top of page

Developing a Flash Drought Index Based on Atmospheric Pressure Variability

Focusing on the Yeongdong region of South Korea, we demonstrate the potential of a region-specific, flash-drought-oriented drought index. This approach not only enhances the understanding of flash drought mechanisms but also provides a new tool that may support monitoring, prediction, and response strategies of future droughts.

가뭄

Motivation

Motivation

 

Changing Nature of Droughts

  • Traditional drought: Long-term precipitation deficit over large areas (typically >1 month).
     

  • Flash drought: Rapid onset caused by low rainfall and high evapotranspiration due to heat. Can develop in days to a few months.
     

 

Circulation Drivers of Flash Droughts

 

Key Factors

  • Cyclone deficiency → Reduced rainfall
     

  • Persistent anticyclone → Strong radiation & high evapotranspiration

 

Mechanism

  • Combined effects rapidly dry the soil
     

  • Pushes moisture below a critical threshold

 

Characteristics

  • Triggered by high temperatures, strong winds, and radiation changes.
     

  • Often linked to climate patterns like La Niña.
     

  • Can occur even when total precipitation seems sufficient.

Goals

Problem Statement

​​

  • Existing drought indices fail to capture rapid soil moisture changes
     

  • Monthly-based observations → delays in detecting short-term flash droughts
     

  • Limits on-site response and management
     

 

Research Approach

​​

  • Develop a new index (CFDI) that integrates atmospheric circulation and land-surface conditions
     

  • Regionally tailored design suitable for Yeongdong region
     

 

Core Objectives

​​

  • Enable early detection of flash droughts
     

  • Strengthen practical response and decision-making support

​

 

Project Goal

​​

  • Develop a new drought index that integrates atmospheric dynamics.
     

  • Provides a framework that is practical for site-specific drought monitoring

Methods & Data

Analyzing the limitations of existing drought indices, we collected multivariate data to address their shortcomings and improve early detection performance. Based on the physical mechanisms of flash drought and statistical validity, we formally derive the CFDI (Composite Flash Drought Index) equation using a linear combination method. This approach aligns with methodological precedents set by established indices, aiming to create a robust tool for improved flash drought prediction.

Limitations of Existing Drought Indices and Directions for Improvement

​​

  1. They are strongly biased toward single climatic factors.
     

  2. They do not adequately represent time-scale differences in drought response.
     

  3. Their structure is insufficient for capturing the rapid and compound characteristics of flash droughts.
     

→ Integrated drought index 
multiple climatic variables + information across different time scales.

Rationale for the CFDI’s Linear Combination Based on Index Precedents

  1. Flash droughts is the outcome in an "additive" manner of  multiple independent atmospheric factors.
     

  2. Indices like SPEI, ESI, and FDCI all share the same linear sum.
     

  3. Linear combination after z-score standardization  is the statistical standard for multivariate climate indices. 
     

→ Define the CFDI as a weighted linear combination with SCFI, SII, AO, and ENSO 

Data Acquisition and CFDI Construction

  1. Synoptic Cyclone Frequency Index (SCFI)
     

  2. Synoptic Intensity Index (SII)
     

  3. Arctic Oscillation (AO)
     

  4. El Niño–Southern Oscillation (ENSO)

     

The CFDI was computed as: CFDI = −SCFI − SII − ENSO + AO

Limitations of Existing Drought Indices and Directions for Improvement

First, we introduce the existing drought indices. The numbers attached to each index represent the accumulation period of precipitation.​​

Standardized Precipitation Index


: Evaluates drought conditions by standardizing precipitation over a specific period.

 


Drought categories by SPI value
 

  • SPI ≥ 1.00 : Wet
     

  • 0.99 ~ -0.99 : Normal
     

  • -1.00 ~ -1.49 : Mild drought (Advisory)
     

  • -1.50 ~ -1.99 : Moderate drought (Watch)
     

  • SPI ≤ -2.00 : Severe drought (Warning)
     

  • SPI ≤ -2.00 for more than 20 days : Extreme drought (Emergency)

This study focuses on Gangneung, Gangwon Province, in August 2025, analyzing SPI3/6/9, SPEI3/6/9, and PN3/6/9 to evaluate how well each index captures flash drought characteristics.

​

​

​

​

[1] PCA-Based Structural Analysis of Drought Indices

​

 














PCA was performed to assess correlations and information redundancy among the drought indices.
 

  • The first principal component (PC1) explains over 85% of the total variance,
    → indicating that multiple indices repeat essentially the same pattern rather than providing distinct drought information.
     

  • All indices show very similar temporal behavior,
    → making it difficult to distinguish drought timing, intensity, or underlying causes.
     

  • However, real-world droughts result from complex interactions among multiple factors—precipitation, evapotranspiration, soil moisture, groundwater levels, temperature, and more.
     

  • Current indices rely primarily on precipitation, offering only a partial and simplified representation of droughts.
    → Highlighting the need for improved drought indices that integrate multiple contributing factors.




     

[2] Correlation Analysis Using Heatmap Visualization

​

​

​

​

​

​

​

​

​

​

​

​

​

  • Indices within the same category (SPI3–6–9, SPEI3–6–9, PN3–6–9) show very high correlations,
    → reflecting repeated information with only the accumulation period differing.
     

  • Cross-category correlations are not consistent:
     

    • SPI–SPEI: approx. 0.7–0.85
       

    • SPI–PN: approx. 0.85–0.95
       

    • SPEI–PN: approx. 0.65–0.85
      → showing that precipitation (SPI), evapotranspiration (SPEI), and soil moisture (PN) respond to different climatic drivers.
       

  • Significant differences appear across time scales (3, 6, and 9 months),
    → with some combinations—particularly those involving 9-month indices—showing weakened correlations.
    → Flash droughts intensify within days to weeks, so long-term indices struggle to capture such rapid changes.
     

As signals of precipitation deficits, increased evapotranspiration, and soil-moisture depletion become distributed across different indices,
→ no single index is capable of capturing the rapid, multi-factor nature of flash droughts.

Rationale for the CFDI’s Linear Combination Based on Index Precedents

[1] Atmospheric Dynamics: Linear Combination is Justified by Independent Mechanisms

 

Flash droughts are not caused by a single factor, but rather by the additive and independent influence of several mechanisms:

 

 

 

 

 

 

 

 

 

 

​

​

​

​

​

​

​

​

​

[2] Methodological Precedent: Linear Combination is Standard in Flash Drought Research

​

The use of a weighted linear sum of standardized variables is an already established and verified methodology in flash drought research globally.

 

​

​

​

 

 

 

 

 

 

 

​

 

 

 

 

 

 

 

The common structure of
standardizing multiple atmospheric factors → weighted linear sum
provides strong methodological justification for the CFDI's design.

​

​

 

 

[3] Statistical Basis: Z-score Linear Combination is the Standard for Multivariate Indices

 

 

Research on flash drought prediction has progressed toward combining various precursors to improve early detection performance. Standardizing variables with different units and scales (e.g., to z-scores) and then combining them via a weighted linear sum is a standardized approach.

 

 

Statistical Validity: Z-score Standardization
 

​

​​

​​

​

  • Eliminates Differences in Scale: By standardizing input variables with different units (e.g., hPa, °C, mm) into z-scores (mean 0, variance 1), differences in variable units and variance are eliminated.

  • Allows Interpretation of Weights: This enables the direct interpretation of weights calculated from regression analysis as the relative contribution of each variable, allowing for physically interpretable contribution scaling.
     

This methodology is common to FDCI, EDDI, and other multivariate index designs. Comprehensive reviews conclude that linear regression-based multivariate models are widely used and effective for practical forecasting.

​

Therefore, defining the CFDI as a weighted linear combination of standardized SCFI, SII, AO, and ENSO is justifiable from the perspective of atmospheric dynamics, statistics, and methodology. This construction method is reasonable and consistent with existing scientific accessibility and practices.​​​​​​​​

Standardized Precipitation-Evapotranspiration Index

​

P (Precipitation) – PET (Potential Evapotranspiration) 

→ Linear combination of two variables

​

Vicente-Serrano, S. M., Beguería, S., & López-Moreno, J. I. (2010). A multiscalar drought index sensitive to global warming: The Standardized Precipitation Evapotranspiration Index (SPEI). Journal of Climate, 23(7), 1696–1718.  

Screenshot 2025-11-27 151454.png

Data Acquisition and CFDI Construction

Directional Unification
& CFDI Calculation

[1] Synoptic Cyclone Frequency Index (SCFI)

 

Definition Number of low-pressure events (below a defined threshold) detected per day.

 

Data 1-hour interval sea level pressure (SLP) observations in Gangneung, August 2025.

 

Method

  1. Daily mean SLP was calculated.

  2. A statistical threshold was defined using: ​

​

      P = μ − 5σ​

​

to detect extreme events (probability ≈ 2.87×10⁻⁷)

 [ We used: μ=1013.2 hPa, σ=4.1 hPa   →   μ−5σ = 999.7 hPa ]

​

* Why μ − 5σ? 

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

 

​

​​​

​

​

​

​

​

​

​

​

 

3. If daily pressure fell below the threshold, SCFI = 1, otherwise SCFI = 0.
 

This approach is frequently used in previous studies
(Enz et al., 2023; Löptien, 2005; Crawford et al., 2021)
Future work may explore multiple thresholds or sensitivity analysis.

​

​

 

[2] Synoptic Intensity Index (SII)

 

Definition Captures the lowest pressure value per day if it falls below the threshold. Otherwise, SII = 0.

 

Uses the same SLP data as SCFI.
(μ=1013.2 hPa, σ=4.1 hPa → μ−5σ = 999.7 hPa
Thus using less than 999.7hpa per day)

​

​

 

[3] Arctic Oscillation (AO)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Daily AO index obtained from NOAA.
The observed value was directly applied to CFDI.

The Influence Pathway of AO: High-latitude wave–Upper-level circulation change → Reinforcement of the Korean Peninsula dry pattern

 

 

 

[4] El Niño–Southern Oscillation (ENSO)

 

 

 

 

 

 

 

 

 

 

 

 

​​​​

ONI-based ENSO index (3-month scale).
Since temporal resolution differs, randomized noise was added to approximate daily variations.

The Influence Pathway of ENSO: Tropical SST–Walker Circulation change → Reinforcement of the lower-level high-pressure system and reduction in precipitation

​

​

 

 

​

​

 

After standardizing all indices using z-scores, the variables were adjusted so that higher CFDI values correspond to stronger drought conditions:

​​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​Thus, higher CFDI values correspond to more intense drought conditions.

 

While the weighting coefficients (a, b, c, d) could ideally be calibrated using statistical approaches such as regression or logistic modeling, we adopted a simplified framework and assigned a = b = c = −1 and d = +1.
 

Accordingly, the CFDI was computed as:

​

CFDI = −SCFI − SII − ENSO + AO​

Setting the threshold at μ − 5σ identifies exceptionally low pressure events under the normal assumption — events that are extremely unlikely in a Gaussian world. This helps detect only the most extreme cyclones for SCFI calculation.

Screenshot 2025-11-27 001920.png
Screenshot 2025-11-27 001920.png

Note: The threshold μ−5σ was derived under a normality assumption. The normality assumption is an approximation; sensitivity to the threshold should be tested. We  assumed a normal distribution and used k=5 to identify extreme low-pressure events. The optimal value of k may be refined through future experimental and analytical evaluations.

Original Effect

More cyclones → drought relief

​

​

CFDI Direction

 - (negative)

Results

[1] CFDI during August 2025, Gangneung

​​​

To evaluate drought severity in the Gangneung–Yeongdong region during August 2025,
we developed the Combined Flash Drought Index (CFDI).
CFDI is calculated as a linear combination of four meteorological/oceanic variables:

​

CFDI = -1·SCFI + -1·SII + 1·AO + -1·ENSO

​​

​

​

​

​

​

 

 

 

​

​

​​​

​

Positive increases in CFDI corresponded with worsening drought conditions, aligning well with reported severe drought in late August—particularly the onset around August 21. 

​​

​

​

​

​

 

[2] What Makes the CFDI Unique

​

​

1. Captures Rapid Changes 

Time-series analysis of the CFDI revealed distinct spikes and drops. These patterns corresponded well with observed drought intensification, highlighting the CFDI’s ability to reflect real and rapidly evolving hydro-meteorological conditions.

​

2. Combined Multiple Drivers Into One Index

Instead of relying on separate indicators, CFDI merges precipitation deficits, atmospheric dynamics, and land-surface responses into a single interpretable value.

​

3. Improves Early Detection and Local Relevance

CFDI highlights rapid intensification signals earlier than SPI/SPEI, and its design can be tailored to regional climate characteristics—making it highly practical for site-specific drought monitoring.

​

4. Enhances Operational Use for Monitoring & Decision-Making

With clearer thresholds and more responsive behavior, CFDI provides actionable information for early warning, resource management, and drought preparedness.

About Us

References

◼ United Nations Office for Disaster Risk Reduction (UNDRR). (2023). Drought. Terminology. https://www.undrr.org/understanding-disaster-risk/terminology/hips/mh0401#:~:text=A%20drought%20is%20a%20period,month%20(WMO%2C%202023)
 

◼ National Integrated Drought Information System (NIDIS). (n.d.). What is drought? Flash drought. U.S. Drought Portal. https://www.drought.gov/what-is-drought/flash-drought
 

◼ 동아일보. (2025, August 26). [사회] ‘폭염·무강수 이어지며 돌발가뭄 확산’. https://www.donga.com/news/Society/article/all/20250826/132261749/2
 

◼ 한겨레. (2025, September 18). [사회] ‘돌발가뭄 경고음…한반도도 안전지대 아니다’. https://n.news.naver.com/mnews/article/028/0002748845

 

◼ Jun Li, Zhaoli Wang, Xushu Wu, Jie Chen, Shenglian Guo, Zhenxing Zhang, A new framework for tracking flash drought events in space and time, CATENA, Volume 194, 2020, 104763,ISSN 0341-8162,

https://doi.org/10.1016/j.catena.2020.104763.

 

◼ Jun Li, Zhaoli Wang, Xushu Wu, Jie Chen, Shenglian Guo, Zhenxing Zhang, A new framework for tracking flash drought events in space and time, CATENA, Volume 194, 2020, 104763, ISSN 0341-8162, 

https://doi.org/10.1016/j.catena.2020.104763.

 

◼ Lee, H. J., Nam, W. H., Yoon, D. H., Mark, D. S., Brian, D. W., & Otkin, J. A. (2022). 국내 돌발가뭄 사상의 메커니즘 분석. J. Korean Soc. Hazard Mitig, 22(3), 25-35.

 

◼ 박창균. "Spring precipitation variability over East Asia and its link to the drought-prone condition." 국내박사학위논문 서울대학교 대학원, 2021. 서울

 

◼ Christian, J. I., Hobbins, M., Hoell, A., Otkin, J. A., Ford, T. W., Cravens, A. E., Powlen, K. A., Wang, H., & Mishra, V. (2024). Flash drought: A state of the science review. WIREs Water, 11(3), e1714. https://doi.org/10.1002/wat2.1714

 

◼ Korea Meteorological Administration. (n.d.). Automated synoptic observing system (ASOS) data. Data Portal. https://data.kma.go.kr/data/grnd/selectAsosRltmList.do?pgmNo=36
 

◼ Korea Meteorological Administration. (n.d.). Radiosonde observation data. Data Portal. https://data.kma.go.kr/data/hr/selectRdsdRltmList.do?pgmNo=49
 

◼ Korea Meteorological Administration. (n.d.). Surface weather analysis charts. Data Portal. https://www.weather.go.kr/w/image/chart/analysis.do
 

◼ Water Resources Management Information System. (n.d.). Hydrological and meteorological data (river, dam, and reservoir discharge). Ministry of Land, Infrastructure and Transport. https://www.wamis.go.kr/wkw/flw_dubobsif.d

​

◼ Christian et al.(2024) - *Flash drought: A state of the science review

​

◼  He, S., Gao, Y., Li, F., Wang, H., & He, Y. (2017). Impact of Arctic Oscillation on the East Asian climate: A review. Earth-Science Reviews, 164, 48–62.

 

◼ Wang, B., Xiang, B., & Lee, J.-Y. (2013). Subtropical High predictability establishes a promising way for monsoon and tropical storm predictions. PNAS, 110(8), 2718–2722.

 

◼ Anderson, M. C., Hain, C., Otkin, J., Zhan, X., Mo, K., Svoboda, M., Wardlow, B., & Pimstein, A. (2011). An intercomparison of drought indicators based on thermal remote sensing and NLDAS multimodel simulations. Journal of Hydrometeorology, 12(5), 1230–1244.

 

◼ Byun, H. R., & Wilhite, D. A. (1999). Objective quantification of drought severity and duration. Journal of Climate, 12, 2747–2756.

 

◼ Vicente-Serrano, S. M., Beguería, S., & López-Moreno, J. I. (2010). A multiscalar drought index sensitive to global warming: The Standardized Precipitation Evapotranspiration Index (SPEI). Journal of Climate, 23(7), 1696–1718.

 

◼ Christian, J. I., Basara, J. B., Otkin, J. A., Hunt, E. D., & Xiao, X. (2024). A review of flash drought: Definitions, processes, and emerging predictive approaches. NOAA Institutional Repository.

 

◼ Ford, T. W., & Labosier, C. F. (2017). Meteorological and soil moisture-based indicators for flash drought early warning.

 

◼ Journal of Hydrometeorology, American Meteorological Society.

 

◼ Hobbins, M. T., Wood, A. W., Streubel, D. P., & Werner, K. (2016). The Evaporative Demand Drought Index. Journal of Hydrometeorology, American Meteorological Society.

 

◼ SciSpace (n.d.). Literature summaries on multivariate drought indicators and linear-combination methods. Retrieved from SciSpace database.

bottom of page