Introduction
1. Background
Observational data directly measure atmospheric states but suffer from spatial inhomogeneity, noise, and indirect measurement of certain variables. Numerical models offer physically consistent atmospheric representations, yet remain vulnerable to model imperfections and sensitivity to initial conditions. Data assimilation (DA) addresses these limitations by optimally combining observations with model outputs, producing reanalysis datasets that are both spatially complete and physically coherent.
Daily maximum temperature (Tmax) is one of the most critical variables in surface climate monitoring, serving as a key indicator for human health, energy demand, and agricultural impact. However, accurately representing Tmax in reanalysis products remains challenging, as it is strongly influenced by local surface conditions — including terrain complexity, land-sea contrast, and urban heat effects — that are difficult to capture at coarse grid resolutions.
2. Motivation
Heatwaves, defined by sustained periods of extreme Tmax, are highly nonlinear phenomena that are disproportionately sensitive to small initial errors — making their accurate representation a critical challenge. The summer of 2024 (JJA) recorded the highest number of heatwave days in the past five years (KMA), underscoring the urgency of evaluating how well current datasets capture such extremes. This raises a key question: do data-assimilated reanalysis products better represent the Tmax distribution and heatwave characteristics — including frequency, intensity, and spatial structure — compared to raw observational data?
3. Objectives
This study evaluates how accurately ERA5 reanalysis reproduces daily maximum temperature and heatwave characteristics compared to ASOS station observations during the summer of 2024 (JJA), focusing on:
- Representation of daily maximum temperature (Tmax) distributions
- Spatial distribution of biases between coastal and inland regions
- Characteristics of heatwave frequency, duration, and intensity