# DyME vulnerable populations

## Dataset

### Clim-recal data
Clim-recal applied bias correction method with the UK Climate Projections 2018 (UKCP18) dataset, providing more accurate climate projection from 2020 to 2100 for the Whole UK.

Currently, the clim-recal dataset includes **highest temperature**, **lowest temperature** and **precipitation**, the current spatial resolution of data points is **2.2KM**, the temporal resolution is **daily**.

Humidity data will be updated in the future.

### SPC data
The Synthetic Population Catalyst (SPC) data includes various [data sources](understanding_data_sources.qmd) and outputs as a single file. The data include socio-demographic, health, salary and daily activity data per person, and information about the venues where people conduct those activities.

The attributes included in this dataset can be found on page 46 of [this documentation](https://alan-turing-institute.github.io/uatk-spc/Synthetic-Population-Catalyst.pdf).

## Research Themes

How to predict the risk of heat exposure in the future combining the climate and demographic projection?

* [name= @Bowen] Comments: I wrote this research question as a provisional one, please feel free to edit/write down a new one here.

**Research question/obectives**:

TBD

## Proposed Methods

1. **Finding the risky area for heat exposure**

Defining the risky area for heat exposure may combine different climate indicators including the temperature, humidity etc., and the non-climate environment factors (i.e. shade of trees and buildings) could be considered if possible.

A reference [here](https://a816-dohbesp.nyc.gov/IndicatorPublic/data-explorer/climate/?id=2191#display=map) is the Heat Vulnerability Index of NYC.

2. **Finding the vulnerable populations for heat explosure**

The vulnerable populations can be selected from the SPC dataset as it already includes various health information for individuals. The SPC data includes socio-economic and accommodation characteristics.

A demo has been presented to show the pipeline for how to connect the population data and clim-recal dataset in Glasgow. If there are any comments/suggestions, please contact @Bowen

This could be combined with step one to generate a more comprehensive HVI.

3. **Assess the interaction between risky areas and vulnerable people**

Not has been discussed yet.

From previous discussions between @Ruth and @Bowen: one relatively easy way is to integrate the population into the area at a specific geographical level (i.e. MSOA), similar to the Glasgow demo, therefore, we could get a Heat Vulnerability Index for each area. Another possible way is applying mobility models such as the ABM model or QUANT model to explore the more detailed travel behaviour and its related heat exposure risks. By applying the latter method, we could have a much higher spatial and temporal resolution for the heat exposure risk predictions.

## Litureture Review and Thinking

Key Questions:

### 1. **How to defining the Heat Vulnerability Index?**

A bunch of previous research has answered this question. The variables for defining the HVI could be divided into three categories:

a. Demographic/social-economic factors like poverty, education, ethnicity, living alone, elderly, healthcare, air conditions, GDP per capita

b. Environmental factors like building density, road density, and greenspace area, (shading has been mentioned as the potential factor, but I have not seen any papers used that)

c. Climate factors like hot days, consecutive hot days with Tmax > 30°C and Tmin > 22°C, land surface temperature.

**Selection of Key Parameters:**

Extremely hot weather: various between 30-37.5 °C based on the location, or above 99% percentile of daily Tmax across the year
Heat wave: nomarlly be defined as more than consecutive 3 days

P.S. I noticed that many of the studies selected indicators a bit randomly (e.g., using PCA) or based on the kinds of data they had. I have not found many studies that have set a 'threshold' for temperature or other characteristics or clearly stated the importance of different factors.

**Possible research gap to fill: A dynamic Heat vulnerability index combining all three factors?**

Repesentive literatures:

*Development of a heat vulnerability index for New York State* [(link)](https://doi.org/10.1016/j.puhe.2017.09.006)

*The Construction and Validation of the Heat Vulnerability Index, a Review* [(link)](https://www.mdpi.com/1660-4601/12/7/7220)

*Developing an applied extreme heat vulnerability index utilizing socioeconomic and environmental data* [(link)](https://doi.org/10.1016/j.apgeog.2012.04.006)

### 2. **How to measure/estimate the heat exposure for indivudals?**

Previous research about measuring/estimating individual level heat exposure mainly focuses on occupational heat exposure (e.g., construction workers, farmers or metal industries workers) and indoor heat exposure as people would stay at the same place for a long time in these scenarios.The mainstream estimation method is regression prediction based on measurements from physical equipment installed/worn on site.

Activity-based estimation for heat exposure has been relatively rare in previous research; the few cases here are travel to calculate the total amount exposure in hot weather [ (link1)](https://www.sciencedirect.com/science/article/pii/S2214140515006866?via%3Dihub) [(link2)](https://www.sciencedirect.com/science/article/pii/S1470160X1300349X?via%3Dihub). Here is also a research using ABM to estitmte the sum of heat-exposure for all day [ (link1)](https://www.mdpi.com/2413-8851/2/2/36). Similar analyses for large populations are feasible with the activity-based method we are trying to add to the SPC dataset. In addtion, a key question we need to consider here: What indicators are we trying to output from this modelling to assess heat exposure? Previous research has used concepts like total degree minutes (time * temp) or extreme degree minutes (sum of time when temp > therehold) to measure it.

**A possible research gap** here is how to link the amount of heat exposure to the specific health impact by combining the personal healthy status provided in the SPC data, but it is not easy as it requires more clinical evidence to prove the result.

Repesentive literatures:

*Heat exposure during non-motorized travel: Implications for transportation policy under climate change*[ (link)](https://www.sciencedirect.com/science/article/pii/S2214140515006866?via%3Dihub)

*Opportunities and Challenges for Personal Heat Exposure Research*[ (link)](https://ehp.niehs.nih.gov/doi/full/10.1289/EHP556)

*Modeling Exposure to Heat Stress with a Simple Urban Model*[ (link)](https://www.mdpi.com/2413-8851/2/2/36)

### 3. **What long-term impact about climate change can be predicted?**

The current research has stopped at roughly estimating how many people will be affected by heat waves in the future by combining climate prediction and demographical prediction[ (link)](https://www.nature.com/articles/nclimate2631). Some research tying to predict the heat-related excess mortality under climate change scenarios mostly rely on the medical evidance [ (link)](https://www.nature.com/articles/s41467-021-21305-1). However, these estimations are rough and far away from the individual level. It is hard to obtain detailed and reliable demographical/socioeconomic projects in the long term. SPC data and clim-recal provide opportunities to provide long-term health predictions with high accuracy. Still, as discussed in the last section, if we particularly focus on the health impact, there is a lack of a clear framework to estimate what level of heat exposure may lead to specific health issues.

Repesentive literatures:

*Future population exposure to US heat extremes* [ (link)](https://www.nature.com/articles/nclimate2631)

*Global risk of deadly heat* [ (link)](https://www.nature.com/articles/nclimate3322)

*Projecting heat-related excess mortality under climate change scenarios in China* [ (link)](https://www.nature.com/articles/s41467-021-21305-1)

## Poetential Research Ideas:

1. Mapping the Heat Vulnerability Index dynamically in the long-term prediction

2. Large-scale heat exposure estimation for individuals with the activity-based method

3. Prediction of the individual level heat-related healthy risks/mortality in the future

## Meeting on 16th May

Notes:

NDVI datase

### Topic 1: Specific research questions

-

### Topic 2: Details need to be determined

- Timescale: current or future

- Temporal resolution: days?

- Geographical scale: single city/city-region or whole UK

- Spatial resolution: raster cells or OA/MSOA

- Expecting formatting of output

### Topic 3: Data availability

- SPC data: ready to use for demographical and socio-economic part, activity part is actively developing

- Clim-recal data: Three-cities samples are almost ready, plan for the whole UK?

- Other Dataset: OSM data, built environment data and more...

### Planning Ahead:

- Funding

- Timeline

- Team members

### To-do-list

- Refine the Research Plan (by?)

- Grant Application (aim and due?)

- Next time meeting on ?