Abstract. Accurate, high-resolution spatial data of paddy rice are indispensable for assessing global food security and tracking progress toward Sustainable Development Goal 2 (Zero Hunger). However, a consistent global rice map at medium-to-high resolution has been lacking due to the challenges of cloud contamination and the temporal irregularity of multi-source satellite archives. Here, we present GlobalRice20, the first global 20m resolution paddy rice dataset for the years 2015 and 2024. We developed a "Time-Series-to-Vision" framework (T2VRCM) that transforms heterogeneous optical and SAR time-series into standardized 2D visual representations, specifically designed to handle irregular sampling and missing modalities. The dataset was produced using Sentinel-1/2 and Landsat imagery and rigorously validated against 164,000 reference samples, achieving an overall accuracy of 92.33 %. Cross-comparison with national agricultural statistics reveals a high coefficient of determination (R2 = 0.91 for 2024), confirming the dataset's reliability for national-scale accounting. Spatiotemporal analysis during the first decade of SDGs (2015–2024) indicates a 6.6 % expansion in global rice area, with Africa exhibiting the most significant growth (15.7 %). This dataset fills a critical gap in global agricultural monitoring, providing a baseline for analyzing food production trends and climate impacts. The dataset is available at https://doi.org/10.5281/zenodo.18168302 (Zhang et al., 2026).

Zero Hunger (Sustainable Development Goal 2, SDG 2) serves as a cornerstone for achieving global sustainable development, and is intricately linked with other SDGs exhibiting complex and multifaceted synergies and trade-offs. While the interconnections among indicators referring to food system within environmental domain have been widely investigated, interactions among indicators of all three pillars (social, economic, and environmental) remain under-researched. This study leverages the 2020 Sustainable Development Solutions Network (SDSN) assessment data to construct a global SDG 2-related network comprising 38 targets and 61 indicators, and examine how this network’s structure varies across income levels. The results reveal high-income countries (HICs) have achieved notable advancements in eradicating hunger and improving agricultural productivity, while facing unique challenges of overnutrition. Low-income countries (LICs), by contrast, face persistent constraints in agricultural productivity, infrastructure, and resource access. Across the global SDG 2-related network, SDG 2 targets show direct synergies with 31 targets in other SDGs, covering all studied economic targets, whereas 10 targets exhibit direct trade-offs, all of which are related to the environment. The share of trade-offs declines as income rises, from 28 % in LICs to 13 % in HICs. Synergies mainly occur between economic targets in LICs, while they often occur between economic and social targets in HICs. Trade-offs linked to environmental targets indicate LICs rely more on natural resources, whereas HICs face environmental spillovers. These findings underscore the need for tailored strategies, with LICs prioritizing agricultural productivity and infrastructure, while HICs addressing social equity, social distribution, and environmental sustainability.

Poverty alleviation is critical for improving social equity and overall development, but ongoing climate change poses a significant threat for such efforts. A clear example is China’s targeted poverty alleviation campaign, which has recently achieved substantial success but is now under increasing climate pressure. Using multi-source geospatial data, we propose a novel framework to map the extent and types of hotspot regions (i.e. frontiers) of China’s poverty alleviation campaign and assess their associated climate vulnerabilities following the IPCC approach. We identify multi-dimensional inequalities in the climate vulnerabilities of different frontiers, regions, and population. The most vulnerable frontiers are livestock and agri-economy frontiers in western China: although they support less than 5% of the low-income population, they disproportionally occupy 50% of the area. In contrast, frontiers related to more spatially concentrated industries (e.g., tourism and other industries), which cover less than 20% of the total area while supporting over 70% of the population, show lower vulnerability. These results reveal the uneven development and climate risk in China’s poor regions and highlight the importance of industry agglomeration in combating poverty and mitigating future risks. Our study paves the way for global monitoring of poverty alleviation hotspots, informing climate adaptation policy-making and helping to prevent climate-induced poverty recurrence.

Timely and accurate high-resolution annual mapping of rice distribution is essential for food security, greenhouse gas emissions assessment, and support of sustainable development goals. East Asia (EA), a major global rice-producing region, accounts for approximately 29.3 % of the world's rice production. Therefore, to acquire the latest rice distribution of the EA, this study proposed a novel rice distribution mapping method based on the Google Earth Engine (GEE) platform, producing a 10 m resolution annual rice distribution map (EARice10) of EA for 2023. A new synthetic aperture radar (SAR)-based rice distribution mapping index (SRMI) was firstly proposed and combined with optical indices to generate representative rice samples. In addition, a stacking-based optical–SAR adaptive fusion model was designed to fully integrate the features of Sentinel-1 and Sentinel-2 data for high-precision rice mapping in EA. The accuracy of EARice10 was evaluated using more than 90 000 validation samples and achieved an overall accuracy of 90.48 %, with both the user accuracy and the producer accuracy exceeding 90 %. The reliability of the product was verified by R2 values ranging between 0.94 and 0.98 with respect to official statistics and between 0.79 and 0.98 with respect to previous rice mapping products. EARice10 is accessible at https://doi.org/10.5281/zenodo.13118409 (Song et al., 2024).

Understanding the spatial and temporal distribution of irrigated cropland at the field scale is essential for managing irrigation water use and addressing the water-food nexus. While global and regional irrigation products exist, they often classify irrigated crops based on machine learning principles, where irrigated crops outperform rainfed ones. However, these methods typically lack mechanistic representation and are rarely applicable at the field scale over long time series. Additionally, identifying irrigated cropland in dual-season systems poses challenges due to temporal heterogeneity, leading to potential misclassification. To address these issues, we constructed a 3D canopy feature space including hydrothermal characteristics (1-precipitation/ P, 2-actual evapotranspiration/AET) and spectral characteristic (3-NDVI). This approach is based on two mechanisms: the impact of irrigation on water vapor cycling and its role in promoting crop growth. We introduced a novel cross-region Slope Length Index (SLI) to map irrigated and rainfed crops at the field scale. Our method involved downscaling NDVI and AET using spectral fusion techniques (STF) on Google Earth Engine (GEE), followed by fitting a robust rainfed line (AET =-125.41 + 0.84 x P, R2 = 0.70) at the provincial scale, and calculating the SLI. Then A case of irrigation map (Irri_HNP) was generated by a threshold for crop water supply and demand, achieving >= 38 % accuracy improvement on overall accuracy (OA = 0.973) compared to existing products. The SLI method also exhibited strong stability when generalized to the national scope (AET =-74.41 + 0.82 x P, R2 = 0.73), maintaining robustness in both drought and humid years (AET =-177.08 + 0.82 x P, R2 = 0.69). The method's scalability and transferability have been rigorously validated across diverse regions and environments, spanning from provincial to national scales. This validation achieved an OA of 0.922, demonstrating robust performance under heterogeneous conditions. Furthermore, the framework provides actionable insights for field-scale crop management and agricultural water governance.

The inherent spatial heterogeneity of land types often leads to a class imbalance in remote sensing-based classification, reducing the accuracy of minority class detection. Consequently, current land use datasets are often inadequate for the specific needs of soil erosion studies. In response to the need for soil conservation in dry–hot valley regions, this study integrated multi-source remote sensing imagery and constructed three high-precision imbalanced sample datasets on the Google Earth Engine (GEE) platform to perform land use classification. The degree of class imbalance was quantified using the imbalance ratio (IR), and the impact of sample imbalance on the classification accuracy of different land use types in a typical dry–hot valley was analyzed. The results show that (1) Feature selection significantly improved both classification accuracy and computational efficiency. The period from February to April each year, between 2018 and 2023, was identified as the optimal time window for land use classification in dry–hot valleys. (2) Constructing composite images over longer time scales enhanced classification performance: using a 2020 annual composite image combined with a Gradient Tree Boosting classifier yielded the highest accuracy, indicating that longer temporal synthesis improves classification results. (3) The effect of class imbalance on classification accuracy varied by land type: woodland (the majority class) was least affected by imbalance, whereas minority classes such as cultivated land, garden plantations, and grassland were highly sensitive to imbalance. In imbalanced scenarios, minority classes are prone to omission errors, leading to notable accuracy declines; producer’s accuracy (PA) decreased by 46%, 42%, and 25% for cultivated land, garden plantations, and grassland, respectively, as IR increased (with PA dropping faster than user’s accuracy, UA). Cultivated land was especially sensitive and frequently overlooked under high imbalance conditions compared to gardens and grasslands. Despite overall accuracy improving with higher IR, the accuracy of these minority classes dropped significantly, underscoring the importance of addressing the class imbalance in land use classification for erosion-prone areas.

Timely and accurate mapping of rice cultivation distribution is crucial for ensuring global food security and achieving SDG2. From a global perspective, rice areas display high heterogeneity in spatial pattern and SAR timeseries characteristics, posing substantial challenges to deep learning (DL) models' performance, efficiency, and transferability. Moreover, due to their "black box" nature, DL often lack interpretability and credibility. To address these challenges, this paper constructs the first SAR rice dataset with spatiotemporal heterogeneity and proposes an explainable, lightweight model for rice area extraction, the eXplainable Mamba UNet (XM-UNet). The dataset is based on the 2023 multi-temporal Sentinel-1 data, covering diverse rice samples from the United States, Kenya, and Vietnam. A Temporal Feature Importance Explainer (TFI-Explainer) based on the Selective State Space Model is designed to enhance adaptability to the temporal heterogeneity of rice and the model's interpretability. This explainer, coupled with the DL model, provides interpretations of the importance of SAR temporal features and facilitates crucial time phase screening. To overcome the spatial heterogeneity of rice, an Attention Sandglass Layer (ASL) combining CNN and self-attention mechanisms is designed to enhance the local spatial feature extraction capabilities. Additionally, the Parallel Visual State Space Layer (PVSSL) utilizes 2D-Selective-Scan (SS2D) cross-scanning to capture the global spatial features of rice multi-directionally, significantly reducing computational complexity through parallelization. Experimental results demonstrate that the XM-UNet adapts well to the spatiotemporal heterogeneity of rice globally, with OA and F1-score of 94.26 % and 90.73 %, respectively. The model is extremely lightweight, with only 0.190 M parameters and 0.279 GFLOPs. Mamba's selective scanning facilitates feature screening, and its integration with CNN effectively balances rice's local and global spatial characteristics. The interpretability experiments prove that the explanations of the importance of the temporal features provided by the model are crucial for guiding rice distribution mapping and filling a gap in the related field. The code is available in https://github.com/SAR-RICE/XM-UNet.

Urban land and cropland are two land-use types most significantly affected by human activities, serving as critical indicators for the United Nations’ SDG11 (Sustainable Cities and Communities) and SDG2 (Zero Hunger) . As a rapidly developing major economy, China has experienced accelerated urbanization in recent years, resulting in intensified conflicts between urban expansion and cropland protection, thereby reflecting the trade-offs between SDG11 and SDG2. Using remote sensing-based change data and projections of urban land expansion by 2030, this study applies GIS spatial analysis and indicator-based methods to explore the spatial and temporal characteristics of cropland occupation by urban expansion around the baseline year 2015, both at the national level and within major grain producing region. The study finds that: 1) The most significant cropland loss due to urban expansion is projected in the middle and lower reaches of the Yangtze River, while the Northeast shows the highest rate of decline. Among cities of different sizes, Type-I municipalities exhibit the greatest reduction. 2) In contrast to the national trend, urban expansion in major grain-producing areas is projected to exert an increasingly severe impact on cropland, particularly in the Huang-Huai-Hai Plain, the Southwest, and the Loess Plateau. Type-I and Type-IV municipalities within these regions show the most pronounced effects. 3) Under various Shared Socioeconomic Pathways, the sustainable development scenario has not effectively curbed cropland loss. Urban expansion continues to encroach upon cropland substantially, with cropland being the primary contributor to land conversion. These findings suggest that sustainable development scenarios should be adjusted to better incorporate food security concerns.

In recent years, the demand for rice in Africa has been growing rapidly, and, in order to meet this demand, the rice cultivation area is also expanding rapidly; thus, it is of great significance to monitor the rice cultivation in Africa. The spatial and temporal distribution of rice cultivation in Africa is complex, making it difficult to use phenology-based rice identification methods, and the existing rice distribution products of Africa are all made up of grid-based statistical data with a low resolution, unable to obtain accurate rice field location and available labels. To address these two difficulties, based on time series optical and dual-polarization synthetic aperture radar (SAR) data, this study proposes a sample set construction method by means of fast-coarse-positioning-assisted visual interpretation and a feature-importance-guided supervised classification combining multiple temporal optical and SAR features to reduce the impact of rice diversity in Africa. Firstly, we use the time series statistical features of vertical transmit, horizontal receive (VH) data for fast coarse positioning and screening of possible rice areas and combine multiple auxiliary data for visual interpretation to construct the sample set; secondly, based on the complementary information in SAR data and optical data, the 20 m Africa rice distribution map of 2023 was completed by combining the object-oriented segmentation results of temporal optical images and the pixel-based classification results of temporal SAR data features after feature selection. The average classification accuracy of the proposed method for the validation set is more than 85 %, and the R-2 of the linear fit to various existing statistical data is more than 0.9, which proves that the proposed method can achieve the spatial distribution mapping of rice under complex climatic conditions in a large region, providing crucial data support for rice monitoring and agricultural policy development. The dataset is available at 10.5281/zenodo.13729353 (Jiang et al., 2024).

Accurate delineation of urban-rural boundaries is fundamental for understanding urbanization dynamics and informing sustainable land-use and ecological governance. Yet most existing approaches suffer from limited accuracy and oversimplification of fragmented urban edges. To address these limitations during China’s rapid urbanization, we propose a high-precision framework for nationwide boundary extraction by integrating multi-source remote sensing data. Specifically, we construct a Nighttime Light Adjusted Urban Impervious Surface Index (NAIUI) that fuses impervious surface density with nighttime light intensity to enhance boundary delineation. Applied to China from 1985 to 2020, our approach reveals that urban built-up areas expanded from 22,165 km ^2 to 143,283 km ^2 (annual growth of 3,461 km ^2 ), with marked east-west heterogeneity. Validation shows an overall accuracy of 94.9% and a Kappa coefficient of 81.5%, outperforming existing national-scale products. By providing more reliable, fine-scale boundaries over long time series, the framework can support evidence-based spatial planning, infrastructure allocation, and ecological management, and it offers a practical platform for future integration with deep-learning methods to achieve near real-time monitoring.

Irrigated dryland in China, largely distributed and expanding in arid and semi-arid regions, always poses challenges to water resource and ecosystems, thus deeply impacts the implementation of SDG 2, 6, 15 and so on. Spatial-temporal patterns of irrigated dryland with high resolution help pinpoint hotspots and key issues worth to concern, however, are still lacking in research. This study developed a zonal-specific methodology to identify irrigated dryland and its changes from 2000 to 2015 at national scale with 30 m resolution. The whole China was divided into five zones based on cropping patterns. Key phases of remote sensing data and environmental factors were combined to generate zonal-specific methodologies for the identification of irrigated dryland. Meanwhile changes of features were employed to identify changes of irrigated dryland. Results show that the ratio of vegetation indices to environmental factors exhibits stronger stability in the classification of each zone. That being said, these of slope and texture varied from zone to zone. Moreover, using difference data between two periods for change information extraction demonstrated high accuracy. This method not only avoids the error accumulation caused by overlaying the direct classification results of two periods but also addresses the issue of insufficient classification accuracy due to the shortage of samples and data in historical periods. The average overall accuracy (the kappa coefficient) on generated maps of irrigated dryland in 2000 and 2015 are 88.28% (0.758) and 87.65% (0.744). The irrigated dryland had been increased by 8.84% from 2000 to 2015 and their distribution became denser and shifted towards north of China. With the advancement of economy and technology, the influence of human factors on the distribution of irrigated drylands has gradually intensified. The stable development of irrigated drylands ensures China’s food security and contributes to global food security. Meanwhile, it poses significant challenges to water resources and the ecological environment. This research developed methodologies to obtain accurate irrigated dryland data and its change information, supporting to identify the interaction between SDG 2, SDG 6, and SDG 15, especially in arid and semi-arid regions with vulnerable ecosystem in the world.

The well-facilitated farmland projects (WFFPs) involve the typical sustainable intensification of farmland use and play a key role in raising food production in China. However, whether such WFFPs can enhance the nitrogen (N) use efficiency and reduce environmental impacts is still unclear. Here, we examined the data from 502 valid questionnaires collected from WFFPs in the major grain-producing area, the Huang-Huai-Hai Region (HHHR) in China, with 429 samples for wheat, 328 for maize, and 122 for rice. We identified gaps in N use efficiency (NUE) and N losses from the production of the three crops between the sampled WFFPs and counties based on the statistical data. The results showed that compared to the county-level (wheat, 39.1%; maize, 33.8%; rice, 35.1%), the NUEs for wheat (55.2%), maize (52.1%), and rice (50.2%) in the WFFPs were significantly improved (P<0.05). In addition, the intensities of ammonia (NH3) volatilization (9.9-12.2 kg N ha(-1)), N leaching (6.5-16.9 kg N ha(-1)), and nitrous oxide (N2O) emissions (1.2-1.6 kg N ha(-1)) from crop production in the sampled WFFPs were significantly lower than the county averages (P<0.05). Simulations showed that if the N rates are reduced by 10.0, 15.0, and 20.0% for the counties, the NUEs of wheat, maize, and rice in the HHHR will increase by 2.9-6.3, 2.4-5.2, and 2.6-5.7%, respectively. If the N rate is reduced to the WFFP level in each county, the NUEs of the three crops will increase by 12.9-19.5%, and the N leaching, NH3, and N2O emissions will be reduced by 48.9-56.2, 37.4-42.9, and 46.0-66.5%, respectively. Our findings highlight that efficient N management practices in sustainable intensive farmland have considerable potential for reducing environmental impacts.

Africa faces significant challenges in food security, which are compounded by rapid population growth. The situation is expected to worsen without effective interventions. One of the key obstacles to mitigating food insecurity is the lack of reliable high-resolution data, particularly regarding harvested area datasets, which directly reflect the agricultural situation on the continent. To tackle this challenge, we have developed the African Harvested Area Dataset (AHAD), which covers 22 major crops across the continent at a resolution of 5 arcmin (approximately 10 km) for the years 2000, 2010, and 2020. The dataset is built upon 8 well-used global gridded harvested area datasets, through verifying, merging, calibrating, and confining with available information, including point-specific crop distribution, accurate cropland map data, subnational statistics, and cropping intensity data. In addition to the primary datasets, we also provide data quality assessments for each step of the process. The AHAD could provide geospatial and temporal patterns of harvested area, offering potential for advancing agricultural practices and enhancing food security across Africa.

Soil salinization, a critical form of global soil degradation, threatens agricultural productivity and ecosystem functions. Accurate mapping of soil salinity is essential for sustainable land management and informed decisionmaking. However, conventional optical or radar satellite sensors are often limited in detecting key salinity spectral signatures due to their insufficient thermal infrared (TIR) coverage. TIR remote sensing offers unique advantages for soil salinity assessment, owing to its sensitivity to the emissivity of saline soils within the TIR spectrum, but its application remains underexplored. This study evaluated the suitability and robustness of SDGSAT-1 TIS data for large-scale soil salinity mapping in the Songnen Plain, China, one of the world's three largest soda saline-alkali soil regions. We compared the performance of soil salinity models integrating SDGSAT1 TIS data with those using optical (Sentinel-2) and radar (Sentinel-1 and GF-3) data across several machine learning techniques. Our results demonstrated that incorporating SDGSAT-1 TIS data significantly enhanced soil salinity modeling accuracy, consistently outperforming models based solely on Sentinel-2 optical or Sentinel-1/ GF-3 radar data. The combination of SDGSAT-1 TIS and Sentinel-2 data, optimized using the Gaussian Process Regression model, achieved the highest accuracy (R2 = 0.75, RMSE = 0.65 dS/m). The resulting salinity maps revealed widespread soil salinization across the region, with the majority of the area exhibiting slight to moderate salinity levels, posing substantial challenges to plant growth and ecosystem resilience. This study offers a robust, data-driven validation of TIR's unique sensitivity to soil salinity, emphasizing its potential for integration into large-scale soil salinity mapping frameworks.

In a context of economic growth and increasing urban populations, assessing the impact of urban-cropland interactions on crop production in developed countries offers insights about future global scenarios. We quantitatively assessed the impact of urbanization on crop production in the United States since 2000 by combining earth observation products and crop modelling. We found that urbanization has appropriated 2.1 M ha of highly productive cropland, leading to an overall decline in U.S. cropland area over time and a production loss of 15 million tons of maize, soybean, and wheat. Avoiding the negative impacts of urbanization on crop production and the environment will require proper land-use planning and urban design and yield intensification on existing cropland.

Complete remote-sensing time series with consistent length are important for obtaining reliable analytical results in regional applications. Temporal composite reconciles incomplete raw remote-sensing time series into a time series with fixed and equidistant time intervals, forming the basis for applications like crop classification. To our knowledge, no algorithms or indicators exist for determining the optimal composite interval while quantitatively considering cloud conditions and image acquisition capabilities across different latitudes. As a pioneering effort, we propose an optimal composite interval index (OCII) for producing remote-sensing composite time series. This index first calculates the proportion of valid (cloud-free) composite observations and the information loss from compositing, then describes the trade-off between these two aspects with a simple normalized difference form. We tested OCII using crop classification as an example, accessing classification accuracy with composite time-series data of varying intervals. Experimental results from three regions with different geographic conditions show that OCII suggested 16-day, 25-day and 30-day intervals for the sites with low cloud cover (37.56%), medium cloud cover (56.37%), and high cloud cover (82.11%), respectively. Classification accuracy was poor with either too short or too large composite intervals, and the optimal composite interval derived from OCII achieved a relative accuracy increase of 2.8-10.2%. This underscores the effectiveness of OCII considering differences in data availability at clear and cloudy sites. Calculating OCII requires only the data quality layer and can be easily implemented in various areas using the Google Earth Engine platform. We believe that OCII has great potential for crop classifications and other applications of remote-sensing time-series data.

Agriculture is an important contributor to global carbon emissions. With the implementation of the Sustainable Development Goals of the United Nations and China's carbon neutral strategy, accurate estimation of carbon emissions from crop farming is essential to reduce agricultural carbon emissions and promote sustainable food production systems in China. However, previous long-term time series estimates in China have mainly focused on the national and provincial levels, which are insufficient to characterize regional heterogeneity. Here, we selected the county-level administrative district as the basic geographical unit and then generated a county-level dataset on the intensity of carbon emissions from crop farming in China during 2000-2019, using random forest regression with multi-source data. This dataset can be used to delineate spatio-temporal changes in carbon emissions from crop farming in China, providing an important basis for decision makers and researchers to design agricultural carbon reduction strategies in China.

BACKGROUND:Maize, wheat, rice and soybean production are intimately linked to food security. Identifying the key factors affecting crop yields and determining the countries where increased irrigation and nitrogen application most effectively enhance yields are essential steps towards achieving sustainable development goals and ensuring food security. Identifying these areas is crucially dependent on yield gaps. However, the lack of comparability between different regions in current regional-scale yield gap studies stems from varied methodologies. Moreover, global yield gap research, relying on statistical models and regression methods, tends to neglect the crop growth process. In this study, we used a random forest model, based on statistical and meteorological data, to pinpoint the key factors influencing crop yields. Subsequently, using unified yield data from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP), derived from crop models simulations, we applied the yield gap method to calculate the potential yield increase for four crops across countries, under conditions of full irrigation and nitrogen application. RESULTS:Our research finds that nitrogen application is the main factor affecting yields globally, while irrigation plays a crucial role in the major producing countries. The countries with high potential for yield increases are located at the border between Africa and Eurasia. The global average yield of the four major crops increased 13.7-29.8% under full irrigation, 2.9-39.1% under full nitrogen application and 29.4-97.8% under both conditions. CONCLUSION:This study provides crucial insights into global crop yield changes and their determinants, which are highly important for global sustainable agriculture and food security efforts. © 2024 Society of Chemical Industry.