Gaseous air pollutants pose significant threats to environmental quality and public health, with their diurnal variation patterns critically determining human exposure and health risks. However, the spatiotemporal dynamics and driving factors of these variations remain insufficiently understood at the national level. This study characterizes the diurnal variation of four key pollutants (SO2, NO2, CO, and O3), using hourly observations from 1636 monitoring stations across China spanning 2014-2024. By integrating a Light Gradient Boosting Machine (LightGBM) model with SHapley Additive exPlanations (SHAP), we quantified the non-linear contributions of meteorological, topographic, and socioeconomic drivers to pollutant variability, with the robustness of the results verified through spatial sensitivity analysis. The results reveal pronounced spatial and seasonal heterogeneity in diurnal ranges. Specifically, SO2 is shaped by the synergy of socioeconomic and meteorological factors; NO2 is dominated by socioeconomic intensity but amplified by plateau meteorology; CO is driven by high-frequency emission cycles such as traffic patterns; and O3 is strictly regulated by solar radiation and topography. These findings elucidate the distinct roles of emission intensity, atmospheric stability, and chemical processing in shaping diurnal cycles, underscoring the necessity of region-specific and time-targeted mitigation strategies to minimize population exposure and advance precision air-quality management.

Abstract The Alps are currently considered among the ecoregions with the highest magnitude of average bark beetle disturbance per year in Europe. We present a disturbance characterization based on a unique database including more than 50,000 records of ground‐based bark beetle disturbance observations in the Eastern Alps from 2020 to 2023. The dataset was used to extract precise temporal and spatial information on disturbance events in terms of size, distance, intensity, and frequency. Disturbance events were modeled as spatial point processes based on scale dependency (landscape‐regional) and their deviations from random distributions were assessed. Parameters typically used in forest disturbance models such as clustering degree, intensity slope, and probability scale were retrieved. Additionally, aboveground biomass affected by bark beetle disturbance was estimated. Disturbance metrics were found to be heterogeneous over time, revealing a decrease in single infestation spot sizes as the epidemics progressed in combination with a reduction in inter‐event distances and an increase in spatial frequency. At the landscape scale, point processes showed very distinct distributions in space with nestlike clustering, clustering towards portions of the landscape, or no clustering at all. The clustering degree and the probability scale sharply increased from the build‐up to the highly epidemic phase. Four percent of the total regional aboveground biomass was affected by bark beetle disturbances over the four‐year study period. The extracted disturbance metrics and parameters can help in the correct parameterization of forest disturbance models, thus supporting our capability of predicting future patterns of beetle dispersal and their effects on carbon stocks in the alpine region.

Forests play a central role in the global carbon cycle by serving as critical carbon sinks for atmospheric CO2. Yet, the stability and continued capacity of these sinks are increasingly threatened by a growing number, size, and severity of disturbances. Accurately representing the stochastic nature of disturbance remains a major challenge and a key source of uncertainty in our understanding of carbon cycle dynamics. This study presents a novel framework for deriving disturbance regimes characterized by disturbance rate (μ), gap-size distribution (α), disturbance severity (β), as well as background mortality (Kb) directly from landscape features of high-resolution satellite biomass data. These regimes reflect the characteristics of long-term disturbances at the landscape scale rather than the properties of any single event. Our analysis inverts the model framework developed by Wang et al. (2024), which used a machine learning model trained on a massive synthetic dataset of over 8 million forward model simulations to link known disturbance regimes to spatial biomass patterns. Instead of predicting patterns from regimes, we use observed satellite biomass patterns to infer the underlying disturbance regimes. To ensure robustness, we first identified the optimal spatial resolution for aggregating both simulation and satellite data, minimizing discrepancies in feature value ranges and reducing extrapolation risk. Using this framework, we produced the first globally continuous, observationally constrained dataset of forest disturbance regime parameters and their associated uncertainties, provided at both a 25×25 km tile level and as a gridded 0.25° global product. Additionally, we used a Dissimilarity Index (DIK) to quantify prediction uncertainty and identify potential extrapolation by measuring the divergence of observations from the training set. An empirical evaluation comparing paired forest landscapes with contrasting extremes in their predicted disturbance parameters support the methodology and assumptions used to build the dataset. The datasets are available at https://doi.org/10.17617/3.EF4QGR (Wang et al., 2026). Our global maps of disturbance regimes provide a novel, process-based tool for investigating the coupled dynamics of disturbance, vegetation, and the carbon cycle, with potential applications for improving the representation of stochastic disturbances in large-scale ecosystem models.

Marine chlorophyll concentration is an important indicator of ecosystem health and carbon cycle strength, and its accurate prediction is crucial for red tide warning and ecological response. In this paper, we propose a LSTM-RF hybrid model that combines the advantages of LSTM and RF, which solves the deficiencies of a single model in time-series modelling and nonlinear feature portrayal. Trained with multi-source ocean data(temperature, salinity, dissolved oxygen, etc.), the experimental results show that the LSTM-RF model has an R^2 of 0.5386, an MSE of 0.005806, and an MAE of 0.057147 on the test set, which is significantly better than using LSTM (R^2 = 0.0208) and RF (R^2 =0.4934) alone , respectively. The standardised treatment and sliding window approach improved the prediction accuracy of the model and provided an innovative solution for high-frequency prediction of marine ecological variables.

Forests, which sequester atmospheric CO2 in the form of biomass within long-term reservoirs, are critical to the global land carbon sink. Current land surface models assume a strong coupling between photosynthesis and plant biomass changes, with carbon supply (i.e., carbon assimilation though photosynthesis) has been considered the main driver of plant biomass growth (‘source limitation’). However, the potential for sink limitations to constrain plant biomass growth, where plant biomass change become decoupled from photosynthesis, has been raised and supported by free air CO2 enrichment (FACE) experiment, inventory and tree ring evidence.In this study, we relied on high spatial resolution satellite-based retrievals of above-ground biomass (AGB) and vegetation primary productivity (GPP), to quantify the extent of decoupling between plant photosynthesis and biomass growth at the ecosystem scales over the past decade in the Northern Hemisphere (from 35ºN to 90ºN). We found that the fraction of decoupled area in non-intact forest is 66 ± 9%, significantly higher than in the intact forest. Extensive decoupling was observed across Europe, Russia and Canada. This spatial pattern was verified using multiple satellite-derived and inventory-derived AGB, and GPP data from P model. To investigate the drivers of decoupling, we built a generalized additive model to predict spatial variations in decoupling fractions within non-intact forests. The model suggests that harvest and logging account for most of the decoupling in Europe, while wildfires in Siberia may promote a recovery of coupling due to rapid vegetation regrowth. More importantly, even in intact forests, 56 ± 13% still exhibited the decoupling signals. In western Russia, this decoupling appears to be driven by droughts, likely due to carbon allocation shifts to support metabolism and critical plant functions, thereby constraining biomass growth. In western Canada, decoupling was found in in old-growth, or dense intact forests, where high decomposition, competition, or mortality may result in stable or declining forest biomass over time. Our analysis provides a geographic overview of regions experienced sink limitations to forest biomass growth, as well as insights into the mechanisms regulating terrestrial carbon sequestration. These findings represent a critical step toward improving process-based models and enhancing predictions of terrestrial carbon dynamics under future climate change scenarios.

As China experiences a dramatic surge in carbon emissions from the transportation sector, reducing these emissions has become crucial in mitigating climate change. Since 2020, Navigation Service Providers has collaborated with traffic police departments to integrate the ‘Traffic Light Countdown’ module (TLCM) in the navigation applications in multiple cities across China. This initiative aims to improve vehicle traffic efficiency, reduce transportation carbon emissions, and enhance public driving safety simultaneously. To quantify the effectiveness of this module, we selected a section of Qilu Avenue in Jinan City as a case study. Localization of Motor Vehicle Emission Simulator (MOVES) model parameters to accurately reflect the specific conditions of the study area was achieved by incorporating local traffic, climate, vehicle, and fuel information. Utilizing this localized model, we evaluated the changes in traffic conditions, drivers’ driving behavior and running exhaust emissions in the study. The results indicate that, following the integration of the TLCM, the average vehicle speed on the selected road section increased from 29.5 to 31.8 km h ^−1 . Moreover, the ratio of rapid acceleration per vehicle decreased from 0.08 to 0.05, while the rapid deceleration dropped from 0.04 to 0.01. In terms of carbon emissions, the daily average carbon emissions for this road section decreased by 4.4%, with a particularly significant reduction during the evening rush hours and post-evening rush hours (16:00–22:00 Local Standard Time). This is the first quantitative assess the potential effect of the TLCM on improving road traffic conditions and reducing carbon emissions. These findings demonstrate that the integration of TLCM in navigation applications has not only improved traffic efficiency, reduced the occurrence of risky driving behaviors (rapid acceleration and deceleration) and thus enhanced road safety, but also positively contributed to the reduction of carbon emissions from transportation and the mitigation of climate change.

The European Alps are currently considered among the ecoregions with the highest magnitude of average bark beetle disturbance per year. We present a disturbance regime characterization based on a unique database including more than 50,000 records of ground-based bark beetle disturbance observations in the Eastern Alps for the years 2020 to 2023. The dataset was used to extract precise temporal and spatial information on disturbance events in terms of sizes, distances, intensity and frequency. Disturbance events were modeled as spatial point processes based on scale dependency (landscape-regional) and their deviation from random distributions was assessed. Parameters typically used in forest disturbance models such as clustering degree, intensity slope and probability scale were retrieved. Additionally, above-ground biomass loss was estimated. The disturbance metrics and parameters can help for the correct parameterization of forest disturbance models, and thus supporting our capability of predicting future patterns of beetle dispersal and effects on carbon stocks in the alpine region and beyond.

As a part of the integrated energy system (IES), gas pipeline networks can provide additional flexibility to power systems through coordinated optimal dispatch. An accurate pipeline network model is critical for the optimal operation and control of IESs. However, inaccuracies or unavailability of accurate pipeline parameters often introduce errors in the state-space models of such networks. This paper proposes a physics-informed recurrent network (PIRN) to identify the state-space model of gas pipelines. It fuses sparse measurement data with fluid-dynamic behavior expressed by partial differential equations. By embedding the physical state-space model within the recurrent network, parameter identification becomes an end-to-end PIRN training task. The model can be realized in PyTorch through modifications to a standard RNN backbone. Case studies demonstrate that our proposed PIRN can accurately estimate gas pipeline models from sparse terminal node measurements, providing robust performance and significantly higher parameter efficiency. Furthermore, the identified state-space model of the pipeline network can be seamlessly integrated into optimization frameworks.

Rice production faces increasing challenges from climate change and soil degradation. The conversion from double to single-cropping rice over the past decades has further threatened rice self-sufficiency in China. Understanding the spatial and temporal variations of rice yield across different rice-cropping systems is crucial for creating adaptation strategies. Here we used a process-based modelling approach combined with a nationwide field dataset from 1981 to 2020 to evaluate rice yield gaps and temporal yield variabilities for single and double rice-cropping systems, and further assessed their underlying determinants in China. We showed that single rice had the largest yield gap and the greatest temporal variability in yield, followed by late rice and early rice. The coefficient of variation (CV) for actual yield ranged from 6 % to 64 %, 4 % to 36 %, and 5 % to 28 % for single rice, late rice, and early rice, respectively. Regions with unstable yields were primarily located in southwestern (for single rice) and southern China (for late rice), and determinants of yield stability varied across subregions. Overall, the combined effects of climate and soil factors generally reduce yield stability. Improved management, such as appropriate sowing dates, precise fertilization, and cultivars with favorable traits, significantly enhanced the stability. Socio-economic factors including sufficient labor and advanced agricultural mechanization also contributed to closing yield gaps and stabilizing yield. This study provides spatial insights for developing region- specific strategies to ensure a sufficient and stable rice supply.

Accurate simulation of canopy photosynthesis is essential for predicting dry matter accumulation and crop yield. However, most current crop models overlook the effect of vertical distribution of leaf nitrogen and chlorophyll content on photosynthetic capacity at different canopy layers, resulting in greater uncertainties and weaker mechanistic explanation. Here, we developed a novel canopy photosynthesis model that establishes a bridge between chlorophyll content and photosynthetic nitrogen (PN, defined as total leaf nitrogen minus non-photosynthetic nitrogen) across different canopy heights, and then employs chlorophyll content as a reliable proxy forsimulating photosynthesis. The model was calibrated and validated using data from five field experiments under diverse treatments. Results indicate that leaves at higher canopy positions, receiving more light, contain higher nitrogen content and chlorophyll to support greater photosynthetic rates. The nitrogen extinction coefficient (KN), which characterizes the decline in available of leaf nitrogen, decreases exponentially with increasing LAI, varying among canopy depths, cultivars and growth stages. Chlorophyll shows a stronger correlation with photosynthesis compared to leaf nitrogen. By capturing these dynamics, the model enhances the accuracy of photosynthesis prediction by 60%, particularly correcting the overestimation of canopy photosynthesis and dry matter accumulation during post-flowering. These findings advance the understanding and modelling of canopy-scale photosynthesis in crop models and provide insights for better integration with chlorophyll-related remote sensing data.

Surface ozone (O3) pollution showed a continuous increasing trend during the recent decades in China, posing an increasing threat to food security. A wide range of yield reductions have been reported and thus more studies are needed to narrow down the uncertainty resulting from spatiotemporal accuracy of O3 metrics and extrapolation methods. Based on a high spatial resolution (0.1°) hourly surface O3 data, here we analyzed the spatiotemporal O3 pollution patterns and impacts on yield, production and economic losses for wheat, rice, and maize in China during 2005-2020. The accumulated O3 exposure over a threshold of 40 ppb (AOT40) increased by 10 % during 2005-2019, and a decrease of 5.56 % was observed in 2020 due to the COVID-19 lockdowns. Rising O3 pollution reduced national level wheat, rice and maize yields by 14.51 % ± 0.43 %, 11.10 % ± 0.6 %, and 3.99 % ± 0.11 %, respectively. A Business-As-Usual projection suggested that the relative yield loss (RYL) would potentially reach 8 %-18 % at the national scale by 2050 if no emission control is implemented. COVID-19 lockdowns in 2020 led to significantly reduced RYL for maize (0.52 %) and rice (2.17 %) but not for wheat (0.11 %), with the largest reduction (1.88 %-9.4 %) in North China Plain, highlighting the potential benefits of emission control. Our findings provided robust evidence that rising O3 pollution has significantly affected China's crop yields, production and economic losses, underscoring the urgent need to curb O3 pollution to safeguard food security, particularly in densely populated and industrialized regions.

The land carbon cycle is fundamental in regulating atmospheric CO2 dynamics from seasonal to centennial scales. The fine equilibrium between photosynthetic gains spent in metabolic costs and/or lost in mortality underpins the contribution of terrestrial ecosystems to the global carbon cycle. Uncertainties and divergent hypotheses on the role of climate in regulating their underlying mechanisms hamper our current diagnostic and prognostic abilities despite growing evidence on ecosystem vulnerability to present and future changes in climate. However, quantitative knowledge of the contributions of different carbon cycle processes regulating carbon uptake and release still need to be improved, inflating the uncertainties in future projections of net land-atmosphere carbon exchanges. In this study, we rely on satellite-based Earth observation retrievals of above-ground biomass and vegetation primary productivity to reconstruct the land-atmosphere carbon exchange dynamics over the last two decades, through the application of a three-box model at the pixel level. Our approach confidently reproduces 60% of the observed variability in atmospheric CO2 growth rate (CGR) over throughout 1997-2019 (R = 0.78, p-val < 0.05), with a low RMSE of 1.0 PgC yr-1. We further detail CO2 release from vegetation dynamics emerging from quick turnover induced by wildfires and leaves senescence, as well as the slow turnover from plant and soil decomposition mechanisms. This allows us to differentiate between transient and lagged effects on land-to-atmosphere fluxes. The carbon release, characterized by a lag of over one year, referred to as lagged effects. Globally, the lagged responses accounted for 50% of the variability in CGR, exceeding three times the contribution of transient fluxes from live vegetation. We have yet to a change or trend in the total contributions of vegetation dynamics to CGR. Yet, the relative role of lagged effects to CGR via decomposition fluxes increased by 50%, possibly due to accelerated mortality and decomposition fluxes. As global warming imposes higher stress on vegetation while increasing temperature-mediated decomposition, our results highlight the importance of quantifying their underlying metabolic responses. Such understanding is instrumental for assessing the contribution of adaptation and mitigation measures that will shape the contribution of the terrestrial carbon cycle to dampen the effects of anthropogenic emissions on global climate.

Lentic systems (lakes and reservoirs) are emission hotpots of nitrous oxide (N 2 O), a potent greenhouse gas; however, this has not been well quantified yet. Here we examine how multiple environmental forcings have affected N 2 O emissions from global lentic systems since the pre-industrial period. Our results show that global lentic systems emitted 64.6 ± 12.1 Gg N 2 O-N yr −1 in the 2010s, increased by 126% since the 1850s. The significance of small lentic systems on mitigating N 2 O emissions is highlighted due to their substantial emission rates and response to terrestrial environmental changes. Incorporated with riverine emissions, this study indicates that N 2 O emissions from global inland waters in the 2010s was 319.6 ± 58.2 Gg N yr −1 . This suggests a global emission factor of 0.051% for inland water N 2 O emissions relative to agricultural nitrogen applications and provides the country-level emission factors (ranging from 0 to 0.341%) for improving the methodology for national greenhouse gas emission inventories.

Abstract Many agricultural regions in China are likely to become appreciably wetter or drier as the global climate warming increases. However, the impact of these climate change patterns on the intensity of soil greenhouse gas (GHG) emissions (GHGI, GHG emissions per unit of crop yield) has not yet been rigorously assessed. By integrating an improved agricultural ecosystem model and a meta‐analysis of multiple field studies, we found that climate change is expected to cause a 20.0% crop yield loss, while stimulating soil GHG emissions by 12.2% between 2061 and 2090 in China's agricultural regions. A wetter‐warmer (WW) climate would adversely impact crop yield on an equal basis and lead to a 1.8‐fold‐ increase in GHG emissions relative to those in a drier‐warmer (DW) climate. Without water limitation/excess, extreme heat (an increase of more than 1.5°C in average temperature) during the growing season would amplify 15.7% more yield while simultaneously elevating GHG emissions by 42.5% compared to an increase of below 1.5°C. However, when coupled with extreme drought, it would aggravate crop yield loss by 61.8% without reducing the corresponding GHG emissions. Furthermore, the emission intensity in an extreme WW climate would increase by 22.6% compared to an extreme DW climate. Under this intense WW climate, the use of nitrogen fertilizer would lead to a 37.9% increase in soil GHG emissions without necessarily gaining a corresponding yield advantage compared to a DW climate. These findings suggest that the threat of a wetter‐warmer world to efforts to reduce GHG emissions intensity may be as great as or even greater than that of a drier‐warmer world.

Abstract Natural and anthropogenic disturbances are important drivers of tree mortality, shaping the structure, composition, and biomass distribution of forest ecosystems. Differences in disturbance regimes, characterized by the frequency, extent, and intensity of disturbance events, result in structurally different landscapes. In this study, we design a model‐based experiment to investigate the links between disturbance regimes and spatial biomass patterns. First, the effects of disturbance events on biomass patterns are simulated using a simple dynamic carbon cycle model based on different disturbance regime attributes, which are characterized via three parameters: μ (probability scale), α (clustering degree), and β (intensity slope). 856,800 dynamically stable biomass patterns were then simulated using combined disturbance regime, primary productivity, and background mortality. As independent variables, we use biomass synthesis statistics from simulated biomass patterns to retrieve three disturbance regime parameters. Results show confident inversion of all three “true” disturbance parameters, with Nash‐Sutcliffe efficiency of 94.8% for μ, 94.9% for α, and 97.1% for β. Biomass histogram statistics primarily dominate the prediction of μ and β, while texture features have a more substantial influence on α. Overall, these results demonstrate the association between biomass patterns and disturbance regimes. Given the increasing availability of Earth observation of biomass, our findings open a new avenue to understand better and parameterize disturbance regimes and their links with vegetation dynamics under climate change. Ultimately, at a large scale, this approach would improve our current understanding of controls and feedback at the biosphere‐atmosphere interface in the present Earth system models.

The land carbon cycle is fundamental in regulating atmospheric CO2 dynamics from seasonal to centennial scales. The fine equilibrium between photosynthetic gains spent in metabolic costs and/or lost in mortality underpins the contribution of terrestrial ecosystems to the global carbon cycle. Uncertainties and divergent hypotheses on the role of climate in regulating their underlying mechanisms hamper our current diagnostic and prognostic abilities despite growing evidence on ecosystem vulnerability to present and future changes in climate. However, quantitative knowledge of the contributions of different carbon cycle processes regulating carbon uptake and release still need to be improved, inflating the uncertainties in future projections of net land-atmosphere carbon exchanges. In this study, we rely on satellite-based Earth observation retrievals of above-ground biomass and vegetation primary productivity to reconstruct the land-atmosphere carbon exchange dynamics over the last two decades, through the application of a three-box model at the pixel level. Our approach confidently reproduces 60% of the observed variability in atmospheric CO2 growth rate (CGR) over throughout 1997-2019 (R = 0.78, p-val < 0.05), with a low RMSE of 1.0 PgC yr-1. We further detail CO2 release from vegetation dynamics emerging from quick turnover induced by wildfires and leaves senescence, as well as the slow turnover from plant and soil decomposition mechanisms. This allows us to differentiate between transient and lagged effects on land-to-atmosphere fluxes. The carbon release, characterized by a lag of over one year, referred to as lagged effects. Globally, the lagged responses accounted for 50% of the variability in CGR, exceeding three times the contribution of transient fluxes from live vegetation. We have yet to a change or trend in the total contributions of vegetation dynamics to CGR. Yet, the relative role of lagged effects to CGR via decomposition fluxes increased by 50%, possibly due to accelerated mortality and decomposition fluxes. As global warming imposes higher stress on vegetation while increasing temperature-mediated decomposition, our results highlight the importance of quantifying their underlying metabolic responses. Such understanding is instrumental for assessing the contribution of adaptation and mitigation measures that will shape the contribution of the terrestrial carbon cycle to dampen the effects of anthropogenic emissions on global climate.

Wood density is an emergent property resultant of tree growth strategies modulated by local edapho-climatic and stand development conditions. It is associated with the biomechanical support of trees and hydraulic conductivity or safety, directly and indirectly influencing a range of ecological processes, including, among others, tree growth, tree resistance to disturbances, and mortality. Tree wood density is also crucial for assessing vegetation carbon stocks by supporting the link between a volumetric retrieval and a mass estimate. Earlier studies based on tree-level wood density measurements have reported significant relationships between wood density, environmental conditions, and tree growth strategies. However, these were either regionally focused or suffering from data availability, lacking a representative large-scale and spatially explicit representation of factors influencing tree wood density. This study collects and collates information from several sources to construct a global database of 28,822 tree-level wood density measurements alongside with a wide set of climate, soils, topography, and Earth observation covariates to support the development of statistical models for wood density. The dataset, consisting of more than 3,000 global covariates, is used for training different machine learning models, including random forest model (RF), light gradient boosting model (LGBM), extreme gradient boosting model (XGBoost), and bagged trees models. The experimental design considers six cross-validation approaches: either random 5-fold; according to two sets of climate classifications, land cover types, ecozones, or latitudinal ranges. Model performances are assessed with the coefficient of determination (R2) and Root-mean-square errors (RMSE) when predicting an independent test subset of wood density. The top ten models show a prominent performance (R2 > 0.67 and RMSE < 0.09), and their ensemble mean, and standard deviation are considered the best estimation and uncertainty in wood density predictions, respectively. Systematic underestimation biases are observed around the low northern latitudes (0º-20ºN), primarily due to the lack of wood density measurements. Further analysis of sources of uncertainties and their quantification support the generation of a global quantitative and spatially explicit representation of wood density. The ecological interpretation and quantitative assessment of global wood density, and associated uncertainties aim to contribute to improving predictions of vegetation biomass and inferring ecosystem resistance under current and future climate scenarios.

In a model simulating dynamics of a system, parameters can represent system sensitivities and unresolved processes, therefore affecting model accuracy and uncertainty. Taking a light use efficiency (LUE) model as an example, which is a typical approach to estimate gross primary productivity (GPP), we propose a Simultaneous Parameter Inversion and Extrapolation approach (SPIE) to overcome issues stemming from plant-functional-type(PFT)-dependent parameterizations. SPIE refers to predicting model parameters using an artificial neural network based on collected variables, including PFT, climate types, bioclimatic variables, vegetation features, atmospheric nitrogen and phosphorus deposition and soil properties. The neural network was optimized to minimize GPP errors and constrain LUE model sensitivity functions. We compared SPIE with 11 typical parameter extrapolating methods, including PFT- and climate-specific parameterizations, global and PFT-based parameter optimization, site-similarity, and regression approaches. All methods were assessed using Nash-Sutcliffe model efficiency(NSE), determination coefficient and normalized root mean squared error, and contrasted with site-specific calibrations. Ten-fold cross-validated results showed that SPIE had the best performance across sites, various temporal scales and assessing metrics. None of the approaches performed similar to site-level calibrations(NSE=0.95), but SPIE was the only approach showing positive NSE(0.68). The Shapley value, layer-wise relevance and partial dependence showed that vegetation features, bioclimatic variables, soil properties and some PFTs are determining parameters. The proposed parameter extrapolation approach overcomes strong limitations observed in many standard parameterization methods. We argue that expanding SPIE to other models overcomes current limits and serves as an entry point to investigate the robustness and generalization of different models.

Air pollution is considered one of the greatest threats to human health. This study combines a land use regression (LUR) model with satellite measurements and a distributed-lagged non-linear model (DLNM). It aims to predict high-resolution ground-level concentrations of nitrogen dioxide (NO2) and particulate matter 2.5 (PM2.5) in the Yangtze River Delta (YRD) and reveal the mechanisms of influence between NO2 and PM2.5 and precursors and meteorological factors. Results showed that the annual average NO2 and PM2.5 in the YRD urban agglomeration 2019 were 39.5 µg/m3 and 37.5 µg/m3, respectively. The seasonal variation of NO2 and PM2.5 showed winter > spring > autumn > summer. There is a compelling and complex relationship between NO2 and PM2.5. Predictors indicate that latitude (Y), surface pressure (P), ozone (O3), carbon monoxide (CO), aerosol optical depth (AOD), residential, and rangeland have positive impacts on NO2 and PM2.5. In contrast, temperature (T), precipitation (PRE), and industrial trees hurt NO2 and PM2.5. DLNM model results show that NO2 and PM2.5 had significant associations with the included precursors and meteorological elements, with lagged and non-linear effects observed. Satellite data could help significantly increase the accuracy of LUR models; the R2 of tenfold cross-validation was enhanced by 0.18–0.22. In 2019, PM2.5 will be the dominant pollutant in the YRD, and NO2 showed a high value in the central and eastern parts of the YRD. High concentrations of NO2 and PM2.5 are present in 86