A 1 km monthly dataset of historical and future climate changes over China

AdMaPlace March 18, 2025

0 Comments

A 1 km monthly dataset of historical and future climate changes over China

Background & Summary

Climate data, such as the basic climate variables of temperature, precipitation, and radiation, and further developed bioclimatic variables of seasonality and evapotranspiration, are widely used in meteorology, hydrology, agriculture, and ecology^1,2,3,4,5. In the context of rapidly changing climate, historical changes and future projections are two important parts for climate datasets to enhance data consistency and applicability^6,7. Meanwhile, climate data with high resolution are preferred for capturing subtle changes, and are further used to better understand the profound impacts of climate change on natural and social systems^8,9.

Common methods to obtain the required climate data include interpolation^10,11 and downscaling^12,13, to facilitate studies at a fine scale by converting point-based or reanalysis products into higher-resolution data. Observation records from meteorological stations are reliable sources of historical climate data. However, studies are always constrained by limited amount and uneven distribution of meteorological stations¹⁴, which highlights the necessity of interpolation. The Australian National University Spline (ANUSPLIN) is professional software to interpolate climate data based on the georeferenced algorithms called the thin plate smoothing splines (TPSS)^15,16. The TPSS algorithm can be viewed as a generalization of standard multi-variate linear regression, so that the ANUSPLIN can perform multi-surface fitting simultaneously with the introduction of different covariates¹⁶. Compared with traditional interpolation methods such as ordinary Kriging and inverse distance weighting¹⁰, TPSS, due to its superiority in capturing the impacts of topography on climate, has been widely used for spatial interpolation of various climate variables, such as temperature¹⁷, precipitation¹¹, and radiation¹⁸, and has also been used for the development of global datasets such as the CRU high-resolution gridded datasets¹⁹ and WorldClim²⁰.

Future climate data are mainly derived from general circulation models (GCMs). However, biases always exist due to scenario setting, model choice, and internal climate variability among different GCMs, which can be corrected by downscaling^13,21. Currently, statistical and dynamical downscaling methods are widely applied. The statistical downscaling assumes that the statistical relationship between large-scale and local climate variables remains stable under future climate scenarios²², while the dynamical method is based on the regional climate models with initial boundary values provided by GCMs²³. Therefore, statistical downscaling is suitable for long-term averaged climate data while saving calculating resources^13,22. Delta correction (DC), an easily operable statistical downscaling method, can be implemented by calculating the anomaly (i.e., delta change) between historical and future climate data and then applying the anomaly to the baseline data^12,24. The DC method has presented robust performance under the background of future scenario¹², historical period²⁵, and even paleoclimate reconstruction²⁶, at global¹², national²⁷, and regional²⁸ scales. Some widely used databases have also been developed based on DC method, such as the WorldClim²⁰ and Chelsa³.

China has a large territory with diverse terrain and complex climate systems, making it difficult to obtain high-resolution climate data²⁹. Also, most meteorological stations in China are established after 1950, resulting in the deficiency in long-term observation data³⁰. Therefore, research on the long-term climate change and its impacts across China is limited. Recently, some climate datasets for China have been developed, such as the monthly temperature and precipitation during 1901–2017²⁵, the precipitation variability during 1961–2015¹¹, and the evapotranspiration during 1981–2015³¹. However, only a few of them contain long-term averaged climate data including a variety of variables covering historical and future periods.

In this study, a bias-corrected 30-year averaged monthly climate dataset was developed at 0.01° × 0.01° (≈1 km × 1 km) grid using ANUSPLIN software and the DC method, based on the latest observation records from meteorological stations in China and GCMs. The dataset includes 28 commonly used climate variables during the historical period of 1991–2020 and future periods of 2021–2040, 2041–2070 and 2071–2100. It was evaluated using the interpolated and downscaled outputs, to ensure data reliability and provide a high-quality long-term averaged climate dataset for climate change impact research.

Methods

Data acquisition

Historical observation dataset

The 30-year averaged surface standard climate dataset for 1991–2020^32,33, the baseline year in this study, was recently released based on the observed records of 2185 meteorological stations (Fig. 1). The dataset provided six types of climate variables including temperature, precipitation, humidity, air pressure, wind, and radiation. We selected five basic variables (Table 1), including monthly precipitation (pr), mean air temperature (tas), minimum air temperature (tasmin), maximum air temperature (tasmax) and mean percentage of sunshine (sunp). These variables, along with the bioclimatic variables developed based on them, are commonly used in the research of climatology and ecology^1,3,7,20. All the 2185 stations provided records of the variables tas, tasmax, tasmin, and pr, and 2093 stations with the records of sunp (Fig. 1). The generation of 30-year averaged monthly climate data followed the guidance of China Meteorological Administration³⁴ and World Meteorological Organization³⁵, to ensure the quality of this observation dataset in compliance with national and international standards.

Table 1 Climate variables provided in this dataset.

Full size table

GCM selection

The Coupled Model Intercomparison Project Phase 6 (CMIP6), overseen by the World Climate Research Programme’s Working Group on Coupled Modelling, is a major international multi-model research activity coordinating historical and future climate data from different GCMs³⁶. Among its various climate products, the historical simulations for 1850–2014 are commonly used. It also provides climate projections for 2015–2100, based on different scenarios of future emissions and land use changes, namely, the Shared Socioeconomic Pathways (SSPs)³⁷.

We used historical simulations (1951–1980, 1991–2014) and future projections (2021–2040, 2041–2070, 2071–2100) from nine GCMs under three scenarios, namely, SSP1-2.6, SSP2-4.5, and SSP5-8.5 (Table 2). To maintain consistency with variables in our historical observation dataset, these GCMs also included monthly data of tas, tasmax, tasmin, pr, and the total cloud cover (clt, i.e., 100 − sunp, for convenience of calculation, following Wang et al.³⁸). This partly restricted our selection of GCMs, as only a limited number of GCMs provided clt data. We further considered the GCMs selected by widely used databases developed based on CMIP6 data, such as the WorldClim²⁰ (https://worldclim.org) and Chelsa³⁹ (https://chelsa-climate.org). As for the SSPs, SSP1-2.6, SSP2-4.5, and SSP5-8.5, the continuations of the RCP2.6, RCP4.5, and RCP8.5 forcing levels in CMIP5 representing low, medium, and high forcing categories, indicated optimistic, moderate, and pessimistic future emission scenarios, respectively^37,40. All the CMIP6 data were downloaded from the data search interface (https://aims2.llnl.gov/search/cmip6/).

Table 2 General circulation models (GCMs) selected in this dataset.

Full size table

Generation of high-resolution historical observed data

Due to the coarse spatial distribution of meteorological stations, interpolation was first required to convert the historical observation dataset during 1991–2020 at geographic coordinates into high-resolution surface data. The interpolation software ANUSPLIN v.4.4 was applied to generate 12 monthly surfaces at a resolution of 0.01° × 0.01° (≈1 km × 1 km) for each variable. In this study, we set longitude and latitude as independent variables, and used elevation as a third independent variable for tas, tasmax, tasmin, and pr to perform full tri-variate spline function. The elevation data were derived from the Shuttle Radar Topography Mission digital elevation model⁴¹, whose original resolution of 90 m was resampled to 0.01° using the bilinear method in this study. For sunp, monthly pr data were used as surface independent covariate, as incorporating the dependence on rainfall allowed for a better representation of complex solar radiation patterns^16,42.

Based on the five interpolated basic variables, 23 bioclimatic variables were further calculated (Table 1). Bio1 to Bio19 were developed to describe specific bioclimatic conditions and thus widely applied in the study of biology and ecology^43,44,45. The variables tas, tasmax, tasmin, and pr were required for the calculation of these 19 bioclimatic variables. Unlike Worldclim, which uses the average of tasmax and tasmin to represent monthly mean temperature⁴⁴, we instead applied updated equations to use tas directly, following Karger et al.³. Bio20 to Bio23 were calculated using tas, pr, and sunp as input into the modified bioclimatic envelope model^38,46 developed from a simple process-based bioclimatic model (STASH)⁴⁷. The general algorithm of STASH^46,47 was to convert monthly data into daily data and then calculate the total annual growing degree days (GDD, i.e., Bio20 and Bio21) and annual potential evapotranspiration (PET, i.e., Bio22) based on the modified Penman-Monteith method⁴⁸. Additionally, the moisture index (MI, i.e., Bio23), was calculated according to the United Nations Environment Programme to indicate dryland types⁴⁹.

GCM bias-correction and downscaling

The DC downscaling method was employed here to perform bias-correction and provide high-resolution, multi-year, monthly averaged climate data. The DC method included the following steps (Fig. 2):

First, we calculated the 30-year averages of tas, tasmax, tasmin, pr, and sunp for the historical simulations (1991–2014) and six future scenarios. Because the CMIP6 historical data ended in 2014, some studies, such as the CMIP6-Endorsed activity⁵⁰, extended the dataset with data under SSP2-4.5, the scenario closest to reality. In this study, historical simulations during 1991–2014 were retained for subsequent calculations, as the data showed a higher correlation with observation data (1991–2020) according to pre-simulation results (Fig. S1).

Second, the monthly anomalies (i.e., delta changes) between historical simulations and future scenarios were calculated. Specifically, the anomaly was calculated as absolute difference for temperature-related variables (Eq. 1) and as relative difference for sunp (Eq. 2) and pr (Eq. 3).

$$triangle {T}_{i}={T}_{{fi}}-{T}_{{hi}}$$

(1)

$$triangle {S}_{i}=frac{{S}_{{fi}}}{{S}_{{hi}}}$$

(2)

$$triangle {P}_{i}=frac{{P}_{{fi}}+0.1}{{P}_{{hi}}+0.1}$$

(3)

where, ∆ is the delta change; T is the temperature variables including tas, tasmax, and tasmin; S is variable sunp; P is variable pr; f is the multi-year averaged climate data under future scenarios; h is the averaged historical simulations during 1991–2014; and i is each month from January to December.

Notably, the use of relative difference for pr and sunp was to avoid negative values when applying the delta change to observed data. Considering the scanty rainfall, which is even close to zero in some areas of China to result in excessively large anomaly values, two extra modifications were applied to the delta change in precipitation^3,12: (1) based on the results of the prior sensitivity analysis testing a series of constants and comparison with other studies^3,12, it was determined that a constant of 0.1 was added to both the historical and future data to avoid division by zero (Eq. 3); (2) the top 2% of anomaly values was truncated to the 98^th percentile value in the empirical probability distribution to avoid potentially unreasonable future precipitation extremes.

Third, those monthly anomalies were resampled to the same 0.01° × 0.01° resolution as the historical observed data. The original resolutions of selected GCMs were different but unexceptionally coarse (Table 2). To obtain uniform and high-resolution data, another three commonly used resampling methods were employed including bilinear, cubic spline, and thin-plate spline, in order of calculation from simple to complex. Given the better and less time-consuming performance of the cubic spline method (Table S1), a total of 3240 resampled surfaces of monthly delta change were produced for all the GCMs, SSPs, and variables.

Finally, the downscaled future climate data were obtained by applying resampled anomalies to high-resolution historical observed data. Addition was applied to temperature variables, while multiplication was used for other variables (Eqs. 4, 5). The ensemble model was derived as the equally weighted average of the nine GCMs. The bioclimatic variables were further derived from the downscaled future basic climate variables following the aforementioned methods.

$${T}_{{Dfi}}={T}_{{Ohi}}+triangle {T}_{{Ii}}$$

(4)

$${{SP}}_{{Dfi}}={{SP}}_{{Ohi}}times triangle {{SP}}_{{Ii}}$$

(5)

where, ∆ is the delta change; T is temperature variables including tas, tasmax, and tasmin; SP is variables including sunp and pr; Df is the downscaled future climate data; Oh is the high-resolution historical observed data during 1991–2020; I is resampled data; and i is each month from January to December.

Accuracy evaluation

The software ANUSPLIN allowed detection of errors between interpolated and observed data through generalized cross validation to evaluate interpolation accuracy. The root mean square error (RMSE; Eq. 6), mean absolute error (MAE; Eq. 7), and coefficient of determination (R²; Eq. 8) were selected as evaluation criteria in this study²⁸. Lower RMSE and MAE values meant smaller deviation between simulated and observed data, while an R² value close to 1 indicated a better model fit.

$${RMSE}=sqrt{frac{1}{n}mathop{sum }limits_{i=1}^{n}{left({{SIM}}_{i}-{{OBS}}_{i}right)}^{2}}$$

(6)

$${MAE}=frac{1}{n}mathop{sum }limits_{i=1}^{n}left|{{SIM}}_{i}-{{OBS}}_{i}right|$$

(7)

$${R}^{2}=frac{mathop{sum }limits_{i=1}^{n}{left({{SIM}}_{i}-{{OBS}}_{i}right)}^{2}}{mathop{sum }limits_{i=1}^{n}{({{OBS}}_{i}-bar{{{OBS}}_{i}})}^{2}}$$

(8)

where, SIM_i is the interpolated or simulated value at position i, OBS_i the observed value at position i, and n the number of samples.

For the downscaled data, probability density functions (PDFs) were used to compare the seasonal trends of observed and downscaled data in different regions of China. PDFs could clearly illustrate the impacts of downscaling methods¹². Specifically, the study area was divided into four regions (Fig. 1): Northeast (NE), Southeast (SE), Northwest (NW), and the Tibetan Plateau (TP). The annual period was further divided into four seasons, each consisting of three months: spring (March-April-May), summer (June-July-August), autumn (September-October-November), and winter (December-January-February). We selected GCM data under SSP5-8.5 in 2071–2100 to generate PDFs, as this scenario deviated the most from the current climate and thus improved the visualization. The Taylor diagram⁵¹ was also used to evaluate GCM performance more concisely. Since no future climate observation exists, some studies have chosen a period with proxy data as an independent experiment^12,52. Here, we used climate data during 1951–1980⁵³ as a reference period to further analyze the downscaling methods. The observed records during 1951–1980 were derived from the same source as the baseline data during 1991–2020 to ensure the representativeness of the reference data. The GCM-simulated historical data during 1951–1980 were downscaled following the aforementioned methods. The observed and downscaled data were then compared. Additionally, the three anomaly resampling methods was also validated using the reference data.

Data Records

The dataset includes averaged monthly gridded (0.01° × 0.01°) historical data during 1991–2020 and bias-corrected future data over ten models (nine GCMs and one ensemble model), three SSPs, and three periods: 2021–2040, 2041–2070 and 2071–2100. For the history period and each future scenario, five basic climate variables were calculated including tas, tasmax, tasmin, pr, and sunp, in addition to 23 bioclimatic variables (Table 1).

All data are provided as zipped GeoTIFF (.tif) files. The file naming rules followed a consistent pattern. The historical data were named following the example China_Variable_1km_1991–2020.tif (e.g. China_pr_1km_1991–2020), where Variable is the abbreviation of 28 variables. Similarly, the future data were named following the example China_Variable_Model_VariantLabel_1km_StartYear-EndYear_Scenario.tif, (e.g. China_tasmin_MRI-ESM2-0_r1i1p1f1_1km_2071–2100_SSP585.tif), where Variable is the abbreviation of 28 variables, Model is the GCM name, VariantLabel is r1i1p1f1 as default in this study, StartYear-EndYear is the future period, and Scenario is the SSP name. For the five basic variables, tas, tasmax, tasmin, pr, and sunp, each file contained 12 layers representing 12 months in ascending order. All data are freely available at the China Science Data Bank⁵⁴.

Technical Validation

ANUSPLIN performance

The ANUSPLIN outputs of historical climate during 1991–2020 showed remained similar trends to observation records in different regions (Fig. 3), with the characteristics of simultaneous rain and heat, and gradual decreases in temperature and precipitation from southeast to northwest with the distance from the ocean^25,55, in contrast to the fluctuation of percentage of sunshine. Specifically, in July, the tas across China ranged from −8.50 °C to 34.26 °C, tasmax from 2.53 °C to 49.83 °C, tasmin from −22.81 °C to 26.65 °C, pr from 1.15 mm to 635.30 mm, and sunp from 19.32% to 85.14%. In contrast, in January, the tas across China ranged from −31.89 °C to 21.09 °C, tasmax from −7.99 °C to 35.28 °C, tasmin from −58.54 °C to 18.42 °C, pr from 0.38 mm to 234.96 mm, and sunp from 7.70% to 85.14%. Except for sunp, the averages of ANUSPLIN were slightly lower than those of observation records. This is caused by the sparse meteorological stations in the relatively cold and dry southwestern region of China, indicating that ANUSPLIN could modify the distribution bias of these stations¹¹.

The criteria of RMSE, MAE, and R² were commonly used to evaluate interpolation quality accurately^10,11,53. Generally, the evaluation results differed due to the seasonal and regional variations in observed data but remained overall stable, indicating that the ANUSPLIN surfaces were reliable (Table 3). The RMSE in the monthly series for tas was below 0.38 °C, tasmax below 0.78 °C, tasmin below 1.45 °C, pr below 12.77 mm, and sunp below 2.57%. The MAE for tas was below 0.28 °C, tasmax below 0.54 °C, tasmin below 1.07 °C, pr below 8.14 mm, and sunp below 1.78%. Additionally, R² for all the five variables was above 0.91. Among the five variables, the evaluation results of tas were optimal and the most stable, while the results for pr and sunp fluctuated mainly in summer. On the one hand, this may be related to the choice of covariates. The influence of elevation was more direct on temperature than precipitation and precipitation-based calculation of percentage of sunshine following the temperature lapse rate^10,16. On the other hand, the interpolation for areas with higher temporal and spatial variability was sensitive to the data processing method of meteorological stations, leading to increasing difficulty in the accurate depiction of precipitation distribution¹¹. Spatially (Table S2), the RMSE for tas was below 0.36 °C (NE), tasmax below 0.63 °C (NE), tasmin below 1.27 °C (NE), pr below 12.03 mm (SE), and sunp below 2.54% (TP). The MAE for tas was below 0.26 °C (NE), tasmax below 0.45 °C (NE), tasmin below 0.94 °C (NE), pr below 6.02 mm (SE), and sunp below 1.83% (TP). R² for all the regions was above 0.92. Although meteorological stations were sparse in the TP, the interpolation results performed better than expected (Table S2). This also demonstrated that ANUSPLIN could remain effective in areas with sparse stations^11,14.

Table 3 The evaluation of ANUSPLIN interpolated results in the monthly series.

Full size table

Delta correction performance

The DC method greatly changed the regional and seasonal distribution of all basic variables (Figs. 4, 5; Figs. S2–S4). After downscaling, PDFs had similar shapes, averages and standard deviations as those of interpolated historical data. All the downscaled data successfully preserved the anomaly (i.e., delta change) between simulated historical and future data. For most regions and seasons, the distance between the PDFs of downscaled and interpolated data closely matched the gap between the historical and future GCM data. The PDFs of GCMs were mostly unimodal, while the DC method modified peak distribution and adjusted density values related to temperature frequency and precipitation intensity, such as the tas (SE) during the winter and pr (NW) during the spring, making PDF shapes closer to observations and thus improve model accuracy. For extreme variables, GCMs tended to overestimate tasmin and underestimate tasmax, which was effectively corrected by the DC method.

From the near optimistic (SSP1-2.6 in 2021–2040) to distant pessimistic (SSP5-8.5 in 2071–2100) future (Figs. 6, 7; Figs. S5–S7), all basic variables exhibited a gradually intensified increase in general, in accordance with the definition of their respective climate scenarios. The increase in average annual tas over China ranged from 0.64 °C to 5.83 °C, tasmax from 0.53 °C to 5.70 °C, and tasmin from 0.64 °C to 5.83 °C. And the change of average annual pr over China varied from −4.35% to 54.34% and sunp from −0.55% to 19.90%. Spatially, climate warming intensified from southeastern to northwestern China. Meanwhile, the increase in precipitation was concentrated in the southeast, contrary to the decrease happening in the northwest. As for sunp, its increase mainly occurred in the Tibetan Plateau.

Similar to the evaluation of ANUSPLIN results, RMSE, MAE, and R² were also used to validate the DC method, as recommended by many DC datasets^25,28,56. The Taylor diagram was also a common tool to compare GCMs^28,52,57. Additionally, we also used historical data during 1951–1980 as reference. Therefore, the evaluation of the DC method in this study was also reasonable. For reference data (Fig. 8), downscaled data had higher fitness with historical observations. The average RMSE of downscaled data across nine GCMs for tas, tasmin, pr, and sunp reduced by 58.63%, 0.69%, 53.35%, and 58.44%, respectively. The average correlation coefficient increased by 3.51%, 2.13%, 26.32%, and 71.35%, respectively. However, for tasmax, the average of RMSE increased by 7.96% and the average of correlation coefficient decreased by 1.56%. Overall, all correlation coefficients were above 0.92 after downscaling, indicating that the DC method could improve the resolution and accuracy of CMIP6 data. Among the basic variables, the downscaling results of temperature performed better (Fig. 8). Usually, the downscaling of temperature was more operable²⁸, partly because of the inherently greater divergence among different GCMs for original precipitation data than temperature (Fig. 8; Fig. 9). Also, it was more constrained to fully explain precipitation, like the bias of AUSPLIN results abovementioned (Table 3). Among the nine GCMs, the models that performed the best for tas, tasmax, tasmin, pr, and sunp before downscaling were CMCC-ESM2, GFDL-ESM4, MPI-ESM1.2-HR, MPI-ESM1.2-HR, MIROC6. After downscaling, the better models were BCC-CSM2-MR, MIROC6, BCC-CSM2-MR, INM-CM5.0, MIROC6, respectively. Among GCMs, downscaled data presented higher stability in the average and variance (Fig. 9).

Additionally, note that bias correction could not calibrate fundamentally a GCM performing poorly²⁶, such as the INM-CM5-0 model for sunp projections (Fig. 8). In this case, the ensemble model was a solution to further correct biases among GCMs²⁷. In this study, we used the arithmetic mean of GCMs as an example. The ensemble model further promoted the performance of DC method. After downscaling, the RMSE for tas, tasmax, tasmin, pr, and sunp decreased by 3.82%, 0.56%, 0.62%, 11.68%, and 19.73%, and the correlation coefficient increased by 0.06%, 0.12%, 0.04%, 1.25%, and 2.62%, respectively. The ensemble model performed better than each GCM for tas, pr, and sunp, and preceded only by the optimal model abovementioned for the rest variables. The ensemble method used here was the simplest approach, and users could further calculate customized ensemble models to satisfy their own demands.

Comparison with previous studies and other products

To compare our dataset with other climate products and previous studies, we extracted monthly averaged data for 1991–2020 from other datasets (Table S3) and analyzed the relationships between these datasets and meteorological records (Table S4). The results showed that our dataset provided more variables which exhibited a better fit with the meteorological records than other datasets. Overall, the gridded climate data interpolated based on station records performed better than those based on multi-source and reanalyzed climate products², especially for the extreme variables and percentage of sunshine (Table S4). Among these datasets, our dataset was established based on more meteorological stations. This probably also resulted in its higher accuracy. In the future, the overall projections would increase all the five basic climate variables in most areas. Similar trends and spatial patterns were also observed in previous predictions based on CMIP6 data^58,59,60,61. For example, it was predicted that the increase in precipitation would reach 57.44% at most in the northwestern China in the 2090 s under SSP5-8.5⁶¹, which was similar to the above 50% future increase occurring in the same region in this study (Fig. 7). However, these datasets differed in values due to the choices of scenarios and GCMs. Therefore, further uncertainty analysis is necessary to determine the suitability of data under different situations before application.

Uncertainty

As validated in this study, ANUSPLIN demonstrated good interpolation performance, by accounting for spatial variations such as elevation and relying on all observed data for modeling, even in data sparse, high-elevation regions⁶². However, it was difficult to fully explain local conditions in areas with complex climate depending on only the interpolation software, particularly for microclimate phenomena such as temperature inversions^62,63. In this case, the most effective solution was supplementing local observation records, and it could also enhance interpolation accuracy by incorporating climate information from neighboring areas¹⁷. Therefore, to enhance the accuracy of ANUSPLIN outputs, the improvement of data sources was also a feasible choice.

Compared with CMIP5, CMIP6 data have improved simulations of temperature and precipitation, particularly for climate extremes⁶⁴. However, the downscaling results of the two extreme variables tasmax and tasmin were still partly biased. Compared with the observed average, the median of downscaled tasmax increased by about 11 °C, while the downscaled average of tasmin decreased by about 10 °C (Fig. 9). The prediction of extreme events tended to be uncertain. Although correct changing trends could be retained by DC⁶⁵, some studies showed underestimation of minimum and overestimation of maximum when applying DC method, which was possibly caused by relatively higher sensitivity of extreme variables to the choice of GCMs and downscaling techniques, as well as fluctuation of climate changing signals in data processing^26,52,66,67. This may be the reason why the improvement in the two extreme variables, tasmax in particular, after downscaling was not as obvious as that of other variables. However, considering that the evaluation results of the basic climate variables were close to each other and confident statistically, and that the influences of extreme variables were inevitably partly eliminated after 30-year averaged data processing¹², we believed that the data were reliable and they should be used more cautiously when analyzing the impacts of extreme temperature on defined period and location.

Except the variable differences, the methods of anomaly resampling could also play a part in the DC performance. The methods used in this study included bilinear, cubic spline, and thin-plate spline, which were all commonly used and helped promote DC accuracy^3,12,25. The final choice of cubic spline was based on its slight advantage in statistical analysis, while the differences among the three methods were minimal (Table S1). This is because, as least in this study, the resampling method was not the main factor for DC. Additionally, the DC performance exhibited spatial variability during the same period, which may be influenced by the accuracy of the baseline data^25,26. For instance, three covariates were introduced in the WorldClim dataset, including elevation, distance to the coast and three satellite-derived covariates, to improve data accuracy²⁰. On this basis, the DC downscaled data had also better accuracy²⁵. In this study, the baseline data of interpolated grid during 1991–2020 incorporated abundant stations and their records from official source, thus ensuring the accuracy of downscaled future data.

Usage Notes

Based on ANUSPLIN software and DC downscaling method, our dataset contained five basic climate variables and 23 bioclimatic variables. Each variable covered the historical period and six future climate scenarios, and thus had a wide range of applicability. Firstly, these variables could be used to study the relationships between climate and species, populations, communities, and ecosystems. Data from multiple periods of each variable could also be used to study the impact of climate change on species distribution, ecosystem patterns and functions. Secondly, considering the current data storage costs, we provide only 28 commonly used variables. However, this did not mean that these data were limited to the variables. Instead, users could still further calculate other customized variables such as actual evapotranspiration and the plant-available moisture index^38,68 based on the dataset. Thirdly, as emphasized above, regardless of the dataset used, users should consider the potential biases and determine whether these biases would affect their research. Specifically, the long-term averaged monthly climate data we provided in this study should be applied more carefully when analyzing extreme climate change and annual variation. In the future, more observation data will be collected, as this remains the most effective and direct method to improve data accuracy. Moreover, with the continuous progress of GCMs, we also believe that GCM outputs will achieve higher precision and accuracy, thereby providing better climate data for various fields.

Categories: Climate and Earth system modelling, Projection and prediction

Energy metabolism in health and diseases

Energy metabolism is indispensable for sustaining physiological functions in living organisms and assumes a pivotal role across physiological and pathological conditions. This review provides an extensive overview of advancements in energy metabolism research, elucidating critical pathways such as glycolysis, oxidative phosphorylation, fatty acid metabolism, and amino acid metabolism, along with their intricate regulatory mechanisms. The homeostatic balance of these processes is crucial; however, in pathological states such as neurodegenerative diseases, autoimmune disorders, and cancer, extensive metabolic reprogramming occurs, resulting in impaired glucose metabolism and mitochondrial dysfunction, which accelerate disease progression. Recent investigations into key regulatory pathways, including mechanistic target of rapamycin, sirtuins, and adenosine monophosphate-activated protein kinase, have considerably deepened our understanding of metabolic dysregulation and opened new avenues for therapeutic innovation. Emerging technologies, such as fluorescent probes, nano-biomaterials, and metabolomic analyses, promise substantial improvements in diagnostic precision. This review critically examines recent advancements and ongoing challenges in metabolism research, emphasizing its potential for precision diagnostics and personalized therapeutic interventions. Future studies should prioritize unraveling the regulatory mechanisms of energy metabolism and the dynamics of intercellular energy interactions. Integrating cutting-edge gene-editing technologies and multi-omics approaches, the development of multi-target pharmaceuticals in synergy with existing therapies such as immunotherapy and dietary interventions could enhance therapeutic efficacy. Personalized metabolic analysis is indispensable for crafting tailored treatment protocols, ultimately providing more accurate medical solutions for patients. This review aims to deepen the understanding and improve the application of energy metabolism to drive innovative diagnostic and therapeutic strategies.

AdMaPlace March 19, 2025

0 Comments

Tissue macrophages: origin, heterogenity, biological functions, diseases and therapeutic targets

Macrophages are immune cells belonging to the mononuclear phagocyte system. They play crucial roles in immune defense, surveillance, and homeostasis. This review systematically discusses the types of hematopoietic progenitors that give rise to macrophages, including primitive hematopoietic progenitors, erythro-myeloid progenitors, and hematopoietic stem cells. These progenitors have distinct genetic backgrounds and developmental processes. Accordingly, macrophages exhibit complex and diverse functions in the body, including phagocytosis and clearance of cellular debris, antigen presentation, and immune response, regulation of inflammation and cytokine production, tissue remodeling and repair, and multi-level regulatory signaling pathways/crosstalk involved in homeostasis and physiology. Besides, tumor-associated macrophages are a key component of the TME, exhibiting both anti-tumor and pro-tumor properties. Furthermore, the functional status of macrophages is closely linked to the development of various diseases, including cancer, autoimmune disorders, cardiovascular disease, neurodegenerative diseases, metabolic conditions, and trauma. Targeting macrophages has emerged as a promising therapeutic strategy in these contexts. Clinical trials of macrophage-based targeted drugs, macrophage-based immunotherapies, and nanoparticle-based therapy were comprehensively summarized. Potential challenges and future directions in targeting macrophages have also been discussed. Overall, our review highlights the significance of this versatile immune cell in human health and disease, which is expected to inform future research and clinical practice.

AdMaPlace March 19, 2025

0 Comments

The coupled effect of climate change and LUCC on meteorological drought in a karst drainage basin, Southwest China

With global warming and the acceleration of the water cycle, the frequency and severity of droughts have progressively increased. Although Southwest China is located inland, the combined effects of global climate change, regional climate anomalies, and human activities have led to a diversified set of driving mechanisms for meteorological droughts. Based on monthly global and regional meteorological factors (10 global factors and 8 regional factors) and land use/land cover data from 1948 to 2023, this study employs classical correlation analysis, wavelet analysis, and Bayesian principles to explore the temporal and spatial evolution characteristics of meteorological droughts in Southwest China, as well as their driving mechanisms. The results show that: (1) between 1948 and 2023, meteorological droughts in Southwest China exhibited a north-south (annual average, spring, and winter) or east-west decreasing trend (autumn) and alternating east-west intensity in summer. The drought frequency ranged from 0.35 to 0.39, generally showing a decreasing trend from southwest to northeast. (2) Global atmospheric circulation significantly influences meteorological droughts in Southwest China, especially during El Niño years, when global atmospheric circulation factors such as Nino 3.4 and SOI have a more prominent impact on drought, particularly in the summer and autumn. In contrast, during La Niña years, drought intensity and frequency are more pronounced in spring and autumn. Regional climate factors, particularly temperature and evapotranspiration, also have a significant effect on drought across different seasons, especially in winter and spring, when higher temperatures and evapotranspiration exacerbate droughts, and precipitation has a relatively weaker effect. (3) Land use changes significantly affect meteorological droughts in Southwest China, with a driving probability ranging from 0.39 to 0.42. Under different climate conditions, the driving probabilities of land use changes are ranked as follows: El Niño years (0.32–0.52) > Normal years (0.31–0.51) > La Niña years (0.27–0.50). In particular, land use changes such as the expansion of built-up areas and the reduction of farmland could intensify the frequency and severity of meteorological droughts in drought-prone areas. Additionally, the reduction of green spaces or forests may also exacerbate droughts, especially during the urbanization process. Notably, during El Niño years, the driving effect of various landforms on drought shows different skewed distributions. Therefore, this study clarifies, to some extent, the evolution and mechanisms of meteorological droughts in Southwest China and provides technical guidance and theoretical support for drought prevention and disaster relief efforts in the Karst region.

AdMaPlace March 19, 2025

0 Comments

Advance in peptide-based drug development: delivery platforms, therapeutics and vaccines

The successful approval of peptide-based drugs can be attributed to a collaborative effort across multiple disciplines. The integration of novel drug design and synthesis techniques, display library technology, delivery systems, bioengineering advancements, and artificial intelligence have significantly expedited the development of groundbreaking peptide-based drugs, effectively addressing the obstacles associated with their character, such as the rapid clearance and degradation, necessitating subcutaneous injection leading to increasing patient discomfort, and ultimately advancing translational research efforts. Peptides are presently employed in the management and diagnosis of a diverse array of medical conditions, such as diabetes mellitus, weight loss, oncology, and rare diseases, and are additionally garnering interest in facilitating targeted drug delivery platforms and the advancement of peptide-based vaccines. This paper provides an overview of the present market and clinical trial progress of peptide-based therapeutics, delivery platforms, and vaccines. It examines the key areas of research in peptide-based drug development through a literature analysis and emphasizes the structural modification principles of peptide-based drugs, as well as the recent advancements in screening, design, and delivery technologies. The accelerated advancement in the development of novel peptide-based therapeutics, including peptide-drug complexes, new peptide-based vaccines, and innovative peptide-based diagnostic reagents, has the potential to promote the era of precise customization of disease therapeutic schedule.

AdMaPlace March 19, 2025

0 Comments

Probabilistic machine learning for battery health diagnostics and prognostics—review and perspectives

Diagnosing lithium-ion battery health and predicting future degradation is essential for driving design improvements in the laboratory and ensuring safe and reliable operation over a product’s expected lifetime. However, accurate battery health diagnostics and prognostics is challenging due to the unavoidable influence of cell-to-cell manufacturing variability and time-varying operating circumstances experienced in the field. Machine learning approaches informed by simulation, experiment, and field data show enormous promise to predict the evolution of battery health with use; however, until recently, the research community has focused on deterministic modeling methods, largely ignoring the cell-to-cell performance and aging variability inherent to all batteries. To truly make informed decisions regarding battery design in the lab or control strategies for the field, it is critical to characterize the uncertainty in a model’s predictions. After providing an overview of lithium-ion battery degradation, this paper reviews the current state-of-the-art probabilistic machine learning models for health diagnostics and prognostics. Details of the various methods, their advantages, and limitations are discussed in detail with a primary focus on probabilistic machine learning and uncertainty quantification. Last, future trends and opportunities for research and development are discussed.

AdMaPlace March 17, 2025

0 Comments

Background & Summary

Methods

Data acquisition

Historical observation dataset

GCM selection

Generation of high-resolution historical observed data

GCM bias-correction and downscaling

Accuracy evaluation

Data Records

Technical Validation

ANUSPLIN performance

Delta correction performance

Comparison with previous studies and other products

Uncertainty

Usage Notes

Related Articles

Responses