEP merged precipitation dataset over the three existing products.