Perceived indoor thermal environment and depressive symptoms among older adults in the Japan Gerontological Evaluation Study

Abstract

Depression is a major global health problem and presents a significant health burden. Although abnormal indoor temperatures are known to be associated with adverse health effects, their link to depression is unclear, especially regarding indoor heat. This study aimed to examine the association between perceived indoor cold or heat and depressive symptoms among Japanese older adults. We used cross-sectional data from the 2022 Japan Gerontological Evaluation Study (JAGES), targeting independent older adults aged ≥ 65 years. The prevalence of depressive symptoms was the dependent variable, while the participants’ self-reported ability of their housing to prevent indoor cold or heat was the independent variable. Prevalence ratios (PRs) and 95% confidence intervals (CIs) were estimated using Poisson regression models with potential confounders as covariates. Additionally, we conducted a stratified analysis by geographical regions to explore regional differences. Of a total of 17,491 participants (49.4% male), 22.8% reported depressive symptoms. After adjusting for confounders, participants living in houses that could not prevent cold or heat had a 1.57 (95% CI = 1.45–1.71) times higher prevalence of depressive symptoms than those living in houses that could prevent cold or heat. In the stratified analysis by geographical regions, a significant association was observed in all areas except for Hokkaido, the northernmost area with the coldest climate. In conclusion, perceived indoor cold or heat was associated with an increased prevalence of depressive symptoms among older adults. Further research is expected to investigate the impact of improving the indoor thermal environment, such as by installing insulation, on the prevention of depression.

Introduction

Depressive disorder is a major global health problem¹, ranked as the second leading cause of years lived with disability (YLDs) and the 12th leading cause of disability-adjusted life-years (DALYs) according to the 2021 Global Burden of Disease Study². Depression in older adults is known to increase the risks of dementia, anxiety, suicide, frailty, and mortality, as well as cognitive and functional impairments^3–5. Therefore, depressive disorder presents a significant health burden, and its prevention is crucial for addressing an aging society.

Recently, there has been growing evidence that indoor cold and heat exposures are associated with adverse health effects^6,7. As awareness of the importance of the living environment is increasing, the World Health Organization (WHO) issued Housing and Health Guidelines in 2018, highlighting the health risks associated with abnormal indoor temperatures and recommending a minimum indoor temperature of 18 °C as well as protection from excessive indoor heat⁸. For example, low indoor temperatures are associated with increased risks of cardiovascular and respiratory diseases^9–12, poorer sleep quality¹³, reduced physical performance¹⁴, and lower self-rated health^7,15, whereas high indoor temperatures have been associated with sleep disturbance¹⁶, increased hospital admissions, and higher mortality rates¹⁷. Regarding mental and general health, perceived cold (which is a subjective measure) has been reported to increase risks of psychological distress¹⁸ and affect quality of life (QOL)¹⁹. Moreover, high indoor temperatures have been demonstrated to affect psychological distress and QOL²⁰. Considering that climate change will continue to be one of the major global problems, and increased outdoor temperatures caused by global warming are known to increase mental health issues such as suicide, aggression, and psychological distress and exhaustion²¹, further studies on the health effects of indoor heat are needed. However, evidence of the effects of indoor cold and heat exposures on specific psychiatric diseases such as depression is still lacking, especially in terms of perceived heat.

The indoor thermal environment continues to be a serious issue in Japan. Despite many regions in Japan having four distinct seasons, an estimated 39% of existing houses are uninsulated²². In addition, heating in Japan is mostly intermittent and typically only used in certain rooms such as living rooms and bedrooms. This is evident from the fact that the amount of energy spent on heating is only one-quarter of that in European and American countries, where continuous whole-home heating is standard²³. Perhaps for these reasons, more than 90% of Japanese households do not meet the WHO-recommended minimum indoor temperature of 18 °C in winter²⁴, which could be inferred to lead to various health problems. Indoor heat in Japanese houses during summer is also becoming a serious issue due to rising outdoor temperatures caused by climate change and insufficient insulation²⁵. In 2024, more than 90,000 people were transported by ambulance due to heat stroke, 38% of which occurred inside housings²⁶, highlighting the need to address not only indoor cold in winter but also indoor heat in summer. Moreover, Japan has been experiencing a super-aged society since 2007. Considering that depression in later stage of life increases the risks of dementia, anxiety, frailty, and mortality as mentioned earlier^3–5, and seniors aged over 60 years spend an average of 78% of their time at home²⁷, it is necessary to explore the effects that this issue may have on depression among Japanese older adults. This issue is particularly relevant for functionally independent older adults, who have the ability to make decisions about their own living environment, unlike functionally dependent older adults who typically live in designated facilities.

This study focuses on community dwelling, functionally independent older adults in Japan and explores the hypothesis that exposure to perceived indoor cold or heat is associated with a higher prevalence of depressive symptoms.

Methods

Setting and participants

This cross-sectional study was based on a self-reported questionnaire. We obtained data from the Japan Gerontological Evaluation Study (JAGES) conducted in 2022. The JAGES targeted community-dwelling older adults, aged ≥ 65 years, who were ineligible for long-term care insurance benefits, spanning 71 municipalities in Japan²⁸. Eight questionnaire modules were randomly distributed to cover a wide range of topics, including community resources and support (module A), medical services and medication (module B), life after disaster and COVID-19 (module C), oral health and nutrition (module D), housing and daily life (module E), living environment and life space (module F), physical activity and optimism (module G), and activities of daily living and pain (module H). Among these eight modules, we used module E. The questionnaires were distributed and retrieved via mail. Exclusion criteria included participants who did not provide consent, those with invalid or missing data on sex, age, height, or weight, those certified for long-term care, those with dependent activities of daily living (ADL), and those with incomplete data on housing environment or depressive symptoms. After applying these exclusion criteria and performing multiple imputations for missing data, 17,491 participants were selected for analysis (Fig. 1).

Fig. 1.

Flow diagram of the study participants’ selection.

Dependent variable

We used the prevalence of depressive symptoms as the dependent variable. Depressive symptoms were measured using the Geriatric Depression Scale 15 (GDS-15), which is a validated tool for screening depression in older adults. The GDS-15 is assessed using 15 items, including questions such as “Are you basically satisfied with your life?”, “Have you dropped many of your activities and interests?”, “Do you feel that your life is empty?”, and “Do you often get bored?”²⁹ Scores range from 0 to 15, with scores of 5 or higher indicating depressive symptoms and scores of 4 or lower indicating no depressive symptoms^30,31. The GDS-15 is in the public domain, and permission for its use is not required. More information is available at the following link: https://web.stanford.edu/~yesavage/GDS.html

Main exposure

Whether the participants’ housing could prevent heat or cold was used as the independent variable. From the multiple-selection question “Please choose all difficulties you are facing regarding your housing environment”, the option “Unable to prevent heat or cold” was used to assess whether the participants were living in cold or hot homes. This type of subjective indicator allows participants to respond based on their needs, and reflects the combined effects of indoor temperature, housing conditions, and personal preferences¹⁸.

Covariates

Based on previous studies, demographic factors, health status, socioeconomic factors, social factors, and environmental factors were selected as covariates^18,32–34. For demographic factors, sex assigned at birth (men or women) and age (65–69, 70–74, 75–79, 80–84, or ≥ 85 years) were selected. For health status, body mass index (< 18.5, 18.5–24.9, or ≥ 25 kg/m²), current disease (any disease or none), and walking time (< 60 or ≥ 60 min/day) were selected. Regarding current disease status, participants who answered “no” to having any disease under treatment or with lasting aftereffects were classified as having no current disease, whereas all others were classified as having at least one current disease. Subjective cognitive complaints (SCCs) were also included because dementia is associated with depression⁵, and cognitive decline is known to alter sensitivity to abnormal temperatures³⁵. SCCs were assessed using three questions from the Kihon Checklist³⁶: (1) “Do your family or friends point out your memory loss?” (2) “Are you able to look up phone numbers and make calls yourself?” (3) “Do you find yourself not knowing today’s date?” At least one response of “yes” to the first or third question or “no” to the second question was considered indicative of SCC³⁷. For socioeconomic factors, educational attainment (≤ 9, 10–12, or ≥ 13 years), equivalent income (< 2.00, 2.00–3.99, or ≥ 4.00 million JPY), and wealth (total assets of the household including savings, estate and stocks; < 5.00, 5.00–9.99, 10.00–49.99, or ≥ 50.00 million JPY) were selected since these are strongly related to both housing environment and depressive symptoms. The equivalent income was calculated by dividing the household income by the square root of the number of household members. For social factor, marital status (married, widowed, separated, unmarried, or others) was selected. For environmental factors, duration of residence (≤ 5, 5–9, 10–19, 20–29, 30–39, 40–49, or ≥ 50 years), house type (owned house, public rental house, private rental house, or others), and region were selected. House type was included since it has been reported to be associated with mortality risk³².

Statistical analysis

To evaluate the association between perceived indoor cold or heat and depressive symptoms, we employed Poisson regression with sandwich estimators for standard errors and estimated prevalence ratios (PRs) and 95% confidence intervals (CIs). We built the following three models: the crude model (unadjusted), model 1 (adjusted for sex and age), and model 2 (adjusted for all covariates). For the independent variable and covariates, we checked for collinearity based on variance inflation factor³⁸ and confirmed no higher multicollinearity among the variables. Under the assumption that the data were missing at random, we conducted multiple imputation by chained equations (MICE) including all variables used in the analysis to address potential selection bias, generating 20 imputed datasets^39,40. To diagnose convergence, we checked whether there were large variations between each imputation dataset or bias in the patterns. The estimates of each imputed dataset were combined using Rubin’s rules. For the sensitivity analysis, we conducted a complete case analysis to confirm the consistency of the results of multiple imputation data. Additionally, we conducted stratified analyses by geographical regions (each has a distinct climate: Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, and Kyushu) for the purpose of exploring regional differences in the association between perceived indoor cold or heat and depressive symptoms. Despite including possible confounders in the analysis, it is impossible to deny the possibility that unmeasured confounding factors may have affected the results. Therefore, we have also calculated the E-value of the PR to identify the strength of the unmeasured covariates that affected the estimates. Statistical analyses were performed using Stata MP version 17 (Stata Corp., College Station, TX, USA).

Ethical approval

The JAGES protocol and informed consent procedure were approved by the Ethics Committee of Chiba University (M10460) and Tohoku University (37582). Informed consent was obtained from all participants. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Results

Table 1 shows the characteristics of the participants after multiple imputations. Of a total of 17,491 participants selected (age: 74.5 ± 6.1 years [mean ± standard deviation], 49.4% male), 22.8% exhibited depressive symptoms. 5.1% of the participants were living in houses unable to prevent cold or heat, of which 41.7% had depressive symptoms, whereas 94.9% were living in houses that could prevent cold or heat, of which 21.8% had depressive symptoms. The participants’ characteristics before multiple imputation are presented in Supplementary Table 1. While post-imputation and pre-imputation data had similar distributions, relatively high proportions of missing data were observed in covariates such as equivalent income (10.7%) and wealth (16.0%).

Table 1.

Characteristics of study participants after multiple imputation (n = 17,491).

	Total		Depressive symptoms
			No (n = 13,501; 77.2%)		Yes (n = 3,990; 22.8%)
	n	%	n	%	n	%
Indoor thermal environment
Not cold or hot	16,607	94.9	12,986	78.2	3,621	21.8
Cold or hot	884	5.1	515	58.3	369	41.7
Sex
Men	8,643	49.4	6,673	77.2	1,970	22.8
Women	8,848	50.6	6,828	77.2	2,020	22.8
Age (years)
65–69	4,239	24.2	3,238	76.4	1,001	23.6
70–74	5,374	30.7	4,213	78.4	1,161	21.6
75–79	4,041	23.1	3,161	78.2	880	21.8
80–84	2,620	15.0	1,999	76.3	621	23.7
≥ 85	1,217	7.0	890	73.1	327	26.9
BMI
< 18.5	1,304	7.5	930	71.4	374	28.6
18.5–24.9	12,195	69.7	9,490	77.8	2,705	22.2
≥ 25	3,992	22.8	3,081	77.2	911	22.8
Educational attainment (years of education)
≤ 9	3,466	19.8	2,466	71.1	1,000	28.9
10–12	7,658	43.8	5,900	77.1	1,757	22.9
≥ 13	6,207	35.5	5,022	80.9	1,186	19.1
Others	160	0.9	113	70.6	47	29.4
Equivalent income (million JPY)
< 2.00	8,456	48.3	6,045	71.5	2,411	28.5
2.00–3.99	6,831	39.1	5,544	81.2	1,287	18.8
≥ 4.00	2,204	12.6	1,912	86.8	292	13.2
Wealth (million JPY)
< 5.00	4,730	27.0	3,226	68.2	1,504	31.8
5.00–9.99	2,823	16.1	2,102	74.4	722	25.6
10.00–49.99	7,309	41.8	5,909	80.9	1,399	19.1
≥ 50.00	2,629	15.0	2,264	86.1	365	13.9
Marital status
Married	13,008	74.4	10,356	79.6	2,652	20.4
Widowed	2,978	17.0	2,198	73.8	781	26.2
Separated	849	4.9	555	65.4	294	34.6
Unmarried	568	3.2	337	59.3	231	40.7
Others	88	0.5	55	63.0	32	37.0
Subjective cognitive complaints
No	11,963	68.4	9,840	82.2	2,123	17.8
Yes	5,528	31.6	3,661	66.2	1,867	33.8
Walking time
< 60 min	10,959	62.7	8,124	74.1	2,835	25.9
≥ 60 min	6,532	37.3	5,377	82.3	1,155	17.7
Current disease
No	3,366	19.2	2,802	83.2	564	16.8
Yes	14,125	80.8	10,699	75.7	3,426	24.3
House type
Owned house	15,862	90.7	12,401	78.2	3,461	21.8
Public rental house	674	3.9	450	66.8	224	33.2
Private rental house	594	3.4	401	67.6	192	32.4
Others	361	2.1	249	68.8	113	31.2
Duration of residence (years)
< 5	515	2.9	363	70.4	153	29.6
5–9	603	3.4	448	74.2	155	25.8
10–19	1,460	8.3	1,091	74.7	369	25.3
20–29	2,076	11.9	1,627	78.4	449	21.6
30–39	2,829	16.2	2,198	77.7	631	22.3
40–49	4,048	23.1	3,155	77.9	893	22.1
≥ 50	5,960	34.1	4,619	77.5	1,340	22.5

JPY, Japanese yen.

Table 2 presents the results of the modified Poisson regression analysis. A significant association between indoor cold or heat and depressive symptoms was observed in the crude model (PR = 1.91, 95% CI = 1.76–2.08). After adjusting for covariates, the association remained significant (models 1 & 2). In model 2, the participants living in houses that could not prevent cold or heat had 1.57 (95% CI = 1.45–1.71) times higher prevalence of depressive symptoms than those living in houses that could prevent cold and heat. E-value for the adjusted PR was 2.52. Supplementary Table 2 presents similar results obtained from the complete case analysis (Model 2: PR = 1.60; 95% CI = 1.45–1.76).

Table 2.

Association between indoor cold or heat and depressive symptoms (n = 17,491).

Indoor thermal environment	Crude model	Model 1a	Model 2b
	PR (95% CI)	Adjusted PR (95% CI)	Adjusted PR (95% CI)	E-valuec of the adjusted PR from Model 2
Not cold/hot	1.00 (Ref.)	1.00 (Ref.)	1.00 (Ref.)
Cold/hot	1.91 (1.76–2.08)	1.92 (1.77–2.09)	1.57 (1.45–1.71)	2.52

^aAdjusted for age and sex.

^bAdjusted for sex, age, body mass index, educational attainment, income, wealth, marital status, subjective cognitive complaints, walking time, current disease, house type, and duration of residence.

^cE-value presents the minimum strength of association on the risk ratio scale that an unmeasured confounder would need to have with both exposure and outcome to fully explain away the observed association conditional on included covariates.

Abbreviation: PR, prevalence ratio; 95% CI, 95% confidence intervals.

In the stratified analysis by geographical regions, a significant association between perceived indoor cold or heat and depressive symptoms was observed in those living in the following regions: Tohoku (PR = 1.58, 95% CI = 1.31–1.90), Kanto (PR = 1.64, 95% CI = 1.34–2.02), Chubu (PR = 1.57, 95% CI = 1.36–1.81), Kinki (PR = 1.43, 95% CI = 1.03–2.00), Chugoku (PR = 1.35, 95% CI = 1.01–1.78), and Kyushu (PR = 1.80, 95% CI = 1.36–2.41), in order from north to south (Supplementary Table 3). No significant association was observed in those living in Hokkaido (PR = 1.37, 95% CI = 0.93–2.03), the northernmost region with the coldest climate.

Discussion

This cross-sectional study of independent older adults demonstrated an association between perceived indoor cold or heat and depressive symptoms. The prevalence ratio was 1.91 in the crude model and 1.57 in model 2, which adjusted for all covariates. This indicates that 37% of the excess prevalence ratio was explained by the covariates included as potential confounders. In the stratified analysis by geographical region, a significant association was observed in all areas except Hokkaido, the northernmost region, with the strongest association found in Kyushu, the southernmost region. Since depressive symptoms are a well-established predictive factor for depression, our findings imply an association between perceived indoor cold or heat and depression⁴¹.

Our findings are consistent with previous studies regarding indoor cold, but also suggest a potential association between indoor heat and depression. In terms of indoor cold, previous studies have shown that living in cold housing significantly increases the risk of depression⁴². The inability to adequately warm homes, known as energy poverty, has been reported to increase the likelihood of depression/anxiety by about 50%⁴³. Indoor cold is known to be a more serious issue in regions with mild winters than in those with severe winter conditions, due to generally lower levels of insulation. For example, in Europe, winter mortality is greater in countries with milder climates than in those with more severe winter conditions⁴⁴. The lack of significant association observed in Hokkaido is consistent with this previous finding, and could be attributed the highest living room temperatures in Japan due to high insulation levels and the approach of heating the entire building continuously^24,33. Regarding indoor heat, to the best of our knowledge, there is limited evidence of the effects of perceived indoor heat on depression. The strongest association observed in Kyushu, which has hot and humid summers but mildly cold winters⁴⁵, implies that difficulties with indoor heat in the summer may also contribute to depressive symptoms, although it is impossible to determine if this is solely the effect of indoor heat. We believe this implication is important as it raises concerns that indoor heat caused by global warming may lead to increased depression in the future. Taken together, it could be implied that abnormal indoor thermal environments have effects on depression, especially in mild winter regions and severely hot regions.

Based on previous studies, several mechanisms can be hypothesized to explain the effects of indoor cold or heat on depression. In terms of cold housing, first, high systolic blood pressure⁴⁶ and brain vasoconstriction⁴⁷ caused by a cold environment have been suggested to lead to late-life depression through focal fiber tract disruption, altered functional connectivity, and altered regional brain function⁴⁸. Second, low ambient temperature affects the function of the hippocampus⁴⁹, which is associated with depression⁵⁰. On the other hand, in a hot environment, difficulties with brain cooling, oxygenation and increased blood flow⁵¹ may contribute to depressive symptoms and mental disorders⁵². Furthermore, hyperthermia increases the permeability of the blood–brain barrier (BBB)⁵³, which is known to lead to depression⁵⁴. Both cold and hot indoor environments are known to impair sleep quality^13,16, which is associated with increased depressive symptoms⁵⁵. It has been reported that bedroom temperatures in Japan are approximately 4 °C lower than living room temperatures²⁴, which is possibly leading to sleep disturbances and depressive symptoms.

Our findings suggest that improving thermal environments, even in mild winter regions, is necessary to prevent depression among community-dwelling, functionally independent older adults. Possible approaches include thermal insulation, heat supply, air-conditioning, ventilation^56,57, and relocation. However, the implementation and running costs of heating, air-conditioning, and ventilation systems can be prohibitively expensive. While this population has the autonomy to consider relocation, this option should be approached cautiously, as it is known to lead to relocation stress syndrome (RSS), characterized by confusion, depression, anxiety, apprehension, and loneliness in older adults^58,59. Moreover, ageing in place has been reported to improve the QOL in low-income people^60,61. Thus, insulation, which is associated with health benefits and reductions in healthcare expenditures^42,62, could be considered a potentially effective and practical approach to improving the indoor thermal environment. Importantly, low-income individuals are known to be at higher risk of heat-related mortality⁶³ because they are more likely to live in poorly insulated homes and less likely to afford insulation improvements. As such, public support for housing improvements among functionally independent older adults, especially those with low-income, is critical; otherwise, improvements in insulation could lead to further inequalities⁸.

Several limitations of this study should be acknowledged. First, reverse causation is a potential issue since this is a cross-sectional study. There is a possibility that depression leads to living in excessively cold or hot housing, contrary to our hypothesis. A cross-sectional design was the only option because the question item used to assess the exposure variable (“unable to prevent cold or heat”) was first introduced in the 2022 survey. As longitudinal data will become available from the 2025 survey onward, future cohort studies are expected to confirm the temporal relationship between indoor thermal environments and depressive symptoms. Second, the question option used as the independent variable, “Unable to prevent heat or cold,” cannot distinguish between heat and cold. Thus, it is impossible to determine whether the participants are struggling with indoor cold, indoor heat, or both. As this survey was conducted in the winter months (November–December), participants’ responses are more likely to reflect indoor cold rather than indoor heat. Although we inferred which of them might be the risk of depression by conducting stratified analysis based on geographical regions with distinct climates, it is necessary to measure indoor thermal environment using separate options for clearer insight. Third, this independent variable is binary and subjective; thus, there is no information on the extent to which the participants are affected by the abnormal thermal environment, or the actual indoor temperature. Subjective measures, such as the question used in this study, can capture perceived needs and reflect combined effects of temperature, housing conditions, and preferences¹⁸. However, responses may vary between occupants of the same house. Objective measures, although more consistent, require significant resources, such as cost and effort⁶⁴, and may not reflect time spent in specific rooms. Given the gap between perceived coldness and measured indoor temperatures³³, and that older adults may be less sensitive to cold yet more vulnerable to its health impacts³³, future studies should include both subjective and objective measures to validate and expand upon these findings. Fourth, there is a possibility of measurement bias since both the independent and dependent variables are self-reported. However, the dependent variable, the GDS-15, has been validated in previous studies and is widely used as a screening tool for depressive symptoms in older adults. While the independent variable has not been formally validated, potential misclassification is likely to be non-differential and would thus bias the results toward the null. Fifth, our data does not reflect a nationally representative sample because it only includes community-dwelling, functionally independent older adults. Although the 71 participating municipalities were selected to include all regions of the country, the sample does not reflect the balance between urban and rural areas. Sixth, unmeasured confounding may have affected the results. To assess the potential influence of unmeasured confounders, we calculated E-value of the prevalence ratio of depressive symptoms. The obtained E-value was 2.52, indicating that there is an extremely low possibility of unmeasured confounding factors that are strongly associated with both independent and dependent variables.

Conclusion

The present study revealed that perceived indoor cold or heat exposure is associated with greater prevalence of depressive symptoms. Further research is expected to investigate the impact of improving the indoor thermal environment, such as by installing insulation, on the prevention of depression.

Author contributions

MI: Conceptualization, Formal analysis, Methodology, Writing – original draft. A.K.: Conceptualization, Formal analysis, Writing – review and editing. M.H. Conceptualization, Methodology, Writing—review and editing. K.K.: Conceptualization, Methodology, Writing—review and editing. K.O.: Conceptualization, Methodology, Supervision, Writing—review and editing. K.T.: Conceptualization, Methodology, Supervision, Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This study used data from JAGES (the Japan Gerontological Evaluation Study). This study was supported by Grant-in-Aid for Scientific Research [19K02200, 20H00557, 20H03954, 20K02176, 20K10540, 20K13721, 20K19534, 21H00792, 21H03196, 21K02001, 21K10323, 21K11108, 21K17302, 21K17308, 21K17322, 22H00934, 22H03299, 22J00662, 22J01409, 22K01434, 22K04450, 22K10564, 22K11101, 22K13558, 22K17265, 22K17364, 22K17409, 23K16320, 23H00449, 23H03117, 23K19793, 23K21500, 23K24557] from JSPS (Japan Society for the Promotion of Science), Health Labour Sciences Research Grants [19FA1012, 19FA2001, 21FA1012,22FA2001, 22FA1010, 22FG2001], the Research Funding for Longevity Sciences from National Center for Geriatrics and Gerontology, Research Institute of Science and Technology for Society [JPMJOP183] from the Japan Science and Technology (JST), a grant from Japan Health Promotion & Fitness Foundation, contribution by Department of Active Ageing, Niigata University Graduate School of Medical and Dental Sciences (donated by Tokamachi city, Niigata), TMDU priority research areas grant and National Research Institute for Earth Science and Disaster Resilience. The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the respective funding organizations.

Data availability

The datasets used in this study from JAGES are not publicly available because of ethical and legal restrictions. However, the datasets used in this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-15922-9.