FDPAS – 2026  ·  National Conference of Frontier Development in Physics and Atmospheric Science

Variability of Pre-Monsoon Thermal Stress
over Coastal Balasore, Odisha
Decadal Climatology of Tmax, Heat Index & Wet-Bulb Temperature
using IMDAA Reanalysis (1981–2020)

Parthasarathi Tolaa, Samar Kumar Ghosea, Uma Charan Mohantyb

a P.G. Department of Atmospheric Science, Fakir Mohan University, Balasore, Odisha  |  b Centre for Climate Smart Agriculture, Siksha 'O' Anusandhan University, Bhubaneswar, Odisha

IMDAA 3-hourly · 12 km MAMJ · 1981–2020 4 Decades · D1–D4 Balasore District, Odisha Tmax  ·  HI  ·  TW
Scroll
01 — Abstract

Overview of the Study

Balasore district, situated on the northern coast of Odisha, faces a particular kind of heat problem during the pre-monsoon season (March through June; MAMJ). Continental temperatures peak while Bay of Bengal humidity rises sharply through May and June, and the combination produces thermal conditions that raw temperature alone cannot capture. This study examines daily maximum temperature (Tmax), Heat Index (HI), and Wet-Bulb Temperature (WBT) across Balasore district over four decades from 1981 to 2020, using IMDAA 3-hourly reanalysis (NCMRWF, 12 km resolution) cropped to the district administrative boundary.

HI is computed from co-located 3-hourly temperature and relative humidity at each timestep using the Rothfusz (1990) polynomial before daily aggregation, conserving the nonlinear temperature–moisture coupling that pre-aggregated fields obscure. Tmax shows a 2 to 4 degree inland-to-coast gradient in every decade. Interior grid cells warm at 0.2 to 0.4 degrees per decade; coastal cells warm slower. In May, spatially averaged HI places the district within the NWS Danger zone as a climatological mean across all four decades.

HI trends are largest along the coastal strip — the spatial contrary of the Tmax pattern — because coastal moisture amplifies the heat index response while simultaneously moderating raw temperature. WBT values in the 2011–2020 decade approach physiologically significant thresholds over coastal grid cells, with outdoor workers and fishing communities carrying the highest unmitigated exposure. Temperature-only metrics misidentify both the magnitude and location of maximum heat stress increase in this district.

02 — Introduction

Why Balasore, and Why Humidity Matters

Balasore district lies on the northern coast of Odisha, between approximately 21°–22°N, 86°–87°E. The pre-monsoon months bring the year's highest surface temperatures to this region. Bay of Bengal moisture rises sharply through May and June as the southwest monsoon approaches, so heat stress here is not simply a temperature problem — it is the joint product of temperature and humidity. A monitoring framework that uses only temperature cannot estimate the full hazard faced by the population.

Heat wave frequency over Odisha and eastern India has increased over recent decades (Rohini et al., 2016). Most published studies rely on coastal station networks or on global reanalyses that cannot resolve the land–sea moisture gradients operating at district scale. IMDAA, produced by NCMRWF at 12 km resolution with 3-hourly output, provides the spatial and temporal detail needed for a physically consistent district-scale climatology.

During MAMJ, relative humidity over coastal grid cells is consistently 10–20 percentage points higher than over interior cells at the same time of day. This moisture surplus, sourced from Bay of Bengal surface evaporation, separates the heat stress experience of coastal residents from that of interior residents — despite the interior recording higher raw Tmax.

The two compound indices examined in this study — Heat Index (HI) and Wet-Bulb Temperature (WBT) — are designed to capture exactly this coupling. The nonlinear Rothfusz equation means that at high relative humidity, even a modest increase in temperature produces a disproportionately large increase in HI. Bay of Bengal moisture simultaneously damps the Tmax response at the coast through oceanic thermal inertia and amplifies the HI response through this nonlinear term.

12 km IMDAA spatial resolution
3-hr Temporal resolution
40 yr Record length (1981–2020)
~22 Grid cells in Balasore district
03 — Study Area & Objectives

Balasore District and Research Goals

Study area map — Balasore district

Study Region

Balasore district is bounded approximately between 21°–22°N and 86°–87°E on the northern Odisha coastline. The district contains a persistent inland-to-coast thermal gradient driven by Bay of Bengal sea-surface influence. The Subarnarekha and Budhabalanga rivers create estuarine zones along the eastern coastal strip where near-surface relative humidity is sustained at high levels throughout MAMJ. Grid cells over these zones show the largest HI–Tmax differences in the analysis.

Research Objectives

  1. 01 Characterize the MAMJ decadal climatology of Tmax, Heat Index, and Wet-Bulb Temperature over Balasore district using IMDAA 3-hourly reanalysis at 12 km resolution.
  2. 02 Quantify where and how fast each index has changed across four decadal windows (1981–1990, 1991–2000, 2001–2010, 2011–2020), using the first decade as the reference baseline.
  3. 03 Identify whether compound heat stress trends are spatially consistent with raw temperature trends, and determine whether temperature-only thresholds mislocate the maximum heat stress increase within the district.
04 — Data & Methodology

Dataset, Index Computation and Analysis Framework

Dataset

IMDAA Reanalysis

IMDAA 3-hourly 2 m temperature (TMP_2m) and relative humidity (RH_2m) are extracted for each MAMJ month from 1981 to 2020. The domain is masked to the Balasore district administrative boundary using rioxarray.clip(all_touched=True), retaining any grid cell intersecting the boundary.

Tmax Extraction

Daily Maximum Temperature

Daily Tmax is the maximum of the eight 3-hourly T2m values per day. The timestep with the highest temperature is identified using argmax, and the co-located RH value at that same timestep is extracted for use in heat index calculations. This timestep-first pairing preserves the physically correct nonlinear co-variation between temperature and humidity.

Heat Index — Rothfusz (1990)

Heat Index Formula

HI = −42.379 + 2.049T + 10.143RH
    − 0.225T·RH − 6.84×10⁻³T²
    − 5.48×10⁻²RH² + 1.23×10⁻³T²·RH
    + 8.53×10⁻⁴T·RH² − 1.99×10⁻⁶T²·RH²
(T in °F; RH in %; output converted to °C)

HI is computed at each 3-hourly timestep from co-located T2m and RH2m before resampling to daily maxima. Computing HI from pre-aggregated daily fields would systematically underestimate peak apparent temperature because the equation is nonlinear in relative humidity.

Wet-Bulb Temperature — Stull (2011)

WBT Approximation

WBT is derived from T2m and RH2m at each 3-hourly timestep following the Stull (2011) empirical approximation. Daily maximum WBT is then retained. The 35°C WBT threshold from Sherwood and Huber (2010) is used as the physiological survivability reference for assessing coastal risk.

Sherwood & Huber (2010): sustained WBT ≥ 35°C is the approximate upper limit of human survivability under prolonged outdoor exposure.
Decadal Framework

Four Decades

The 40-year record is divided into four equal decadal windows: D1 (1981–1990) serves as the reference baseline. D2 (1991–2000), D3 (2001–2010), and D4 (2011–2020) are assessed as anomalies relative to D1. All anomaly maps and decade comparisons are computed on this basis.

Trend Estimation

Theil–Sen's Slope

Spatial trends at each grid cell are estimated using Theil–Sen's slope (Sen, 1968) applied to the full 1981–2020 annual series. This non-parametric estimator is robust to outliers and is used for all spatial trend maps. Results are expressed in degrees Celsius per decade.

05 — Results

Scientific Analysis and Interpretation

All three indices are examined separately across the four decadal windows. Select an index below to explore the spatial patterns, temporal trends, and distributional changes.

Spatial Warming Pattern and Decadal Change

A persistent inland-to-coast Tmax gradient of 2 to 4°C characterises Balasore in every decade. The interior-western cells are consistently warmer than the eastern coastal strip. This gradient is widest in May, when the pre-monsoon heat trough lies deepest over the interior and sea breezes moderate temperatures along the coast.

Tmax · Spatial Anomaly
Decadal Tmax Anomaly relative to D1 (1981–1990)
Tmax anomaly vs D1

Interior-western cells show warming of 0.3–0.6°C in D2 and 0.6–1.0°C in D3 and D4. Coastal cells record the smallest anomalies in each comparison period. The inland-to-coast thermal contrast within the district therefore widens over the study period — not because the coast is cooling, but because the interior warms faster.

Interior warming: 0.2–0.4°C decade⁻¹. Land–sea Tmax contrast widens from D1 to D4.
Tmax · Trend Map
Sen's Slope of MAMJ Tmax (°C decade⁻¹) · 1981–2020
Tmax Sen slope trend map

Theil–Sen's slope confirms interior warming of 0.2 to 0.4°C per decade. The coastal strip shows lower trend magnitudes, consistent with oceanic thermal inertia dampening surface warming along the Bay of Bengal shoreline. Stippling marks cells where the trend is statistically significant at the 95% confidence level.

Fastest warming is in the interior-west. Compare this spatial pattern against HI trend (Fig. 8) — they are opposite.
Distributional Shift Across Decades

The shift in Tmax is not confined to the mean — the full distribution has moved upward. In May, the D4 lower quartile lies close to the D1 upper quartile, meaning a below-median May in the most recent decade is about as warm as an above-median May in the 1981–1990 baseline.

Tmax · Distribution
MAMJ Tmax Distributional Shift by Decade
Tmax decade boxplot overlay

In May, the D4 lower quartile approaches the D1 upper quartile. The entire distribution has shifted upward — not merely the extreme tail. This is the key result for heat risk assessment: median days are now as hot as extreme days were four decades ago.

Tmax · PDF Shift
Kernel Density Estimates of MAMJ Tmax — D1 to D4
Tmax PDF shift D1 to D4

Rightward shift and upper-tail widening from D1 to D4 confirm a full distributional change, not only a mean shift. The D4 curve is displaced to higher values relative to D1 with a heavier upper tail — meaning more frequent extreme-heat days even beyond the shift in the median.

Distribution shift confirmed: rightward displacement + heavier upper tail in D4 vs D1.
Temporal Trend and Extreme-Day Frequency
Tmax · Hot Days · Interactive
Annual MAMJ Hot Days (Tmax ≥ 40°C) Frequency · 1981–2020
Loading…

The annual count of MAMJ days exceeding 40°C shows an upward trend across the 40-year record. Year-to-year variability is visible, with notable peaks coinciding with documented heat wave years over eastern India. This is among the most policy-relevant results in the analysis, directly quantifying how the frequency of extreme days has changed.

Tmax · Hovmöller · Interactive
Hovmöller Diagram — Year × Month MAMJ Tmax Anomaly
Loading…

This is the only plot in the analysis that simultaneously shows the seasonal progression of the MAMJ heat signal and the inter-decadal warming in a single frame. The secular warming from D1 toward D4 is clear in the cross-decadal direction, while the month-by-month seasonal structure within each year is visible on the vertical axis.

Monthly Mean Tmax — Four Decades Compared
Tmax · Monthly Bar · Interactive
Mean Daily Tmax per Month — 4 Decades Compared (MAMJ, 1981–2020)
Loading…

Every month shows D4 bars sitting above or near the 1981–2020 climatological mean (dashed line). April and May record the highest absolute values across all decades. March shows the largest spread between D1 and D3, while June shows the smallest inter-decadal range — consistent with monsoon-onset moisture moderating the temperature response in that month.

HI Climatology and Coastal Amplification

HI exceeds Tmax across the whole district in every decade and every month. The excess is largest along the Bay of Bengal coastal strip where relative humidity is highest. The inland-to-coast HI gradient is much smaller than the Tmax gradient, because coastal humidity offsets the inland temperature advantage and raises HI values closer to those of the interior.

HI · Climatology · 1981–2020
MAMJ Mean Heat Index — Full Period Baseline
HI climatology map

HI values are uniformly higher than Tmax at every grid cell, particularly along the coastal strip. The spatial pattern differs from the Tmax climatology: the inland-to-coast gradient is compressed because coastal humidity elevates HI at the eastern cells. For May, spatially averaged HI sits in the NWS Danger category (≥ 41°C) as a climatological mean across all four decades.

May HI places the district in NWS Danger category as a climatological mean — not as a rare event.
HI · Anomaly · vs D1
MAMJ HI Anomaly (°C) for D2, D3, D4 relative to D1
HI anomaly 4-panel

At coastal grid cells, D4 HI anomalies are 1.5 to 2 times the magnitude of the corresponding Tmax anomalies at the same locations. This is the central finding of the HI analysis: the same Bay of Bengal moisture that slows coastal Tmax warming amplifies the HI response through the nonlinear humidity term in the Rothfusz equation.

Coastal HI anomalies in D4 are 1.5–2× larger than Tmax anomalies at the same grid cells.
HI Trend Pattern — The Spatial Reversal

The most scientifically important spatial result in this study is the reversal between the Tmax trend pattern and the HI trend pattern. The zone with the fastest Tmax increase and the zone with the fastest HI increase are on opposite sides of the district.

HI · Trend Map · Sen's Slope
Sen's Slope of MAMJ HI (°C decade⁻¹) · 1981–2020
HI trend map

The HI trend is largest along the coastal strip — the spatial opposite of the Tmax trend map. Bay of Bengal moisture simultaneously damps the Tmax response at the coast through oceanic thermal inertia and amplifies the HI response through the nonlinear humidity dependency of the Rothfusz equation. A heat action plan built on Tmax thresholds will direct resources toward the interior, while the coastal population — which experiences the larger increase in compound heat stress — goes underserved.

"Tmax warms fastest inland. HI warms fastest at the coast." — The spatial reversal is driven by Bay of Bengal moisture.
HI · Distribution
MAMJ HI Distributional Shift by Decade
HI decade boxplots

The HI distributional shift mirrors the Tmax shift but at higher absolute values and with a larger inter-decadal spread. The D4 distribution is shifted toward and within the NWS Danger category across all months, with May showing the most extreme upper-tail values.

NWS Danger Category — Threshold Exceedance

The NWS Danger threshold (HI ≥ 41°C) marks conditions where heat stroke is possible with prolonged exposure. The frequency and spatial distribution of days exceeding this threshold directly quantifies where and how often the district enters dangerous heat stress territory.

HI · Danger Exceedance · Spatial
Mean Annual MAMJ Days with HI ≥ 41°C (NWS Danger)
HI Danger exceedance map

The spatial map of Danger-category day frequency shows where within the district the population is most frequently exposed to HI ≥ 41°C. Coastal cells, despite lower Tmax, should show higher exceedance counts due to sustained humidity — this is the clearest spatial proof that temperature-only thresholds mislocate the primary risk zone.

HI · Danger Days
Annual Count of MAMJ Danger-Level HI Days · 1981–2020
HI Danger exceedance time series

The upward trend in annual Danger-category day count quantifies how rapidly the district is spending more time in conditions hazardous to human health. June shows a progressive increase across decades without a comparable Tmax signal, identifying increasing monsoon-onset moisture loading as an independent driver of heat stress in that month.

June HI rises across D1→D4 without a matching June Tmax signal. Monsoon-onset moisture loading is an independent heat stress driver.
WBT Climatology and Physiological Context

Wet-Bulb Temperature represents the lowest temperature to which air can be cooled by evaporating water into it. For human physiology, WBT is the more direct measure of heat stress than either dry-bulb temperature or heat index, because the body's cooling mechanism — sweat evaporation — becomes ineffective when WBT is high, regardless of dry-bulb temperature alone.

TW · Climatology · 1981–2020
MAMJ Mean Wet-Bulb Temperature — Full Period Baseline
WBT climatology map

The WBT climatology shows the highest values along the coastal strip and over the estuarine zones of the Subarnarekha and Budhabalanga rivers, where sustained Bay of Bengal moisture keeps near-surface RH elevated throughout MAMJ. Interior cells record lower WBT despite higher dry-bulb temperatures, because lower relative humidity allows greater evaporative cooling.

TW · Anomaly · vs D1
MAMJ WBT Anomaly for D2, D3, D4 relative to D1
WBT anomaly vs D1

WBT anomalies in D4 are concentrated along the coastal strip, consistent with the HI pattern. The estuarine zones near the river mouths show the largest anomalies, reflecting the combination of warming temperatures and increasing moisture from Bay of Bengal SST warming. The pattern differs from the Tmax anomaly spatial structure.

WBT Trend and Distributional Shift
TW · Trend Map · Sen's Slope
Sen's Slope of MAMJ WBT (°C decade⁻¹) · 1981–2020
WBT trend map

The WBT trend pattern mirrors the HI trend — coastal strip shows the highest trend magnitudes, the spatial inverse of Tmax. This spatial consistency between HI and WBT trends confirms that the moisture amplification pathway, rather than the temperature pathway, drives the compound heat stress signal along the Balasore coast.

TW · Distribution
MAMJ WBT Distributional Shift by Decade
WBT decade boxplots

WBT values in the D4 decade approach physiologically significant levels during peak MAMJ months over coastal grid cells. Sherwood and Huber (2010) place 35°C WBT as an approximate survivability ceiling under sustained exposure. District-level averages smooth over the hottest local patches in the estuarine zones, so the actual risk at sub-district scale is likely higher than area averages suggest.

WBT in D4 approaches the 35°C physiological limit (Sherwood & Huber, 2010) at coastal grid cells during peak MAMJ hours.
Occupational Risk — WBT Threshold Exceedance

WBT thresholds carry direct occupational implications. Outdoor agricultural workers and fishing communities operating during peak MAMJ hours face the highest unmitigated exposure, without the option of air-conditioned shelter. The exceedance maps identify where within Balasore this population-level risk is most acute.

TW · High Risk Exceedance · Spatial
Mean Annual MAMJ Days with WBT ≥ 28°C
WBT exceedance map 28C

WBT ≥ 28°C represents a High Risk threshold for outdoor workers. The spatial map shows where in the district this threshold is exceeded most frequently during MAMJ, guiding where heat action planning resources should be directed. Coastal and estuarine cells are expected to show the highest exceedance counts.

06 — Key Summary

Principal Findings

Six conclusions drawn from the IMDAA 3-hourly data over Balasore district for MAMJ 1981–2020.

a
Tmax warming is spatially uneven. Interior-western cells warm at 0.2 to 0.4°C per decade. Coastal cells warm slower, widening the pre-existing land–sea thermal contrast across four decades.
b
The MAMJ Tmax distribution shifts upward from D1 to D4. In May, the D4 lower quartile approaches the D1 upper quartile — indicating a shift of the full distribution, not only the extreme tail.
c
May HI places the district in the NWS Danger category as a climatological mean across all four decades. June HI rises progressively without a comparable Tmax signal, identifying monsoon-onset moisture loading as an independent heat stress driver.
d
HI Sen's slopes are largest at the coastal strip, opposite to the Tmax trend pattern. Humidity amplification through the nonlinear Rothfusz equation produces coastal HI anomalies 1.5 to 2 times larger than Tmax anomalies at the same locations.
e
WBT trends in D4 approach physiologically significant levels in the coastal zone. The highest unmitigated exposure risk falls on outdoor agricultural workers and fishing communities operating during peak MAMJ hours.
f
Tmax-only heat alert criteria misidentify both the location and magnitude of maximum heat stress increase. Humidity-inclusive thresholds, differentiated between coastal and interior zones, are needed for effective heat action plan design in Balasore.
07 — Future Directions

Next Steps and Open Questions

Direction 01

Station-Level Validation

Validation of IMDAA-derived heat stress fields against in-situ surface observations within Balasore district is needed before these climatologies are applied in operational settings. IMDAA at 12 km resolution cannot fully resolve the fine-scale estuarine moisture gradients near the Subarnarekha and Budhabalanga river mouths.

Direction 02

CMIP6 Future Projections

Extension to bias-corrected CMIP6 regional downscaling would provide a future risk horizon for coastal Odisha under different emissions scenarios. IMDAA serves as the observational benchmark against which such downscaled products should be evaluated for heat stress consistency.

Direction 03

Solar Radiation Load

WBT in this study is derived from temperature and relative humidity following Stull (2011) and does not account for solar radiation load. Outdoor physiological WBT under direct sunlight is higher than the values shown here. Integration of WBGT with the radiation term would improve the occupational risk assessment.

Direction 04

Compound Event Analysis

Systematic analysis of compound hot-humid events — days where Tmax exceeds a thermal threshold AND RH exceeds a humidity threshold simultaneously — would directly quantify how the frequency of the most dangerous conditions has changed across decades and is projected to change further.

08 — References

Literature Cited

  1. 1 NCMRWF, 2021: IMDAA Regional Reanalysis: Scientific Documentation. National Centre for Medium Range Weather Forecasting, Ministry of Earth Sciences, India. Tech. Rep. NCMRWF/TR/2021-01.
  2. 2 Gogoi, P. P., Vinoj, V., Swain, D., et al., 2019: Land use and land cover change effect on surface temperature over Eastern India. Scientific Reports, 9, 8859.
  3. 3 Naveena, N., Satyanarayana, G. C., Rao, D. V. B., Srinivas, D., and Madhusudana Rao, K., 2021: An emerging threat of heat waves in a humid climate of Hyderabad, India. Natural Hazards, 105, 1359–1373.
  4. 4 Rohini, P., Rajeevan, M., and Srivastava, A. K., 2016: On the variability and increasing trends of heat waves over India. Scientific Reports, 6, 26153.
  5. 5 Rothfusz, L. P., 1990: The heat index equation. NWS Southern Region Tech. Attachment SR/SSD 90-23.
  6. 6 Roxy, M. K., and Coauthors, 2020: A reduction in marine primary productivity driven by rapid warming over the tropical Indian Ocean. Geophysical Research Letters, 43, 826–833.
  7. 7 Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall's tau. Journal of the American Statistical Association, 63, 1379–1389.
  8. 8 Sherwood, S. C., and Huber, M., 2010: An adaptability limit to climate change due to heat stress. Proceedings of the National Academy of Sciences, 107, 9552–9555.
  9. 9 Stull, R., 2011: Wet-bulb temperature from relative humidity and air temperature. Journal of Applied Meteorology and Climatology, 50, 2267–2269.