DATA-150

Arthur Kim

Professor Brewer

14 Dec 2021

Climate Change Vulnerability: Using Gender-Disaggregated CDR’s to Mitigate Gender Disparity

Word Count: 1861

Summary

We propose an investigation utilizing gender-disaggregated call detail records (CDR’s) in order to test their effectiveness at mitigating gender disparities in climate change vulnerability.

Motivation and Reasoning for this Study

Climate change is a dire issue that affects the entire world and brings about many ramifications. In most places around the world, men and women do not suffer the same way from climate change due to a number of factors such as differences in lifestyles and stereotypic gender perceptions (6, 8). As a result, men and women have different specific needs when it comes to addressing their vulnerabilities (8). One of the biggest gaps in the literature is the lack of adequate gender-disaggregated data in climate studies (6). This makes it very easy to have to rely on assumptions and generalizations as opposed to facts, which results in inadequate climate policies and action (6). With our investigation, we will produce an objective, real-time measurement of the distributions of each gender in vulnerable areas, facilitating better judgement and analysis. We anticipate that wherever we act, people of all genders will benefit from this because it will facilitate the enactment of more effective gender-focused climate policies, aid allocation, and proactive adaptive measures, overall reducing vulnerability.

Proposed Solution and Plan

As stated previously, in order to address the lack of sufficient gender-disaggregated data for studying climate change vulnerability, we will use gender-disaggregated CDR’s in order to generate high spatial and temporal resolution maps of the distributions of each gender in climate-vulnerable areas.

CDR’s provide spatial and temporal data about people by both localizing them to a particular base transceiver station (BTS) when the make a phone call, and recording the time and duration of the call (3). CDR’s also contain information about the gender of the caller, allowing for them to be gender-disaggregated and used to study the dynamics of each gender separately (4). When the CDR’s of a large number of people are lumped together, we get an overall picture of the population dynamics and distribution.

Using Earth System Models (ESM’s) combined with machine learning and AI, we have the ability to accurately forecast climate disasters well in advance (5). In fact, one source of ESM data, the Gravity Recovery and Climate Experiment (GRACE) allowed for scientists to forecast the flooding of the Mississippi River six weeks in advance back in 2011 (10). All ESM’s are aggregated into a single database called Coupled Model Intercomparison Project Phase 5 (CMIP5), which is what is primarily used in order to make disaster forecasts (5).

Using ML and AI with CMIP5, we will look for an imminent climate disaster such as a flood or tropical storm, and then determine the area in the world most vulnerable to it. Given the timeliness of these forecasts, we anticipate that we’ll have sufficient time to carry out our operations. We will then fly to the area in the world where the disaster will strike. We will then contact the local cellular providers in order to obtain as many CDR’s as possible. We will then disaggregate them based on the gender of each caller in order to work with them separately.

In order to make use of the CDR’s we obtain, we will follow a procedure taken by Csáji et al. First we will use a Voronoi Tesselation the area of interest based on the BTS locations in order to assign each point in space to its nearest BTS (Figure 1). We will the use Delaunay Neighborship in order to designate “frequent locations.” To the CDR data we obtain, we will apply k-means clustering in order to work out a spatial and temporal pattern of caller activity and dynamics. We will then compare our population distribution maps of both men and women in order to work out the gender distribution in each area of our chosen location. In their original study, when Csáji et al. compared their CDR-based assessments to independent statistics from the Portuguese National Institute of Statistics (INE), they obtained a correlation of 0.92 (Figure 2) (2). Because of this high correlation, we are confident that our study will produce accurate and meaningful results. We will then examine the overall gender distributions across the vulnerable areas in order to look for any unequal distributions. We will then present our gender-disaggregated population distribution map to the local government and aid organizations so that they can take appropriate adaptive measures. Additionally, when the climate disaster strikes, the aid organizations will not only know how much aid to allocate, but also the proportions of each gender-specific aid.

We hypothesize that by being more aware of gender disparities in climate change vulnerability and helping them more effectively, we can spread more awareness and help make progress towards greater gender inequality in the areas we explore. image (2)

Figure 1: Based on the call patterns of individuals over time and the BTS’s they are localized to, we can get a rough sense of their trajectory (2). Figure A also showcases a Voronoi Tessellation, used in order to localize all points in space to the nearest cellular antenna (2).

image (1)

Figure 2: The pattern of call dynamics worked out by Csáji et al. using k-means clustering with k=3 (1). Figure b shows that they fall nicely with INE data, supporting their validity as a measure of population dynamics (1).

Objective of our Inquiry

Overall, the objective of our research is to validate the use of gender-disaggregated CDR’s in the study of climate change vulnerability. We are hoping that by providing a greater amount of objective gender-disaggregated data, this will pave the way for future studies of gender inequality and facilitate efforts to mitigate it. Additionally our real-time gender-disaggregated population distribution will be used by government organizations in order to make more effective gendered policies, and allocate specific gendered-aid, depending on the gender distributions in each area at the time of the disaster.

However, we anticipate a number of obstacles in our research, but we have planned solutions to overcome them. For one, because the disaster forecasts often only leave a window of a few weeks before the disaster actually strikes, traveling to the location of interest and collecting and analyzing the CDR data in time will prove tricky. Luckily however, after doing some research, we anticipate that wherever we choose to travel, we should be able to get same-day flight tickets and get there fairly quickly. Despite our relatively short window of time, we believe that because some ESM’s can make predictions several weeks in advance as opposed to days, they are our best means we have to forecast a disaster and choose a location to conduct our study. Additionally, we have to ensure that we can actually obtain the CDR’s from the local cellular providers. In order to do so, we will contact the cell companies and explain our operations, and then we should be able to obtain the CDR’s fairly easily.

One major logistical problem we anticipate is that depending on where we travel, it is very possible that a significant portion of the population does not have access to cell phones. Therefore, using CDR’s alone to create gender-disaggregated population maps could result in undercoverage. In order to mitigate bias in our results, based on poverty and technology-access estimates from independent statistics, we will also conduct household surveys as needed. In fact, Dujardin et al. in their Sustainability article, “Mobile Phone Data for Urban Climate Change Adaptation: Reviewing Applications, Opportunities and Key Challenges” discuss how when CDR’s are combined with other sources of knowledge, their effectiveness at improving adaptive capacities increases substantially (3). By supplementing our CDR data with household surveys where needed, we will be able to create accurate gender-disaggregated population distribution maps wherever we go.

Because our study will help mend the lack of gender-disaggregated data, the entire climate-vulnerable populations of our study will benefit because they will experience better climate policies and aid reception, which will improve their adaptive capacity. This will allow for better public health due to the decrease in risk of mortality and other conditions of poverty.

Why This Plan Should be Considered

In his book, Development as Freedom, Amartya Sen defines human development as the increase in socioeconomic freedoms and security, and wellbeing as a whole (9). He also defines poverty as the lack of said freedoms (9). Based on this, it is clear that climate change itself negatively affects human development. Because climate disasters like floods and tropical storms are oftentimes lethal, they diminish public health by increasing mortality (7). Additionally, in Sub-Saharan Africa particularly, climate disasters negatively affect food security (1). Azzarri et al. in their World Development article, “Climate and poverty in Africa South of the Sahara. World Development” show that Sub-Saharan African agriculture is negatively impacted by floods and temperature rises, which consequently reduces the amount of food available, leading to many other problems such as malnutrition and worse wellbeing (1). Evidently, climate change has disastrous consequences for human development. Because our plan involves mitigating these issues, it will facilitate human development as well as adaptive capacities.

Additionally, as discussed previously, Lau et al. in their Nature Climate Change article, “Gender equality in climate policy and practice hindered by assumptions,” discuss how because of the inadequate amount of gender-disaggregated data, researchers and policymakers rely on gender stereotypes and generalizations, resulting in poor solutions to addressing gender disparities in climate study (6). Because our plan will produce an objective measurement of the real-time gender distributions in vulnerable areas, it will reduce the influence of assumptions and stereotypes in vulnerability assessments. As stated before, this will allow for aid to be more effectively allocated as well as more proactive mitigation of gender disparities from government organizations. Because our research aims to mend an issue that has long-plagued climate science, its success would be a major contribution to the field, so we believe this is an endeavor worth taking.

Overall, we believe our primarily-CDR-based method despite its shortcomings is best because it allows for the creation of high spatial and temporal resolution population distribution maps using readily available data. Additionally, as mentioned previously, they provide accurate and meaningful results, with Csáji et al. having obtained a correlation of 0.92 (2). Additionally, because the distribution data has a high temporal resolution, it is able to more precisely determine the dynamics of people in vulnerable areas compared to traditional censes (3), which is more useful for aid organizations when they need to act as quickly as possible in emergency situations. Because of the enormous benefits for human development and future climate studies, we believe that our research is worth funding.

Objections and How We Will Address Them

We recognize that there are several potential shortcomings with our research plan. As stated previously, we are aware of the fact that in many parts of the world, CDR’s themselves cannot form a complete map of the population gender distribution due to risks of undercoverage. However, as discussed earlier, we believe we can mitigate this issue by incorporating household surveys in addition to CDR analysis, ensuring that the areas most likely to be overlooked by CDR’s are properly taken into account. Although household surveys lack the temporal dimension of CDR’s, the fact that we will be conducting them near the time of the climate disaster should provide a sufficient estimate of the gender distributions in areas undercovered by CDR’s.

Admittedly, because of the complexity and vast differences across cultures and areas of the world, even without bias, there’s no guarantee that analyzing CDR’s will prove as effective as it did in Csáji et al.’s original study if used in different parts of the world. In order to ensure the veracity of our results, we will compare them to independent statistics from the area if available. Should our measurements prove less than ideal, we can compensate with survey data in order to form as complete a picture as possible.

Anticipated Costs

2021-12-10 (2)

References:

  1. Azzarri, C., & Signorelli, S. (2019). Climate and poverty in Africa South of the Sahara. World Development, 125, 104691. https://doi.org/10.1016/j.worlddev.2019.104691
  2. Csáji, B. C., Browet, A., Traag, V. A., Delvenne, J.-C., Huens, E., Van Dooren, P., Smoreda, Z., & Blondel, V. D. (2013). Exploring the mobility of mobile phone users. Physica A: Statistical Mechanics and Its Applications, 392(6), 1459–1473. https://doi.org/10.1016/j.physa.2012.11.040
  3. Dujardin, S., Jacques, D., Steele, J., & Linard, C. (2020). Mobile Phone Data for Urban Climate Change Adaptation: Reviewing Applications, Opportunities and Key Challenges. Sustainability, 12(4), 1501. https://doi.org/10.3390/su12041501
  4. Goel, R., Sharma, R., & Aasa, A. (2021). Understanding gender segregation through Call Data Records: An estonian case study. PLOS ONE, 16(3). https://doi.org/10.1371/journal.pone.0248212
  5. Huntingford, C., Jeffers, E. S., Bonsall, M. B., Christensen, H. M., Lees, T., & Yang, H. (2019). Machine Learning and Artificial Intelligence to aid climate change research and Preparedness. Environmental Research Letters, 14(12), 124007. https://doi.org/10.1088/1748-9326/ab4e55
  6. Lau, J. D., Kleiber, D., Lawless, S., & Cohen, P. J. (2021). Gender equality in climate policy and practice hindered by assumptions. Nature Climate Change, 11(3), 186–192. https://doi.org/10.1038/s41558-021-00999-7
  7. Pugatch, T. (2019). Tropical Storms and Mortality Under Climate Change. World Development, 117, 172–182. https://doi.org/10.1016/j.worlddev.2019.01.009
  8. Rautio, A., Kukarenko, N., Nilsson, L. M., & Evengard, B. (2021). Climate change in the Arctic—the need for a broader gender perspective in data collection. International Journal of Environmental Research and Public Health, 18(2), 628. https://doi.org/10.3390/ijerph18020628
  9. Sen, A. (2019). Development as freedom. MTM.
  10. Tapley, B. D., Watkins, M. M., Flechtner, F., Reigber, C., Bettadpur, S., Rodell, M., Sasgen, I., Famiglietti, J. S., Landerer, F. W., Chambers, D. P., Reager, J. T., Gardner, A. S., Save, H., Ivins, E. R., Swenson, S. C., Boening, C., Dahle, C., Wiese, D. N., Dobslaw, H., … Velicogna, I. (2019). Contributions of GRACE to understanding climate change. Nature Climate Change, 9(5), 358–369. https://doi.org/10.1038/s41558-019-0456-2