Search

It has been widely reported that anthropogenic influences are detectable with high confidence in global warming. However, whether human activities have an impact on snowfall changes is still unclear. Here we based on phase 6 of the Coupled Model Intercomparison Project (CMIP6) multi-forcing dataset and the regularized optimal fingerprint method, the detection and attribution of various grades of snowfall (including annual snowfall (>0.1 mm/day), light snowfall (<2.5 mm/day), intense snowfall (>5 mm/day), and their corresponding days) changes in Eurasia (20 ° -90 °N , 10 °W -180 °E) were carried out. Results show that anthropogenic activity forcing (ANT) and greenhouse gas forcing (GHG) well reproduced the spatial-temporal characteristics of snowfall indices. The ANT influence is robustly detected in the decreased trend of snowfall days (snowday), light snowfall, and light snowfall days (light_day) at the 90% confidence interval, clearly separated from the natural forcing. Moreover, the GHG signals are detectable for decreases in the three snowfall indices, which only could be distinguished from the natural and aerosol forcings for snowday and light snowfall. Thus, anthropogenic activities may considerably account for the decreases in snowday, light snowfall, and light_day across Eurasia, wherein the changes in the first two indices dominated by GHG emissions. However, human influence detections fail for intense snowfall, and it is hard to detect on regional scales, except for North Asia. Finally, by the end of this century (2081–2100), the observation-constrained projections based on the detection and attribution analysis under two SSPs (new Shared Socioeconomic Pathway) scenarios exhibit that the scaled snowday, light snowfall, and light_day are expected to decrease by approximately 13.9 days (28.3 days), 24.8% (48.7%), and 4.3 days (8.8 days) under the SSP2–4.5 (SSP5–8.5) scenario with reference to the current climate (1995–2014). Our study highlights the need to improve climate model performance in simulating extreme snowfall to clear whether and to what extent human influence impacts it. In the past 35 years, the decrease in snowfall days (snowday), light snowfall, and light snowfall days (light_day) can be attributed to human influence. However, CMIP6 simulations may underestimate these observed changes. We utilize the observationally constrained techniques based on detection results to constrain the projections for snowfall changes. Our results show that by the end of the 21st century, the scaled snowday, light snowfall, and light_day are expected to decrease by approximately 13.9 days (28.3 days), 24.8% (48.7%), and 4.3 days (8.8 days) under the SSP2–4.5 (SSP5–8.5) scenario with reference to the current climate. [Display omitted] • In the past decades, the observed snowfall decreased while extreme snowfall increased across most of Eurasia. • The decrease in snowfall days and light snowfall could be attributed to anthropogenic influence dominated by GHG emissions. • CMIP6 models may underestimate the observed changes in snowfall. • The observation-constrained projections show a larger decrease for snowday and light snowfall than the raw CMIP6 outputs. [ABSTRACT FROM AUTHOR]

• A modified isotope-based method was developed to estimate the evapotranspiration partitioning, and it can be applied on various time scales. • The modified method and the traditional method are equivalent at a 15-minute time scale. • The modified method is capable of partitioning evapotranspiration at high-frequency and distributed the high sensitivity parameter δ ET into other three less sensitive parameters. • The high-frequency method is advantageous in avoiding the extrapolation of the keeling plot regression line, and enabling the calculation of confidence interval of evapotranspiration partition. • The high-frequency method might be useful when investigating evapotranspiration partitioning under short-term extreme weather events and flush agricultural irrigation. To better understand water and energy cycles, numerous efforts to partition evapotranspiration (ET) into evaporation (E) and transpiration (T) have been made over the recent half century. One of the analytical methods is the isotopic approach. The isotopic composition of ET (δ ET) is a crucial parameter in the traditional isotope-based ET partitioning model, which however, has considerable uncertainty and high sensitivity. Here we proposed a modified T fraction in total ET (F T) calculation using Keeling plot slope (k), the atmospheric vapor concentration (C v), and the isotopic composition of atmospheric vapor (δ v), to avoid the direct use of δ ET. Following the traditional method, we used the Craig-Gordon model for the isotopic composition of evaporation (δ E) and chamber method for the isotopic composition of transpiration (δ T) in our modified method. The modified F T calculation method (F T (m)) can be applied at a 15-min time scale using the average values (F Ti (mp)) and at a 1 Hz time scale for high-frequency method (F Ti). The modified method was verified by both theoretical derivations and field observations. F Ti (mp) was equivalent to those using the traditional isotopic method at a 15-min time scale. However, F Ti eliminated the highly sensitive parameter δ ET , and redistributed the sensitivity of δ ET into three less sensitive parameters. Additionally, F Ti has two main advantages. First, the high-frequency method avoids the extrapolation of the Keeling plot regression line intercept. Second, the high-frequency method can produce a 95% confidence interval of F T in a measurement cycle (e.g., 15 min). The calculated confidence interval was different from that of traditional uncertainty analysis. The high-frequency method might be useful when investigating evapotranspiration partitioning under short-term extreme weather events and flush agricultural irrigation. [ABSTRACT FROM AUTHOR]