A Model between Near-Surface Air Temperature Change and Dynamic Influencing Factors in the Eastern Tibetan Plateau, China

Wentao Xu, Qinjun Wang, Dingkun Chang, Jingjing Xie, Jingyi Yang2022 · DOI

TL;DR

This study develops a statistical model to understand the complex interplay of factors influencing near-surface air temperature changes in the Eastern Tibetan Plateau (ETP). By analyzing 20 years of data, the research identifies key drivers like vegetation and snow cover, revealing their varying importance across seasons and providing a crucial step towards better climate change predictions and mitigation strategies for this sensitive region.

What Problem Does This Paper Address?

Climate change is a global phenomenon with profound impacts, and regions like the Tibetan Plateau (TP) are particularly vulnerable due to their sensitivity to climatic shifts. While global warming is well-established, pinpointing the specific drivers and their quantitative contributions to temperature changes in complex environments like the ETP remains a challenge. Existing research often focuses on broad spatial-temporal patterns or individual factors, leaving a gap in understanding the dynamic, interconnected influences on temperature at a regional level. This paper aims to fill this gap by developing a quantitative model that links near-surface air temperature changes in the ETP to a suite of dynamic influencing factors, providing a more nuanced understanding of regional climate dynamics.

Methods

The researchers employed a multi-faceted approach to model temperature changes in the ETP:

  • Data Standardization: To account for the varying magnitudes and dimensions of different influencing factors (e.g., CO2 concentration vs. vegetation cover), data was standardized. This involved calculating anomalies (deviations from the normal level) and then standardizing these anomalies to a common scale. This step is crucial for ensuring that no single factor unduly dominates the analysis due to its scale.
  • Single-Factor Correlation Analysis: Before building a comprehensive model, the study first assessed the individual correlation between standardized temperature change and each dynamic influencing factor (carbon dioxide concentration, vegetation cover, urban cover, snow cover, water cover, and bare land cover). This helped identify the most relevant factors for each period (annual and seasonal).
  • Least Squares Multivariate Statistical Analysis: The core of the modeling approach involved using multiple regression analysis. This statistical technique establishes a mathematical relationship between a dependent variable (temperature change) and one or more independent variables (the identified influencing factors). The model aims to quantify the contribution of each factor to the overall temperature change.
  • Statistical Significance Testing: To ensure the reliability of the model and its parameters, the study utilized t-tests and p-values. These tests determine whether the observed relationships between factors and temperature change are statistically significant or likely due to random chance. Factors with non-significant p-values were removed from the model.

Datasets

The study utilized a range of datasets, primarily from 2003 to 2020, for the Eastern Tibetan Plateau:

  • Air Temperature: Measured daily data from 45 meteorological stations.
  • Carbon Dioxide Concentration: Monthly reanalysis data from the Copernicus Atmospheric Monitoring Service (CAMS).
  • Vegetation Cover: Monthly MODIS/Terra Vegetation Indices (MOD13A3).
  • Urban Cover: Derived from nighttime light index products (A Prolonged Artificial Nighttime-light Dataset of China).
  • Snow Cover: Daily cloud-free snow cover products for the Tibetan Plateau.
  • Water Cover: Derived from MODIS/Terra Vegetation Indices.
  • Bare Land Cover: Derived from MODIS/Terra Vegetation Indices.
  • Digital Elevation Model (DEM): A 1 km resolution dataset for the region.

The study focused on a 100 km buffer zone around each meteorological station to capture local environmental influences.

Key Insights and Ideas

  • Seasonal Sensitivity: The ETP's temperature is not influenced by a static set of factors; their importance shifts significantly between seasons. This highlights the need for season-specific climate models and adaptation strategies.
  • Vegetation and Snow Dominance: Vegetation cover and snow cover emerged as the most consistently influential dynamic factors across various periods. This underscores the critical role of land surface characteristics in modulating regional climate.
  • Data Standardization is Key: The paper emphasizes that standardizing diverse data types is essential for a meaningful multivariate analysis. Without it, factors with larger inherent scales could mask the contributions of others.
  • Urbanization's Limited Impact (in this region): Despite being a known driver of temperature change in many areas, urban cover showed insignificant influence in the ETP. This is attributed to the region's low urbanization levels and dispersed urban layouts.
  • Negative Snow-Temperature Correlation: The study consistently found a negative correlation between snow cover and temperature. This aligns with the understanding that snow reflects solar radiation, leading to cooling.

Key Results

  • Model Establishment: A multivariate statistical model was successfully established for temperature change in the ETP across annual and seasonal periods.
  • Factor Identification: The study identified specific dynamic influencing factors that significantly affect temperature in different periods. For instance, in autumn, carbon dioxide concentration, vegetation cover, and snow cover were key drivers.
  • Correlation Strength: The paper provides correlation coefficients for various factors and periods, indicating the strength and direction of their relationships with temperature change. For example, in winter, snow cover had the strongest negative correlation (-0.392), while vegetation cover had a positive correlation (0.360).
  • Model Fit: The determination coefficients (R²) for the models across different periods ranged from 0.1711 (winter) to 0.3863 (autumn), indicating a moderate fit. The paper notes that while these R² values might seem modest, they represent a significant step in understanding the complex interactions in this region.

Limitations and Future Work

  • Data Resolution and Scale: While the study used various datasets, the spatial and temporal resolutions might not perfectly capture all microclimatic variations. The 100 km buffer zone, while practical, might also smooth out finer-scale localized effects.
  • Model Fit: The R² values, while statistically significant, suggest that other unmeasured factors or more complex interactions are at play in driving temperature changes in the ETP.
  • Future Work: The authors suggest that this research provides a theoretical basis and direction for similar analyses in other regions, particularly the "three polar regions." Further work could explore more sophisticated modeling techniques, incorporate additional influencing factors (e.g., atmospheric circulation patterns), and investigate the feedback mechanisms between temperature change and influencing factors more deeply.

Summary updated

BibTeX
@misc{xu2022model,
  author = {Wentao Xu and Qinjun Wang and Dingkun Chang and Jingjing Xie and Jingyi Yang},
  title = {{A Model between Near-Surface Air Temperature Change and Dynamic Influencing Factors in the Eastern Tibetan Plateau, China}},
  year = {2022},
  doi = {10.3390/s22166196},
}
Figure