Climate-sensitive urban energy poverty indicator framework

Conceptual framework

The proposed approach, termed the Climate-sensitive Urban Energy Poverty Indicator Framework, consists of five indicators, illustrated in Fig. 5 and described as follows: (1) the Energy Gap Indicator, which quantifies the relative difference between the energy required by the household to achieve adequate hygrothermal comfort conditions and the energy actually consumed; (2) the EEPR, which assesses the proportion of the household’s residual income needed to cover the dwelling’s actual energy consumption; (3) the Climate Energy Demand Gap Indicator, which estimates the gap in energy demand by considering the dwelling’s LCZ in comparison to a rural reference zone; (4) the Basic Equipment Deficiencies Indicator, which identifies the lack of essential cooling equipment and thermal insulation during summer periods; and (5) the Thermal Discomfort Exposure Indicator, which quantifies the percentage of time during which indoor conditions fall outside adequate hygrothermal comfort ranges.

**Fig. 5: Conceptual scheme of the climate-sensitive urban energy poverty indicator framework.**

Study area

Temuco (38°44′ S, 72°35′ W), with a total communal population of approximately 292,000 inhabitants according to the 2024 Census⁵², is one of the main urban centers in southern Chile and has experienced a sustained process of urban expansion characterized by heterogeneous densification and pronounced socio-spatial fragmentation. The city is located in an intermediate depression surrounded by mountain ranges, a topographic condition that limits natural ventilation and favors heat accumulation in the urban environment. This study focuses on households located within the urban area of Temuco, according to the Plan Regulador Temuco, PCR-02⁵³.

According to the Köppen climate classification, Temuco has a temperate oceanic climate (Cfb) with dry summer influence. However, over the last decade, a sustained increase in summer temperatures has been observed, reaching a maximum of 40.2 °C in 2019 (~2 °C above values recorded in previous decades)⁵⁴. Under current climate change projections, increasing temperatures are expected in the region, potentially intensifying thermal conditions within the urban environment. Combined with recent changes in land use and urban morphology, these factors contribute to the development of Urban Heat Islands (UHI), as documented in previous studies^55,56,57.

The coexistence of highly densified areas, zones with discontinuous vegetation cover, and peri-urban areas with different heat dissipation capacities generates a complex microclimatic mosaic, which is relevant for analyzing residential thermal exposure during extreme heat events. In addition, structural conditions of the housing stock increase household thermal vulnerability. In Temuco, approximately 80.0% of households rely on firewood for heating, reflecting a traditional energy use pattern and a low penetration of active cooling systems such as air conditioning, thereby reinforcing the dependence on passive thermal conditioning strategies within the analyzed urban context⁵⁸.

Household data collection

Household-level information was collected through three complementary processes. First, in situ surveys were conducted in each dwelling (n = 53) to characterize socioeconomic and energy conditions. For the socioeconomic characterization, households were classified according to the Chilean Socioeconomic Group (SEG) framework developed by the Asociación de Investigadores de Mercado y Opinión Pública (AIM)⁵⁹, which is widely used in urban and energy-related studies. This framework distinguishes five socioeconomic groups, defined according to per capita income levels and predominant educational characteristics, and ordered from the highest to the lowest income level: ABC1, C2, C3, D, and E.

The socioeconomic survey included questions related to total household income, number of occupants, housing tenure (rent or mortgage), and expenditures associated with basic needs. These variables were subsequently used to estimate the residual income of each household. In addition, an energy survey was conducted to characterize energy consumption and HVAC equipment, identifying the availability of cooling systems (air conditioning and exhaust systems), energy use patterns, and recording the reported actual energy consumption (REC), including consumption associated with appliances and lighting (AL).

Second, a planimetric survey of each dwelling was carried out to document geometric properties, envelope characteristics, construction materiality, window-to-wall ratio, orientation, and shading conditions, with the aim of evaluating their influence on thermal performance through energy simulations.

Finally, indoor air temperature and relative humidity were recorded using Air-Q Internet of things (IoT) devices installed in the living room of each dwelling, approximately 1 m above floor level and away from windows, electronic equipment, and direct sources of heat or moisture⁶⁰. The sensors have nominal accuracies of ±0.5 °C for temperature and ±2.0% for relative humidity⁶¹. Before deployment, they were evaluated simultaneously for 48 h under controlled indoor conditions, showing mean differences of 0.5 °C and 1.1%, respectively. Measurements were recorded at 2-min intervals from December 21, 2024, to March 20, 2025, corresponding to the time of year with the highest frequency and intensity of heatwaves in south-central Chile⁶². To minimize the risk of data loss, the online connectivity of the Air-Q devices was checked twice a week throughout the monitoring period. In addition, each device stored a local backup on a microSD card, allowing the records to be preserved in the event of a temporary network disconnection. As part of the quality control process, the temperature and relative humidity records were subjected to a plausibility check before the analysis. Anomalous or physically inconsistent values, including those outside the operating range of the sensors, were identified and excluded. In cases of prolonged interruptions in online connectivity, the records stored locally on the devices’ microSD cards were used to recover the information and complete the data series. Only the valid records available for each dwelling were considered in the calculation of the indicator.

This summer period is characterized by high levels of solar radiation and reduced nocturnal thermal amplitude, conditions that are critical for assessing residential thermal discomfort. The collected data were used to estimate the percentage of time outside hygrothermal comfort ranges (19 °C–25 °C and 30%–70% RH) and to characterize the actual thermal exposure experienced by occupants during the analysis period. Data processing and statistical analyses were performed using Python 3.13.6 with the pandas, NumPy, and SciPy libraries.

All collected information was integrated into a georeferenced database, enabling each dwelling to be linked to its urban context and the corresponding environmental monitoring records.

LCZ classification

The urban morphological and thermal classification was developed using the LCZ scheme proposed by Stewart and Oke (2012), which enables the identification of homogeneous urban typologies in terms of morphology, surface cover, and thermal behavior³⁷. For its implementation, the LCZ Generator online tool (Ruhr University Bochum; https://lcz-generator.rub.de), developed by Demuzere et al. (2021), was used. This tool is based on a semi-automated classification system that combines remote sensing techniques and machine learning¹⁰.

The inputs used for LCZ classification included digital cartography, high-resolution orthophotographs, digital elevation models (DEM), and municipal cadastral data. Spatial data processing and cartographic analyses were performed using QGIS version 3.40.5. All datasets were processed within a common spatial reference system and refined through visual quality control to eliminate inconsistencies and duplications^63,64. The classification considered key structural and surface variables, including average building height, impervious surface fraction, vegetation cover, building density, and surface reflectance. Each variable was normalized and weighted according to LCZ standard criteria. The analysis was conducted at a spatial resolution of 100 × 100 m, selected to balance spatial accuracy and computational efficiency. As a result, seven predominant LCZ classes representative of Temuco’s urban morphological mosaic were identified. Based on the resulting LCZ map, the dwellings included in the study were georeferenced and assigned to their corresponding LCZ class according to their spatial location (Fig. 6).

**Fig. 6: Urban microclimatic classification of Temuco based on Local Climate Zones during the summer period (December 21, 2024–March 20, 2025).**

Urban temperature estimation

The climatic record was obtained from two official meteorological stations located in Temuco (Maquehue and Ñielol), complemented by a network of intra-urban stations installed in areas representative of different urban thermal zones (Fig. 6). These stations were strategically located in areas with contrasting building density and vegetation cover, with the aim of capturing thermal differences associated with urban morphology.

The mean air temperature used to generate climate files for stations associated with each LCZ was estimated using a prediction model developed by Martínez-Soto et al. (2024), which integrates data from both fixed meteorological stations and the intra-urban monitoring network in Temuco⁶⁵. This model enabled the extrapolation of point-based measurements across the entire urban area, providing representative average thermal values for each zone of Temuco under summer conditions. The model presents a mean residual prediction error of 1 °C.

Based on this urban climate characterization, the dwellings were digitally reconstructed and energy-modeled in DesignBuilder version 7.0.2.006 using the EnergyPlusV9-4-0 calculation engine. Each model incorporated the specific dimensions and construction characteristics recorded during the survey, including the materials of opaque and glazed elements, the window-to-wall ratio, and the orientation of the dwelling. In addition, an infiltration rate of 6 air changes per hour (ACH) was considered, in accordance with the criteria established in the Temuco Atmospheric Decontamination Plan⁶⁶. The simulations were conducted for the study period under hygrothermal comfort conditions, considering a temperature range between 19 °C and 25 °C for the activation of HVAC systems. For each dwelling, the existing heating system and the corresponding fuel type were incorporated according to the information collected through the surveys. To estimate the theoretical cooling demand under comparable conditions, a standardized air-conditioning system was assumed for all dwellings.

Each dwelling was simulated under two climatic scenarios: (1) rural reference conditions, using data from the Maquehue meteorological station; and (2) urban microclimatic conditions, using the climate file associated with the LCZ corresponding to the dwelling location. In this way, the theoretical energy demand (TED) required to maintain hygrothermal comfort conditions under each scenario was estimated.

Indicator framework formulation

The IPE–LCZ framework operates through five indicators that capture different dimensions of urban energy poverty (EP): energy gap, energy affordability, climate gap, equipment and insulation deficiencies, and exposure to hygrothermal discomfort. Each component is calculated at the household level ${\mathcal{i}}$ using data collected during the study period $t$.

$$\begin{array}{c}{{\rm{IPE}}-{\rm{LCZ}}}_{{\mathcal{i}},t}=\left({{EG}}_{{\mathcal{i}},t},{{EEPR}}_{{\mathcal{i}},t},{{CEDG}}_{{\mathcal{i}},t},{{BED}}_{{\mathcal{i}},t},{{TFC}}_{{\mathcal{i}},t}\right)\\ {{EG}}_{{\mathcal{i}},t},{{EEPR}}_{{\mathcal{i}},t},{{BED}}_{{\mathcal{i}},t},{{TFC}}_{{\mathcal{i}},t}\in [0,1]\\ {{CEDG}}_{{\mathcal{i}},t}\in [-\mathrm{1,1}]\end{array}$$

(1)

where ${\mathrm{IPE-LCZ}}_{{\mathcal{i}},t}$ represents the Climate-sensitive Urban Energy Poverty Indicator Framework for household ${{\mathcal{i}}}$ during period t. ${\mathrm{EG}}_{{\mathcal{i}},t}$ denotes the Energy Gap Indicator; ${\mathrm{EEPR}}_{{\mathcal{i}},t}$, the Economic Energy Poverty Ratio Indicator; ${\mathrm{CEDG}}_{{\mathcal{i}},t}$, the Climate Energy Demand Gap Indicator; ${\mathrm{BED}}_{{\mathcal{i}},t}$, the Basic Equipment Deficiencies Indicator; and ${\mathrm{TFC}}_{{\mathcal{i}},t}$, the Thermal Discomfort Exposure Indicator. The value ranges of the indicators are specified in Formula (1), and each indicator is detailed below.

The Energy Gap Indicator (EG) quantifies energy deprivation associated with an insufficient provision of energy relative to the TED required to maintain hygrothermal comfort conditions in the dwelling. This approach is related to situations in which households reduce their energy consumption due to economic or behavioral constraints, a phenomenon described in the literature as hidden energy poverty⁶⁷. The indicator is defined as:

$${{EG}}_{{\mathcal{i}},t}=\left\{\begin{array}{cc}0,\hfill &{\rm{if}}\,{{REC}}_{{\mathcal{i}},t}\ge {{MED}}_{{\mathcal{i}},t,j}\\ 1-\frac{{{REC}}_{{\mathcal{i}},t}}{{{MED}}_{{\mathcal{i}},t,j}},&{\rm{if}}\,{{REC}}_{{\mathcal{i}},t} < {{MED}}_{{\mathcal{i}},t,j}\end{array}\right.$$

(2)

where ${{REC}}_{{\mathcal{i}},t}$ corresponds to the actual energy consumption of household ${\mathcal{i}}$ during period $t$ and ${{MED}}_{{\mathcal{i}},t,j}$ is the modeled energy demand of household ${\mathcal{i}}$ under hygrothermal comfort conditions, estimated using DesignBuilder with the climate file associated with LCZ $j$ in which the dwelling is located. Both variables are expressed in $\left[\frac{{kWh}}{{m}^{2}t}\right]$.

The EEPR represents the economic dimension of energy poverty (EP) and assesses household energy affordability. Its formulation is based on the Ten Percent Rule (TPR), one of the most widely used criteria in EP studies, which establishes a relationship between household energy expenditure and available income to estimate the economic effort required to meet energy needs^68,69,70. In addition, household economic capacity is estimated using the residual income approach proposed by the Minimum Income Standard (MIS), which considers the income available after deducting minimum living costs and housing-related expenses^71,72. The indicator is defined as:

$${{EEPR}}_{{\mathcal{i}},t}=\left\{\begin{array}{cc}1,\hfill &{\rm{if}}\,{{REC}}_{{AI},{\mathcal{i}},t}\cdot {U}_{{AI},t}\ge {{RI}}_{{\mathcal{i}},t}\\ \frac{{{REC}}_{{AI},{\mathcal{i}},t}\cdot {U}_{{AI},t}}{{{RI}}_{{\mathcal{i}},t}},&{\rm{if}}\,{{REC}}_{{AI},{\mathcal{i}},t}\cdot {U}_{{AI},t} < {{RI}}_{{\mathcal{i}},t}\end{array}\right.$$

(3)

where ${\mathrm{CER}}_{{EI},{\mathcal{i}},t}$ corresponds to the actual energy consumption associated with appliances and lighting (AL) of household ${\mathcal{i}}$ during period $t$, expressed in ${\mathrm{CER}}_{{EI},{\mathcal{i}},t}$, and ${U}_{{AI},t}$ is the unit electricity price during the same period, expressed in [€/kWh]. For the calculation of the indicator, a reference unit electricity price of approximately €0.28/kWh (CLP $299/kWh) was used. This value was estimated from the average price per kWh observed in electricity bills from January 2025, a month within the study period. The product ${\mathrm{CER}}_{{EI},{\mathcal{i}},t}\cdot {U}_{{AI},t}$ represents the energy expenditure associated with these uses and is compared with the household residual income ${{RI}}_{{\mathcal{i}},t}$, expressed in [€/t].

Residual income is defined as:

$${{RI}}_{{\mathcal{i}},t}=\max (0;{{TI}}_{{\mathcal{i}},t}-{{HE}}_{{\mathcal{i}},t}-{3\cdot N}_{{\mathcal{i}},t}\cdot {{BE}}_{{pc}})$$

(4)

Where ${{TI}}_{{\mathcal{i}},t}$ corresponds to the total household income during the study period, ${{HE}}_{{\mathcal{i}},t}$ represents housing-related expenditures (rent or mortgage payments), ${N}_{{\mathcal{i}},t}$ is the equivalized number of household members, and ${{GB}}_{{pc}}$ corresponds to the per capita value of the national basic basket, estimated from the monthly average for the year 2025⁷³.

The Climate Energy Demand Gap Indicator (CEDG) captures the differential effect of the urban climate, particularly the thermal increase associated with urban structure or Urban Heat Island (UHI), on residential energy requirements. Several studies have documented that urban climatic conditions influence household energy demand, especially under urban warming scenarios^74,75.

This indicator compares the TED of the same dwelling under two climatic scenarios: rural reference conditions and urban microclimate conditions associated with the LCZ in which the dwelling is located. The indicator is defined as:

$${{CEDG}}_{{\mathcal{i}},t}=\left\{\begin{array}{cc}0,\hfill &{\rm{if}}\,{{TED}}_{{\mathcal{i}},t,j}=0\\ 1-\frac{{{TED}}_{{\mathcal{i}},t,r}}{{{TED}}_{{\mathcal{i}},t,j}},&{\rm{if}}\,{{TED}}_{{\mathcal{i}},t,j}\ne 0\hfill\end{array}\right.$$

(5)

where ${\mathrm{TED}}_{{\mathcal{i}},t,r}$ c corresponds to the theoretical energy demand of household ${\mathcal{i}}$ during period $t$, estimated under rural reference climatic conditions $r$, while ${{TED}}_{{\mathcal{i}},t,j}$ represents the simulated energy demand for the same dwelling using the climate file associated with LCZ $j$ where the household is located. Both demands were estimated using DesignBuilder under hygrothermal comfort conditions and are expressed in $[{kWh}/{m}^{2}\cdot t]$.

The Basic Equipment Deficiencies Indicator (BED) represents the technological and material dimension of energy poverty (EP) and evaluates the availability of basic climate control equipment within the household required to cope with summer thermal conditions^76,77. The inclusion of cooling and ventilation equipment is based on the MEPI approach, which characterizes EP in terms of access to energy services and household equipment⁷⁸. In addition, the absence of thermal insulation in key envelope elements (walls and roof) is incorporated, given its role in the thermal performance of the dwelling⁷⁹. The indicator is defined as:

$${{BED}}_{{\mathcal{i}},t}=\frac{\frac{{{CA}}_{{\mathcal{i}},t}+{{VA}}_{{\mathcal{i}},t}}{2}+\frac{{{ELTI}}_{{\mathcal{i}},t}}{{{TE}}_{{\mathcal{i}},t}}}{2}$$

(6)

Where ${\mathrm{CA}}_{{\mathcal{i}},t}$ and ${{\mathrm{VA}}}_{{\mathcal{i}},t}$ represent the absence of cooling and ventilation equipment, respectively, in household ${\mathcal{i}}$ during period $t$. Both variables are expressed as binary variables, taking a value of 1 when the equipment is absent and 0 when it is present.

On the other hand, ${\mathrm{TE}}_{{\mathcal{i}},t}$ corresponds to the total number of envelope elements considered in household ${\mathcal{i}}$ during period $t$ (walls and roof), while ${\mathrm{ELTI}}_{{\mathcal{i}},t}$ represents the number of these elements lacking thermal insulation. The value of ${\mathrm{TE}}_{{\mathcal{i}},t}$ depends on the dwelling configuration: for single-story dwellings, ${{TE}}_{{\mathcal{i}},t}=2$ (walls and roof), whereas for two-story dwellings ${\mathrm{TE}}_{{\mathcal{i}},t}=3$ (first-floor walls, second-floor walls, and roof).

The Thermal Discomfort Exposure Indicator (TFC) represents the observed thermal performance dimension of energy poverty (EP) and quantifies the proportion of time during which indoor hygrothermal conditions fall outside the acceptable comfort range over the analysis period.

To avoid interpreting situations without perceived thermal discomfort as vulnerability, the calculation of the indicator uses as a reference the mean proportion of time outside the comfort range, computed only for households that report experiencing thermal discomfort due to heat. In this way, the indicator combines instrumental evidence (temperature and relative humidity records) with occupant-reported perception. The indicator is defined as:

$${{TFC}}_{{\mathcal{i}},t}=\left\{\begin{array}{cc}0,\hfill &{\rm{if}}\,{N}_{T,{\mathcal{i}},t}=0\hfill\\ \frac{{N}_{F,{\mathcal{i}},t}}{{N}_{T,{\mathcal{i}},t}},&{\rm{if}}\,{N}_{T,{\mathcal{i}},t} > 0 \;\; \wedge \frac{{N}_{F,{\mathcal{i}},t}}{{N}_{T,{\mathcal{i}},t}} > \alpha \\ 0,\hfill &{\rm{if}}\,{N}_{T,{\mathcal{i}},t} > 0 \;\; \wedge \frac{{N}_{F,{\mathcal{i}},t}}{{N}_{T,{\mathcal{i}},t}}\le \alpha \end{array}\right.$$

(7)

Where ${N}_{T,{\mathcal{i}},t}$ corresponds to the total number of valid temperature and relative humidity records for household ${\mathcal{i}}$ during period $t$, while ${N}_{F,{\mathcal{i}},t}$ represents the number of records in which hygrothermal conditions fall outside the acceptable comfort range.

The threshold α is defined as the mean proportion of records outside the comfort range, calculated only among households that report thermal discomfort due to heat $d{tc}$, and is expressed as:

$${\alpha }=\frac{{\sum }_{{\mathcal{i}}{\cdot M}_{{\mathcal{i}},t,{dtc}}=1}\frac{{N}_{F,{\mathcal{i}},t}}{{N}_{T,{\mathcal{i}},t}}}{\sum _{{\mathcal{i}}}{M}_{{\mathcal{i}},t,{dtc}}}$$

(8)

where ${M}_{{\mathcal{i}},t,{dtc}}$ is a binary variable indicating whether the occupants of household ${\mathcal{i}}$ report thermal discomfort due to heat during period $t$.

Explainer: The CMIP7 emissions scenarios – and how they explore future climate change

London talks raise hopes for green shipping deal

Farmers Are Going Broke Amid Rising Gas Prices, While the Ag Secretary Profits Off Them

Threat of further violence looms after Mexican cartel rampage

‘Two-sided risk’ Medicare Advantage plans improve patient outcomes

An $18bn settlement – and Zuckerberg barely blinked. The tech titans must be stripped of their power, and soon | Jonathan Freedland