Impacts of extreme climate change from 2000 to 2022 on net primary productivity in the Gaoligong Mountains Authors: Tiantian Yang, Wanxiong Zhang, Liuwen Yang, Zifei Gao, Jingjing Wang, Zhuoya Zhang Published: 2026 ABSTRACT AbstractGlobal climate change has increased the frequency and intensity of extreme weather events, significantly impacting the net primary productivity (NPP) of vegetation. Understanding the relationship between NPP and extreme climate events in ecologically sensitive areas is essential for effective ecological strategies. This study analyzed the spatiotemporal distribution characteristics of net primary productivity (NPP) from 2000 to 2022 and its response to extreme climate conditions. Utilizing the flexible space-temporal DAta fusion (FSDAF), the study integrated MODIS and Landsat data from 2000 to 2022 to generate a high-resolution NDVI dataset (30 m, 16-day). The NPP was estimated using the Carnegie-Ames-Stanford approach (CASA) model. We also evaluated the effects of 13 extreme climate indices (ECIs) on NPP in the Gaoligong Mountains. The results showed that (1) annual NPP exhibited an upward trend (slope = 1.5), with the most significant increase occuring in January (slope = 0.35, p < 0.001); (2) the climate in the study area has displayed a clear warming trend, with significant increases in extreme temperature indices (TXx, TNx, TN90p, TX90p, TMAXmean, and TMINmean, p < 0.001), while extreme precipitation indices (RX1day, RX5day), showd a relatively small trend of change and not significant; (3) At the seasonal scale, the responses of NPP to ECIs varied significantly among different vegetation types. The correlations between NPP and ECIs were markedly stronger in spring and autumn than in summer and winter, with temperature-related indices showing the strongest explanatory power for variations in NPP. (4)The response of NPP to extreme temperatures and precipitation is primarily characterized by a lag effect, typically delayed by 1–2 months, and is observed across different vegetation types. (5) extreme temperatures, particularly TX90p, TXx, TMAXmean, and DTR, are the key climatic factors affecting NPP. These results offer insights into the impact of climate extremes on NPP, which can inform future ecological management strategies.Keywords: Gaoligong Mountains, Extreme climate indices (ECIs), Net primary productivity (NPP), FSDAFSubject terms: Climate sciences, Ecology, Ecology, Environmental sciences Introduction As global climate change intensifies, The increased frequency, intensity, and duration of extreme climate events are among the key characteristics of global change and represent some of the most severe challenges to regional sustainability1,2. The carbon cycle of terrestrial ecosystems is a fundamental process driving ecosystem dynamics and plays an irreplaceable role in mitigating increases in atmospheric CO2 concentrations and global temperature rise3. Extreme climate events, such as extreme temperatures, precipitation, and droughts, significantly influence the structure, function, and composition of terrestrial ecosystems, thereby altering the carbon cycle and its feedback mechanisms within the climate system4. Given the limited adaptive capacity of terrestrial ecosystems in the short term, the suddenness and extremity of extreme climate events often exert disproportionately severe impacts on plants and plant communities compared to gradual climate variations, Such events can lead to ecosystem collapse, dramatic reductions in biodiversity, and the loss of ecosystem services5,6. Net primary productivity (NPP) is a critical indicator of terrestrial ecosystem productivity and carbon sequestration capacity, functioning as a key regulator of the ecological environment7. Currents studies of NPP largely rely on MODIS remote sensing imagery8,9, While this approach effectively captures temporal variations of NPP, it lacks sufficient spatial detail, particularly in regions characterized by complex terrain and landscape fragmentation10,11. High spatiotemporal resolution remote sensing imagery is essential for accurately estimation of NPP, especially in analyzing spatial variations and lag effects associated with extreme climate events. However, obtaining high-resolution data is often constrained by high costs and technical complexity12,13. Multi-source remote sensing data fusion techniques provide a cost-effective solution to this challenge14,15. Among these methods, the Flexible Space-temporal DAta Fusion (FSDAF)16 has demonstrated exceptional performance in integrating normalized difference vegetation index (NDVI) data from multiple sources, offering valuable support for generating high-resolution NPP datasets12,17. As a key component of terrestrial ecosystems, vegetation productivity is highly sensitive to extreme climate events18,19. While existing studies primarily focus on temperature, precipitation, and solar radiation20,21, the impacts of extreme climate events on vegetation remain underexplored. With the increasing frequency of global extreme climate events, their effects on vegetation growth and carbon dynamics have become more pronounced22. For example, Under extreme heat stress, photosynthesis in vegetation may weaken or stagnate, while increased respiration surpasses reductions in NPP, resulting in diminished ecosystem carbon sequestration capacity23. Furthermore, the synergistic effects of extreme heat and drought can exacerbate carbon losses24. Although extreme cold events have decreased global warming25, frost events occurring early or late in the growing season can partially offset the benefits of a longer growing seasons on NPP26. In arid regions, increased soil moisture generally enhances ecosystem productivity and carbon accumulation, while extreme precipitation events in humid regions are detrimental to ecosystem productivity and carbon sequestration27,28. Extreme climate events often have short-term impacts on vegetation, and when such events exceed vegetation tolerance thresholds, lag effects may occur. However, there is ongoing debate regarding the lag effects of NPP across different vegetation types in response to extreme climate events29, potentially due to a lack of high spatiotemporal resolution data. Consequently, exploring the spatiotemporal response of NPP to extreme climate events using high-resolution data is a critical scientific endeavor29,30. To address these issues, this study focuses on the Gaoligong Mountains as the research area, utilizing NDVI data from MODIS and Landsat (2000–2022) and integrating them via the FSDAF model to generate an NDVI dataset with high spatiotemporal resolution (30 m, 16 days). The CASA model was used to simulate high-resolution NPP data, which were combined with 11 extreme temperature indices and 2 extreme precipitation indices recommended by the Expert Team on climate change detection and indices (ETCCDI) (http://etccdi.pacificclimate.org/index.shtml). This study investigates the spatiotemporal response of NPP to extreme climate indices (ECIs) aims to address three key questions: (1) What are the spatiotemporal characteristics of NPP in the Goolagong Mountains from 2000 to 2022? (2) What are the trends in ECIs? (3) How does extreme climate change influence NPP? The findings of this study contribute to deepening our understanding of terrestrial ecosystem response mechanisms in the context of global change. They also hold significant importance for elucidating the response mechanisms linking vegetation productivity to extreme climate events, and provide valuable reference for regional ecological and environmental conservation. Data and methodology Data Study area Gaoligong Mountains is located on the border between China and Myanmar, in the southern section of the Hengduan Mountains (Fig. 1). Its topography is characterized by higher elevations in the north and lower elevations in the south. The mountain range connects to the Tibetan Plateau in the north, adjoins the India-Myanmar Peninsula in the south, lies to the east of the Nu Mountains within the Hengduan Mountain range, and borders the India-Myanmar mountains to the west. This range extends in a north-south direction, spanning five latitude zones and covering approximately 440 km. The study area encompasses five counties and districts: Longyang District and Tengchong City in Baoshan, Lushui City and Fugong County in Nujiang Prefecture, and Gongshan County. The climate of this region is primarily influenced by the southwest Asian monsoon originating from the Indian Ocean, which is characterized by abundant rainfall in the summer and limited precipitation in the winter, resulting in distinct wet and dry seasons. Additionally, spatial variations in precipitation and temperature are influenced by the region’s topography31. Location of the study area. The maps in this study were processed and plotted by ArcGIS (10.8.https://developers.arcgis.com/). The map images we use come from National Geographic Information Public Service Platform Standard Map Service Network (http://bzdt.ch.mnr.gor.cn/). The zonal vegetation of Gaoligong Mountains is characterized by evergreen broadleaf forests32. Spanning five latitude zones and featuring unique topography and climate, Gaoligong Mountains fosters a complex ecological environment and rich biodiversity. It is the only region on Earth that preserves extensive areas transitioning from humid tropical forests to temperate forests33. The considerable north-south extent and relatively small east-west extent of the mountain result in minimal longitudinal zonal variation in vegetation, while distinct differences in latitudinal zonation are evident. The intricate topography of high mountain gorges has created spectacular vertical natural landscapes and a three-dimensional climate, promoting abundant biodiversity34. Gaoligong Mountains is recognized as one of the world’s biodiversity hotspots35. It is also referred to as the “Gene Bank of World Species,” designated as a “Biodiversity Conservation Zone” by UNESCO, and classified as an A-level protected area by the World Wildlife Fund. Data sources The remote sensing data utilized in this study comprises Landsat TM/ETM+/OLI (Path/Row: 132/41, 132/42, 132/43, 133/40, 133/41) and MOD13Q1 (h28v06) from 2000 to 2022, sourced from Google Earth Engine (https://earthengine.google.com/). To ensure data reliability, Landsat TM/ETM+/OLI T1_SR products that have undergone atmospheric correction and exhibit cloud cover of less than 10% were selected. The Normalized Difference Vegetation Index (NDVI) was calculated using the near-infrared (NIR) and red bands from the Landsat images, which have a spatial resolution of 30 m. The NDVI images derived from the MODIS 13Q1 product possess a spatial resolution of 250 m and a temporal resolution of 16 days. The preprocessing of the MODIS images involved projection conversion to WGS 84/UTM Zone 47 N and resampling to 30 m to align them with the Landsat NDVI images. By integrating the NDVI data from MODIS and Landsat, high spatiotemporal resolution NDVI data (30 m, 16 days) for the years 2000 to 2022 was generated. Furthermore, the MODIS images were co-registered with the available Landsat images using ENVI (5.6.https://envi.geoscene.cn/) software. To mitigate the variability in extreme climate research associated with short time scales, meteorological data from 18 weather stations in the study area and its surrounding regions, spanning from 1980 to 2022, were selected. This dataset includes daily average temperature, daily maximum temperature, daily minimum temperature, daily total precipitation, and sunshine data. The meteorological data were sourced from the China Meteorological Data Network (http://data.cma.cn/). Daily maximum temperature, daily minimum temperature, and daily total precipitation were utilized to calculate ECIs. The preliminary data for 11 extreme temperature indices and 2 extreme precipitation indices underwent homogenization and quality control, with final calculations performed using the RHtests (V4) package and the RClimDex (V1.9) package (http://etccdi.pacificclimate.org/). Daily average temperature, daily total precipitation, and sunshine data from 2000 to 2022 were aggregated into monthly data to simulate NPP. The sunshine data were employed to calculate solar radiation, referencing methods36. Subsequently, the specialized meteorological data interpolation tool ANUSPLIN37, developed by Hutchinson based on thin-plate spline theory, was utilized, with latitude and longitude as independent variables and elevation as a covariate; elevation data were sourced from the NASADEM (30 m) product. By performing spatial interpolation on the meteorological data within the study area, spatially scaled meteorological raster data were generated. To analyze the short-term climate impact of ECIs on NPP in the study area, as well as the long-term trends and cumulative effects, 13 ECIs were calculated on both monthly and annual scales (Table 1).The maps in this study were processed and plotted by ArcGIS (10.8.https://developers.arcgis.com/). The charts were plotted using Origin (2025b.https://www.originlab.com/) and the R package (4.3.3.https://www.r-project.org/). Table 1Names and definitions of ECIs.Index classificationIndexNameDefinitionUnitExtreme temperature intensity indexTNnMin TminMonthly minimum value of daily minimum temperatureoCTNxMax TminMonthly maximum value of daily minimum temperatureoCTXnMin TmaxMonthly minimum value of daily maximum temperatureoCTXxMax TmaxMonthly maximum value of daily maximum temperatureoCDTRDiurnal temperature rangeDaily temperature range: monthly mean difference between Tmax and TminoCTMAXmeanTmaxAverage of daily maximum temperatureoCTMINmeanTminAverage of daily minimum temperatureoCExtreme temperature frequency indexTN10pCold nightsPercentage of days when Tmin < 10th percentiledaysTN90pWarm nightsPercentage of days when Tmin > 90th percentiledaysTX10pCold daysPercentage of days when Tmax < 10th percentiledaysTX90pWarm daysPercentage of days when Tmax > 90th percentiledaysExtreme precipitation intensity indexRx1dayMaximum one- day precipitationMonthly maximum 1- day precipitationmmRx5dayMaximum five- day precipitationMonthly maximum 5- day precipitationmm Names and definitions of ECIs. Methodology FSDAF model The flexible spatio-temporal fusion algorithm (FSDAF) model proposed by Zhu integrates high temporal, low spatial resolution data (such as MODIS) with low temporal, high spatial resolution data (such as Landsat) to generate synthetic imagery with high spatiotemporal resolution16. The input data consists of a pair of low spatial resolution images captured at times t0 and t1, along with the high spatial resolution image at t0. Initially, the high spatial resolution image at t0 (Landsat) undergoes unsupervised classification, and the coverage differences for each class are calculated using the low spatial resolution data (250 m) from both t0 and t1. Subsequently, high-resolution data for t1 (30 m) is predicted, and the residuals of the low-resolution data pixels are computed. A thin-plate spline interpolation function is then employed to predict high-resolution data based on the low-resolution data from t1, with residuals added to the predicted high-resolution data. Finally, neighborhood information is utilized to obtain the final high-resolution imagery prediction38. MODIS data have a high temporal resolution and can provide continuous and stable vegetation dynamic information; however, their relatively low spatial resolution makes it difficult to effectively characterize the spatial heterogeneity of vegetation in the complex terrain of the Gaoligong Mountains. In contrast, Landsat data have a higher spatial resolution (30 m) and can better capture fine-scale variations in mountain vegetation, but their application is limited by revisit cycles and cloud contamination, making it difficult to obtain continuous high-quality time-series data. Therefore, this study employed the FSDAF model to integrate MODIS and Landsat data, fully combining the high temporal resolution advantage of MODIS with the high spatial resolution advantage of Landsat, thereby constructing a high spatiotemporal resolution NDVI dataset to provide reliable data support for NPP estimation. The data fusion process of the FSDAF model was implemented on the Python platform(3.10.http://www.python.org/). The specific processing steps of the FSDAF model are as follows: Where represents the predicted high spatial resolution data at time t1; represents the high spatial resolution data at time t0; indicates the change in pixel resolution between t0 and t1; is the weight determined by spatial distance, time difference, and spectral characteristics; signifies the change in the quantity of category a within b in high-resolution data between t0 and t1; represents the residual assigned from the low-resolution pixel to the high-resolution pixel. CASA model The Carnegie-Ames-Stanford Approach (CASA) model is a process model for estimating NPP based on the efficiency of plant light energy utilization. Previous studies have indicated that this model performs well in simulating NPP in China39,40. Therefore, this study selects the CASA model to simulate the spatiotemporal NPP of the Gaoligong Mountains. The NPP estimation based on the CASA model was conducted using ENVI(Classic 5.3.https://envi.geoscene.cn/) software. with the specific formulas as follows: Where represents the spatial location, and represents the month; is the net primary productivity (g C/m²) of the vegetation at location during month ; represents the photosynthetically active radiation absorbed by the pixel x in month t (g C/m²); denotes the actual light energy utilization efficiency of the vegetation (g C/MJ). Both factors together determine the NPP in the model. and are calculated using the following formulas: Where represents the total solar radiation at pixel during month ; indicates the fraction of photosynthetically active radiation absorbed by the vegetation layer. The constant 0.5 represents the proportion of the total solar radiation that can be utilized by the vegetation as effective solar radiation. NDVI is mainly used to estimate the fraction of photosynthetically active radiation absorbed by vegetation (FPAR), which is further involved in the calculation of APAR and ultimately affects NPP estimation. Therefore, improving the spatiotemporal resolution of NDVI can help enhance the accuracy of NPP estimation. Where represents the maximum light energy utilization efficiency of vegetation under ideal conditions. and denote the stress effects of low and high temperatures on the light energy utilization efficiency of vegetation, respectively. is the moisture stress coefficient, which indicates the impact of moisture on the light energy utilization efficiency. Trend analysis This study employs the Theil-Sen slope (TS) method and the Mann-Kendall (MK) test to analyze the trend of NPP. The TS method is a non-parametric statistical technique for trend calculation that effectively mitigates the influence of data anomalies on the results41. The MK test is also a non-parametric method that does not require measurement values to adhere to a normal distribution and is resilient to missing values and outliers, making it particularly suitable for significant trend analysis in long time series data42. The combination of the TS and MK methods is a valuable approach for identifying trends in long-term sequence data and has increasingly been applied to the analysis of long-term vegetation sequences43. The trend analysis was performed in R(4.3.3.https://www.r-project.org/), and the resulting maps were generated using ArcMap(10.8.https://developers.arcgis.com/) software. The equations for TS and MK are given below: Where indicates the degree of trend in the time series; denotes the median function; and refer to the data corresponding to specific times and . When > 0, the element is on an upward trend and vice versa for a downward trend. Where represents the test statistic; denotes the sign function; and indicate time; is the sample size; is the variance, where n refers to the number of elements in the sequence, is the number of groups formed by grouping the same elements in the sequence, and is the number of elements in the -th group. Where Z represents the standardized test statistic. Given a significance level of α, if |Z|> Z_(1-a/2), the trend is considered significant; otherwise, there is no obvious trend change. In this study, the significance level is set at α = 0.05, so the critical value is ± 1.96. When the absolute value of Z exceeds 1.65, 1.96, and 2.58, it indicates that the trend has passed the significance test at confidence levels of 90, 95, and 99%, respectively. Correlation analysis To study the relationship between ECIs and NPP at different temporal and spatial scales in the Gaoligong Mountains, we use the Pearson correlation coefficient to characterize the correlation between ECIs and NPP. The correlation analysis was performed on the R platform(4.3.3.https://www.r-project.org/). The calculation formula is as follows: Where represents the extreme climate index for the first year; is the NPP for the first year; is the annual average of the extreme climate index, and is the annual average of NPP. The value of the correlation coefficient r ranges between − 1 and 1. The closer the absolute value of is to 1, the stronger the correlation between the variables. Existing research has demonstrated that vegetation exhibits a lag effect in response to climatic factors within 0–3 months44. Consequently, when analyzing the response of NPP to extreme climate events, it is essential to interpret the correlation between the two during both concurrent and lagged periods. This study calculates the correlation coefficients of NPP and ECIs with lags of 0–3 months, using the maximum absolute value to represent the correlation between vegetation NPP and ECIs. Random forest regression A random forest regression model was employed to analyze the importance of 13 ECIs on NPP45. The analysis was performed on the R platform(4.3.3.https://www.r-project.org/). Extreme Climate Indices (ECIs) are a set of climate indicators used to characterize the frequency, intensity, and duration of extreme temperature and precipitation events. Compared with traditional mean climate variables, ECIs can more effectively reflect climate anomalies and their ecological effects. In this study, the 13 selected ECIs mainly cover different types of extreme climate events, including extreme high temperature, low temperature, heavy precipitation, and drought. The purpose of using these indices is to more comprehensively identify the driving effects of different types of extreme climate events on NPP variations and further reveal the response mechanisms of ecosystems to climate extremes. This model uses the percentage increase in mean squared error (%Inc MSE) to evaluate the impact of each independent variable on the dependent variable46. First, a decision tree model Ntree is established, and the mean squared error of the out-of-bag samples (OOB) is recorded as , , …, as shown in the following matrix: The importance score for variable is calculated as follows: Where n represents the number of original data samples, m and represents the number of variables. For each row in the matrix, subtract , , …, and then divide by the standard error to obtain the importance of the variable . A higher score indicates a more significant impact. Results Accuracy validation of NDVI fused with FSDAF In this study, Landsat (2017/11/14) and MODIS (2017/11/17) imagery were used as inputs for the FSDAF model to predict high-spatial-resolution NDVI on 2017/11/17. Due to the lack of cloud-free Landsat imagery corresponding exactly to the target date, Landsat NDVI (2017/11/14), with a short temporal interval of 3 days, was used as reference data to perform a near-date validation of the predicted results and evaluate the accuracy of the fused imagery.The results showed that the predicted NDVI exhibited high consistency with the actual Landsat NDVI in terms of spatial distribution patterns, while preserving more spatial detail information than the MODIS NDVI data.The scatter plot illustrates the relationship between the predicted NDVI and the actual Landsat NDVI. The fitting results were satisfactory (R² = 0.73, p < 0.01, RMSE = 0.0916), with most sample points distributed close to the y = x line (Fig. 2). These results indicate that the FSDAF model can effectively integrate Landsat and MODIS data and accurately reconstruct high spatiotemporal resolution NDVI data in the Gaoligong Mountains. Previous studies have also commonly used this method to validate the accuracy of fused models9,47. Reliability assessment of the FSDAF for NDVI fusion of Landsat and MODIS data. The left image in this study were processed and plotted by ArcGIS(10.8.https://developers.arcgis.com/); The map images we use come from National Geographic Information Public Service Platform Standard Map Service Network (http://bzdt.ch.mnr.gor.cn/); The Chart on the right in this study was plotted using the R package(4.3.3.https://www.r-project.org/). Spatial and temporal variation characteristics of NPP From 2000 to 2022, the annual average NPP in the Gaoligong Mountains was 300.60 g·C/m², with a minimum value of 240.345 g·C/m² recorded in 2000 and a maximum value of 377.431 g·C/m² in 2009. The spatial distribution of the multi-year average NPP is shown in Fig. 3a. High NPP values (> 400 g C/m²) are predominantly concentrated in the southern Gaoligong Mountains, particularly in regions with relatively high forest coverage, such as the western part of Longyang District and most areas in Tengchong City. In contrast, low NPP values (< 120 g·C/m²) are mainly distributed in the northern part of Gongshan County, as well as in Fugong County and the high-altitude regions on both the eastern and western sides of Lushui City. The spatial distribution distribution of NPP variation trends indicates that over the past 23 years, NPP in the Gaoligong Mountains has exhibited a fluctuating but overall upward trend (slope = 1.5). Area with an increasing NPP trend cover 64.78% of the Gaoligong Mountains, while areas with a decreasing trend account for 35.22%. Significant increases in NPP are observed in parts of southern Gongshan County, Fugong County, the western side of Lushui City, and northern Tengchong City, In contrast notable decreases are concentrated in parts of northern Gongshan County and southern Tengchong City (Fig. 3b). Monthly NPP variation trends in the Gaoligong Mountains are illustrated in Fig. 4. Except for May、June and November, which display a downward trend, NPP in the other months shows an upward trend, with the most significant increase occuring in January (slope = 0.35, p < 0.001). (a) The main figure shows the spatial distribution of the multi-year average NPP in the Gaoligong Mountains, while the inset illustrates the temporal variation trend of the multi-year average NPP. (b) The main figure presents the spatial pattern of NPP variation trends, with the inset highlighting areas where the variation trends reach a significance level of p < 0.05.The maps in this study were processed and plotted by ArcGIS(10.8.https://developers.arcgis.com/). The map images we use come from National Geographic Information Public Service Platform Standard Map Service Network (http://bzdt.ch.mnr.gor.cn/). (a) The side chart in the figure is drawn using Origin (2025b.https://www.originlab.com/) . Temporal trend of monthly NPP in Gaoligong Mountains from 2000 to 2022.These Charts is drawn using Origin (2025b.https://www.originlab.com/). Spatial and temporal variations of ECIs Significant interannual differences are observed in the ECIs for the Gaoligong Mountains (Fig. 5). Among the 11 extreme temperature indices, TXx, TNx, , TN90p, TX90p, TMAXmean, and TMINmean exhibit significant increasing trends (p < 0.001). In contrast, DTR, TN10p, and TX10p show declining trends, with the decrease in DTR being particularly notable (slope = -0.019, p = 0.016), indicating that nighttime temperatures are rising faster than daytime temperatures. The trends in extreme precipitation indices are relatively weak and lack statistical significance. For example, Rx1day indicates a slight increasing trend (slope = 0.07), while Rx1day demonstrates a decreasing trend (slope = −0.62). In summary, the study area exhibits a pronounced and persistent warming trend, whereas changes in precipitation indices are not statistically significant, and there is no clear evidence of a drying trend. Spatiotemporal of ECIs in the Gaoligong Mountains from 1980 to 2022 (Panels a-m depict the spatial variation trends of ECIs in the Gaoligong Mountains from 1980 to 2022, while each individual subfigure shows the temporal variation trends of the corresponding extreme climate index over the same period.) The maps in this study were plotted by ArcGIS (10.8.https://developers.arcgis.com/). The map images we use come from National Geographic Information Public Service Platform Standard Map Service Network (http://bzdt.ch.mnr.gor.cn/). The attached Chart in the above figure are all drawn using Origin (2025b.https://www.originlab.com/). The spatial variability of ECIs across the Gaoligong Mountains reveals distinct trends. TNx, TN90p, TMINmean, TX90p, and TMAXmean exhibit an overall increasing trend across the entire study area. In contrast, TXx and TXn demonstrate predominantly increasing trends in the northern Gaoligong Mountains but a decreasing trend in the southern region. Conversely, TN10p and TX10p display increasing trends in the central and northern regions while showing decreasing trends in the southernmost areas. This pattern contrasts with the spatial distribution of TNn, suggesting that the reduction in extreme cold events is more pronounced in the central and northern regions. For DTR, an increasing trend is observed in the southern Gaoligong Mountains and lower-altitude areas, whereas most other regions exhibit a declining trend. Regarding the spatial distribution of extreme precipitation indices, Rx1day shows increasing trend in the northern Gaoligong Mountains and Longyang District, while decreasing trends are concentrated in Lushui and Tengchong City. Similarly, Rx5day primarily shows an increasing trend in Longyang District, while other regions demonstrate a decreasing trend. Response of NPP to extreme climate Annual NPP response to extreme climate Annual NPP shows a strong correlation with extreme temperature indices, such as TXx, TNx, TXn, and TMAXmean (Fig. 6). The increase in the intensity and frequency of extreme temperatures significantly promotes vegetation growth. TX10p is the only extreme temperature index that exhibits a significant negative correlation with NPP, suggesting that low temperatures may somewhat inhibit photosynthesis and growth potential in vegetation. A significant negative correlation is observed between extreme precipitation indices and NPP, primarily in the northern regions of Gaoligong, including Gongshan, Fugong, and Lushui. The response of NPP to ECIs varies among different vegetation types, with extreme temperature indices having a greater impact on NPP across these types. The NPP of barren land shows a significant positive correlation with TXn and a significant negative correlation with TXx. Forest and shrubland ecosystems are more sensitive to extreme temperature events, with temperature fluctuations contributing to enhanced photosynthesis and biomass accumulation. However, both forest and shrubland NPP exhibit a significant negative correlation with TX10p, indicating that low-temperature events suppress vegetation growth and productivity in these areas. Spatial correlation between NPP and ECIs in the Gaoligong mountains. The maps in this study were processed and plotted by ArcGIS 10.8. The Chart we use come from National Geographic Information Public Service Platform Standard Map Service Network (http://bzdt.ch.mnr.gor.cn/). (Panels a-m show the relationships between NPP and various ECIs, while panel n illustrates the correlation between NPP for different vegetation types and the respective extreme climate indices.The chart was plotted using the R package (4.3.3.https://www.r-project.org/).) Responses of NPP of different vegetation types to extreme climate indices at the seasonal scale The results showed that the responses of NPP to extreme climate indices (ECIs) varied considerably among vegetation types at the seasonal scale, with pronounced temporal heterogeneity among seasons (Fig. 7). Overall, the correlations between NPP and ECIs were stronger in spring and autumn than in summer and winter, and temperature-related indices exhibited the strongest explanatory power for NPP variations. Correlation patterns between NPP and extreme climate indices in different seasons. The data processing and figure generation for this figure were completed using Python (3.10.http://www.python.org/). In spring, NPP of most vegetation types was positively correlated with TMAXmean, TMINmean, TNn, TNx, and TXx. Among them, shrubs exhibited the strongest response to TMINmean, indicating that increases in minimum temperature during spring significantly promoted shrub productivity. Strong positive correlations were also observed between shrubs and TNn as well as TNx, suggesting that changes in extreme low-temperature conditions played an important regulatory role in shrub growth. Forests also showed high sensitivity to temperature-related indices, particularly TMAXmean and TXx, while grasslands responded more strongly to TMINmean, TNn, and TNx. In contrast, DTR generally exhibited negative correlations with all vegetation types in spring, with the strongest negative relationships observed in barren land, cropland, and shrubs, indicating that larger diurnal temperature ranges may suppress vegetation carbon accumulation. In addition, RX1day and RX5day showed moderate positive correlations with most vegetation types, suggesting that intense precipitation events in spring could alleviate water limitation and thereby promote vegetation recovery and NPP increase. In summer, the correlations between NPP and ECIs weakened considerably across all vegetation types. TMAXmean remained positively correlated with grasslands and forests, indicating that higher summer temperatures still promoted vegetation productivity to some extent, although the effect was substantially weaker than in spring and autumn. Meanwhile, DTR shifted from negative correlations in spring to weak positive correlations in summer, particularly for forests and grasslands, suggesting that changes in diurnal temperature range during summer may enhance heat utilization efficiency. TX90p also exhibited positive correlations with forests and grasslands, implying that more frequent high-temperature events could still exert a certain promoting effect on some vegetation types. However, the overall correlations in summer were relatively weak, indicating that vegetation growth during this period may already approach thermal saturation, thereby limiting the influence of extreme climate variations on NPP.Autumn was identified as the season with the strongest responses of NPP to ECIs. Cropland showed strong positive correlations with TXx, TNx, TMINmean, and TMAXmean, among which TXx exhibited the highest correlation, followed by TNx and TMINmean. Barren land also showed strong positive correlations with TMAXmean and TXx, indicating that warming conditions in autumn significantly enhanced regional vegetation productivity. Similarly, shrubs exhibited strong positive relationships with TNx and TXx. In addition, RX1day and RX5day were positively correlated with most vegetation types in autumn, suggesting that increased precipitation could alleviate late-season water limitation and delay vegetation senescence. In contrast, DTR remained predominantly negatively correlated with NPP, indicating that larger diurnal temperature ranges may reduce the efficiency of vegetation utilization of heat and water resources, thereby constraining NPP increase.In winter, the correlations between NPP and ECIs were generally the weakest, and negative correlations became more prevalent. Cropland exhibited the strongest negative correlation with DTR, indicating that increased diurnal temperature range in winter significantly inhibited cropland NPP. Barren land also showed negative correlations with RX5day and RX1day, suggesting that intense winter precipitation events may be unfavorable for maintaining vegetation productivity. Furthermore, TXx was negatively correlated with barren land, shrubs, and grasslands, indicating that extreme high-temperature events in winter had limited positive effects on vegetation carbon fixation. Nevertheless, some indices still showed weak positive correlations; for example, cropland exhibited positive correlations with TX10p and TN90p, implying that reduced low-temperature events in winter may help maintain basic physiological activities in agricultural ecosystems. Among vegetation types, cropland was the most sensitive to extreme climate variations, especially in autumn when strong correlations were observed with TMAXmean, TMINmean, TNx, and TXx. Shrubs and grasslands were more responsive to TMINmean, TNn, and TXx, whereas forests exhibited relatively stable correlations overall, suggesting a certain buffering capacity against short-term climate fluctuations. Lagged response of monthly NPP to climate extremes Significant differences are observed in the lag effects of various ECIs on NPP, as well as notable variations in lag effects among different vegetation types (Fig. 8). Overall, the monthly NPP exhibits a weaker response to TN10p, TX10p, TN90p, and TX90p compared to other ECIs, with a lag effect of 1–3 months for the latter. Specifically, for extreme temperature indices, NPP responds to DTR, TNx, TXn (except for forest, where the peak response occurs in February), TNn (except for forest and grassland, where the peak response occurs in February), TMINmean, and TX10p with the maximum response observed at a lag of 1 month. The response to TXx peak at a lag of 2 months. The responses of shrubland, cultivated land, and barren land to TMAXmean peak at a lag of 1 month, while forest and grassland show their peak at a lag of 2 months. Regarding extreme precipitation, the lag response times of NPP to Rx1days and Rx5days are consistent across vegetation types, with shrubland and cultivated land showing a lag response time of 1 month (p < 0.001). Shrubland and cultivated land exhibit the maximum response at a lag of 1 month, while grassland, forest, and barren land reach their maximum response at a lag of 2 months. Additionally, the correlation coefficients between NPP and extreme temperature and precipitation indices at a lag of 1 month are generally higher than those at a lag of 2 months, indicating that the effects of extreme climate events on NPP are more significant in the short term. Lagged response of NPP to ECIs for different vegetation types. (Note: “Lag0” indicates no time delay and “Lag0-3” indicates a time delay of 0–3 months.) These Charts are all drawn using Origin (2025b.https://www.originlab.com/). Importance ranking of ECIs on NPP A random forest regression model was employed to analyze the impact of ECIs on NPP in the study area. Validation indicated that the model performed optimally with 500 decision trees and a feature number (mtry) set to 10. Based on these parameters, the importance ranking of ECIs derived from the random forest analysis (Fig. 9). Overall, the most significant ECIs influencing NPP in the Gaoligong Mountains are extreme temperatures, specifically DTR, TX90p, TXx, and TMAXmean. However, the effects vary by vegetation type; TX90p is the most critical factor affecting NPP in forested and cultivated areas, while TXx has the most substantial impact on NPP in shrubland and bare land. For grassland NPP, the most important factor is TXn. The importance rankings of Rx1day and Rx5day are relatively low, indicating their minor contribution to the NPP of vegetation in the Gaoligong Mountains. The impact of ECIs on cultivated land is greater than on grassland, indicating that extreme weather events may have a significant impact on agricultural production. The effect on forests is relatively smaller, which may be due to the greater ecological resilience and resistance of forests compared to cultivated land and grassland vegetation. Importance ranking of ECIs on NPP. These Charts are all created using the R package (4.3.3.https://www.r-project.org/). Conclusion This study employed the FSDAF to integrate Landsat and MODIS data from 2000 to 2022, in conjunction with the CASA model, to derive high spatiotemporal resolution NPP. It also analyzed the effects of ECIs on NPP. The main conclusions are as follows: The average annual NPP in the Gaoligong Mountains from 2000 to 2022 was 300.60 g·C/m², indicating a fluctuating upward trend in annual NPP changes (slope = 1.5). With the exception of May, June and November, where NPP displayed a fluctuating downward trend, NPP exhibited an upward trend in all other months, with January demonstrating the most significant increase (slope = 0.35, p < 0.001). The average annual NPP in the Gaoligong Mountains from 2000 to 2022 was 300.60 g·C/m², indicating a fluctuating upward trend in annual NPP changes (slope = 1.5). With the exception of May, June and November, where NPP displayed a fluctuating downward trend, NPP exhibited an upward trend in all other months, with January demonstrating the most significant increase (slope = 0.35, p < 0.001). The extreme temperature events in the study area generally showed an upward trend. Significant increases (p < 0.001) were observed in indices such as TXx, TNx, TN90p, TX90p, TMAXmean, and TMINmean. In contrast, the trends of extreme precipitation indices RX1day and RX5day were relatively minor. Overall, the study area is experiencing a continuous warming trend. The extreme temperature events in the study area generally showed an upward trend. Significant increases (p < 0.001) were observed in indices such as TXx, TNx, TN90p, TX90p, TMAXmean, and TMINmean. In contrast, the trends of extreme precipitation indices RX1day and RX5day were relatively minor. Overall, the study area is experiencing a continuous warming trend. The primary factor affecting NPP in the Gaoligong Mountains is extreme temperature. Changes in NPP showed a significant positive correlation with extreme temperature indices and a notable negative correlation with extreme precipitation indices. This relationship was particularly pronounced in shrubs and crops. The primary factor affecting NPP in the Gaoligong Mountains is extreme temperature. Changes in NPP showed a significant positive correlation with extreme temperature indices and a notable negative correlation with extreme precipitation indices. This relationship was particularly pronounced in shrubs and crops. The responses of NPP to extreme climate indices varied significantly among different seasons, with the strongest responses occurring in spring and autumn, while relatively weaker responses were observed in summer and winter. Temperature-related indices exerted stronger effects on NPP variations than precipitation-related indices. In addition, the sensitivity to extreme climate change differed among vegetation types, with cropland showing the strongest response, whereas forests exhibited relatively stable ecological buffering capacity. The responses of NPP to extreme climate indices varied significantly among different seasons, with the strongest responses occurring in spring and autumn, while relatively weaker responses were observed in summer and winter. Temperature-related indices exerted stronger effects on NPP variations than precipitation-related indices. In addition, the sensitivity to extreme climate change differed among vegetation types, with cropland showing the strongest response, whereas forests exhibited relatively stable ecological buffering capacity. NPP responds to extreme temperatures and precipitation with a lag of one to two months. The monthly NPP responses to TN10p, TX10p, TN90p, and TX90p were significantly weaker than those to other ECIs. NPP responds to extreme temperatures and precipitation with a lag of one to two months. The monthly NPP responses to TN10p, TX10p, TN90p, and TX90p were significantly weaker than those to other ECIs. Discussion Variation characteristics of ECIs and NPP During the period from 2000 to 2022, the intensity and frequency of extreme temperatures in the Gaoligong Mountains exhibited significant increases, consistent with observations in other regions48. In contrast, the trend of extreme precipitation intensity in Gaoligong Mountains is not significant increase. This difference may be due to the relatively small magnitude of extreme precipitation trend and the limited number of selected indices, which may not fully capture the overall variation of extreme precipitation. The NPP in the Gaoligong Mountains increased from 2000 to 2022, aligning with studies from other regions in Yunnan Province49,50, indicating that the Gaoligong Mountains play an important role in the terrestrial ecosystem’s carbon cycle. The annual NPP also exhibits a spatial trend of “more in the south and less in the north,” but is far below the Yunnan provincial average of 890.97 g C/m²49. Spatially, both the northern and southern NPP show a marked decreasing trend. The decline in northern NPP may be due to altitude limitations, as plants in high-latitude regions is typically constrained by thermal conditions (radiation and temperature), the reduction in solar radiation negatively impacts plant growth51. In contrast, the decrease in southern NPP may be attributed to rising temperatures and increased precipitation. Frequent precipitation, along with rising temperatures, promotes increased cloud cover, affects photosynthetic efficiency, and reduces NPP52,53. Model studies have found that in years with a longer growing season, if increased summer respiration is offset and exceeded by higher GPP in early spring, the annual NPP will increase54. Temporally, the NPP in June for the Gaoligong Mountains shows a significant decreasing trend. The extension of the growing season leads to increased spring transpiration and carbon uptake, which in turn limits the ability of vegetation to absorb soil nutrients and moisture in June, thus reducing NPP55,56. Although climate change restricts NPP growth in summer, there may be a slight increase in NPP during autumn. However, due to the longer day length in spring compared to autumn, it is expected that spring NPP will be higher than that of autumn57. Furthermore, we observed a significant decline in NPP in November, potentially due to leaf renewal. While evergreen species can photosynthesize year-round, leaf senescence and renewal in late autumn can lead to decreased NPP; the extended growing season alters the rate of leaf renewal, resulting in a decline in NPP46,58. Response of NPP to ECIs As the frequency of extreme climate events rises, an increasing number of researchers are examining the effects of these events on vegetation growth59. In the Gaoligong Mountains, most extreme temperature indices demonstrate a significant upward trend, while extreme precipitation exhibits a slight downward trend, consistent with the findings of He49. The extreme high temperatures enhance the growth of NPP, whereas extreme precipitation tends to inhibit it, though its impact is relatively small. This observation contrasts with findings from other regions in Yunnan and the coastal areas of China49,60. Several extreme high temperature indices are generally beneficial for increasing NPP. This may be attributed to the abundant water resources in the study area; higher temperatures enhance evapotranspiration, thereby mitigating the impact of flood disasters61. Conversely, extreme low temperatures indices can severely hinder plant growth. However, TN10p and TX10p helps reduce the frequency and severity of low-temperature damage, thereby promoting NPP62. Additionally, a reduction in the DTR may lead to decreased rates of photosynthesis and increased respiration, ultimately resulting in lower net photosynthetic accumulation63. In low altitude regions, elevated nighttime temperatures can prolong the growing season, fostering NPP growth. Shrubs exhibit the highest sensitivity to extreme temperature index, typically occurring at higher altitudes where their growth is significantly influenced by low temperatures. An increase in optimal temperature enhances photosynthesis, extends the growing season, and increases NPP60. Extreme precipitation indices have a suppressive effect on NPP, likely due to the mountainous-valley terrain, where soils on gentle slopes are susceptible to erosion. High-intensity rainfall may exacerbate terrain fragmentation and soil erosion on steep slopes, leading to the loss and transfer of soil organic carbon and altering nutrient distribution, which negatively impacts regional NPP4,64. Variations in ecological environments and vegetation cover across different geographical areas result in inconsistent responses of ecosystems to extreme climate conditions65. For example, while extreme high temperatures enhance productivity in the grassland ecosystems of the northwestern Tibetan Plateau, they hinder productivity in the dry valleys of Yunnan66. Furthermore, several regions have reported increased vegetation productivity due to extreme precipitation67; however, there is no significant impact on vegetation productivity in the Loess Plateau during certain seasons or years68. The results of this study indicate that the primary extreme climate factor influencing NPP is extreme temperature. This further emphasizes the positive correlation between the upward trend of NPP fluctuations (slope = 1.5) and warming in the study area. This finding aligns with the research conducted by Cao et al. on the Yellow River basin59. Lagged response of NPP to ECIs An increasing number of studies have found that due to the buffering capacity of ecosystems, plant physiological adaptation processes do not cease immediately following the occurrence of extreme climate events69. Differences in lagged responses can be explained by vegetation physiological traits and ecological mechanisms. Adaptive mechanisms of photosynthesis and respiration may lead to a temporal delay in the transport of photosynthetic products between leaves and roots, and this lag varies among different vegetation types70. In the present study, the observed lag differences indicate that forests may exhibit the longest NPP lag response due to their woody structure, deep roots, and slower transport of photosynthetic products to belowground organs; grasslands, composed of herbaceous plants with short growth cycles and rapid allocation of photosynthates, show the shortest lag. The lag responses of shrublands and croplands are further influenced by human activities such as cultivation and irrigation.The growing season of grasslands and croplands is primarily regulated by heat, whereas plant growth is mainly influenced by soil moisture; thus, different ecosystems respond to climate factors on different temporal scales71,72. In the mid-to-northern regions of China, the impact of extreme temperatures on NPP generally exhibits a lag of approximately one month, while the lag associated with extreme precipitation is relatively longer73. Grasslands respond to climate indices with a shorter lag, with 18.68% of responses occurring within one month74, which is consistent with the short life cycle and rapid growth characteristics of grassland species. Cropland crops are affected by management practices; irrigation can alleviate short-term drought stress, while fertilization promotes growth and alters responses to extreme climate events. For example, when extreme precipitation occurs during periods of low soil moisture, its effect may not appear until one month later75. Shrublands, characterized by a short growing season and cold environment, rely more on individual climate events during growth, resulting in relatively short lag times76.Compared with other vegetation types, forests in the western Gaoligong Mountains exhibit the longest NPP response lag to extreme climate indices, which differs from results reported for the Loess Plateau29. Generally, forest ecosystems have well-developed roots and strong soil and water conservation capacities, resulting in higher response thresholds to extreme events and reduced sensitivity to environmental fluctuations77. In contrast, grasslands and croplands in the Gaoligong Mountains typically begin growing around April, with growth cycles markedly shorter than those of forests, leading to shorter lag times78. Additionally, under the context of asymmetric global warming, the significant reduction in diurnal temperature range accelerates vegetation respiration, inhibits biomass accumulation, and consequently suppresses or delays plant growth79. Evergreen and deciduous forests also generally exhibit longer temperature response lags than grasslands, croplands, and shrublands80.The mechanisms by which temperature and precipitation influence NPP differ: temperature primarily regulates plant growth directly by affecting photosynthesis and respiration, whereas precipitation primarily influences growth indirectly through soil moisture. When acting simultaneously, these factors may produce either synergistic or counteracting effects. Therefore, to accurately assess ecosystem responses to extreme climate events, it is necessary in future studies to distinguish lag pathways among different vegetation types and extreme climate indices. Furthermore, the results of this study indicate that short-term extreme events (lag of one month) have more pronounced effects on vegetation NPP, highlighting the need to consider the short-term cumulative effects of extreme events and their potential ecological risks in future research. Limitations and prospects This study utilized synthetic NDVI images to investigate the impact of extreme climate change on NPP in the Gaoligong Mountains from 2000 to 2022. Compared to MODIS data, the synthetic NDVI data offers advantages in understanding NPP changes at a finer scale81,82. Although significant foundational results have been obtained, this study also has several limitations. The primary inherent uncertainty lies in the remote sensing image data. Due to the influences of atmospheric conditions, cloud cover, and solar angles, terrestrial satellite data may contain uncertainties83. When remote sensing image data represents continuous spectral information, the images can only be displayed through the gray values of discrete points, which inevitably leads to information loss84, which may result in biases in data interpretation.In addition, due to the lack of observation data, this study only conducted accuracy validation between the fused NDVI and actual Landsat NDVI data, without further validation of the final NPP results. In the CASA model, NDVI is one of the key input parameters and is mainly used to estimate the fraction of photosynthetically active radiation absorbed by vegetation (FPAR), thereby influencing NPP estimation. The climate datasets used in this study, including air temperature, precipitation, and solar radiation, were processed using the professional meteorological interpolation tool ANUSPLIN, while vegetation type data were obtained from the CLCD dataset. Therefore, the reliability of the fused NDVI data largely affects the accuracy of the estimated NPP results.Moreover, commonly used NPP products for validation, such as MOD17A3HGF NPP and GLASS NPP, have relatively coarse spatial resolutions and are not well matched with the 30 m resolution NPP data generated in this study, making strict pixel-scale validation difficult. Meanwhile, long-term continuous field-measured NPP data are lacking in the study area. Future studies could combine higher-precision field observations and multi-source NPP datasets to further improve the validation and uncertainty assessment of NPP estimation results. This study was mainly conducted using monthly, seasonal, and annual scale data, with a greater focus on the long-term cumulative responses of vegetation to extreme climate conditions. Therefore, the instantaneous ecological effects and underlying mechanisms associated with short-term extreme climate events were not fully characterized. In addition, due to the limited availability of documented typical extreme events in the study area, this study was unable to establish a detailed correspondence between specific extreme climate events and NPP variation processes. Future research should integrate higher temporal resolution data and information on typical events, such as heatwaves and extreme precipitation events, to further elucidate the dynamic impacts of different types of extreme climate events on vegetation NPP and their ecological response mechanisms at the event scale. Given these uncertainties and limitations, it is necessary to explore finer spatial datasets and more reliable methods to characterize the spatiotemporal variations of net primary productivity (NPP), while future research should integrate higher‑temporal‑resolution data and information on typical extreme climate events to reveal the impact mechanisms of extreme climate on NPP at the event scale, thereby enabling a more accurate identification of factors influencing vegetation change. Acknowledgements The author acknowledgement the financial support by Funding: This work was supported by the Yunnan Fundamental Research Projects (Grant 202301BD070001-246), the National Natural Science Foundation of China (Grant 42377444), the National Key Research and Development Program of China (Grant 2024YFF0807901), the Second Tibetan Plateau Scientific Expedition and Research(STEP)program(2024QZKK0301) and the Yunnan Fundamental Research Projects (Grant 202501BD070001-025). Author contributions Conceptualization, T.Y. and Z.Z.; Data curation, T.Y.; Funding acquisition, Z.Z.; Project administration, Z.Z.; Supervision, J.W.; Validation, T.Y. and W.Z.; Writing – original draft, T.Y.; Writing – review & editing, T.Y., W. Z., L.Y., Z. G., J.W. and Z. Z. Funding This work was supported by the Yunnan Fundamental Research Projects (Grant 202301BD070001-246), the National Natural Science Foundation of China (Grant 42377444), the National Key Research and Development Program of China (Grant 2024YFF0807901), the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2024QZKK0301) and the Yunnan Fundamental Research Projects (Grant 202501BD070001-025). Data availability The datasets generated and analyzed in this study are available from the corresponding author on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Citations Data Availability Statement The datasets generated and analyzed in this study are available from the corresponding author on reasonable request.