2015 Progress Report: How will cleaner cooking and lighting practices impact regional air quality and climate in the Sahel of Africa?EPA Grant Number: R835424
Title: How will cleaner cooking and lighting practices impact regional air quality and climate in the Sahel of Africa?
Investigators: Hannigan, Michael P. , Dickinson, Katie , Dukic, Vanja , Hayden, Mary , Monaghan, Andrew , Oduro, Abraham , Wiedinmyer, Christine
Institution: University of Colorado at Boulder , National Center for Atmospheric Research , Navrongo Health Research Center
EPA Project Officer: Keating, Terry
Project Period: June 1, 2014 through May 31, 2017 (Extended to May 31, 2018)
Project Period Covered by this Report: June 1, 2015 through May 31,2016
Project Amount: $1,500,000
RFA: Measurements and Modeling for Quantifying Air Quality and Climatic Impacts of Residential Biomass or Coal Combustion for Cooking, Heating, and Lighting (2012) RFA Text | Recipients Lists
Research Category: Global Climate Change , Air Quality and Air Toxics , Tribal Environmental Health Research , Climate Change , Air
The overarching goal for this proposed work is to develop a better understanding of the social, physical, and climatological determinants of regional emissions and air quality linked to cooking and lighting practices in the African Sahel. To reach that goal, we have four specific objectives: (1) test hypotheses about the impact of different cooking technologies on behavior and emissions at the local scale; (2) develop a comprehensive set of emissions measurements from traditional cooking and lighting practices, as well as from cleaner burning alternatives; (3) develop realistic scenarios of regional-scale technology adoption and emissions by scaling up the observed social data and derived emissions relationships; and (4) assess how clean cooking and lighting practices could impact regional air quality and climate.
Our planned measurements in the Kassena Nankana (KN) district in northern Ghana were completed by the middle of the second year of the project. Since this research project built off a National Science Foundation (NSF) Dynamics of Coupled Natural and Human Systems (CNH) project, our experienced team was really able to start measurements during the first month in our first year; this allowed us to collect 18 months of data in the first 18 months of the project. Details of those measurements are given below. As such, half way through the second year, we were ahead of our proposed timeline. Then we hit a personnel snag. Our student lead on emissions measurements, Didier Muvandimwe, decided that he needed to take a 4-month break to explore a once-in-a-lifetime opportunity. He subsequently decided not to return to the Ph.D. program. We were slow to find a replacement for Didier as we were hoping that he would return to the Ph.D. program. Bringing on a new student also means training that student, which can be very time consuming. In this case, a new student would not only have to learn how to undertake the emissions measurements but also need to be able to undertake organic analysis as he/she would be involved in subsequent source apportionment. As of summer 2016, we were able to recruit a new student, David Pfotenhauer, and have begun training him. We hope that David, with existing team members Ricardo Piedrahita and Evan Coffey, will be able to complete analysis of collected data during Year 3. Our current plan for Year 3, beyond data analysis, includes a few more field emission measurements at alternative areas in the Sahel of Africa; likely Rwanda and an additional site in Ghana.
For the NSF CNH project, our team implemented a stove intervention study in 200 rural homes in the KN District in northern Ghana, see the included map. We used two stoves for each intervention home: a clay lined rocket stove built in Ghana and a forced draft gasifier stove powered by a battery that was solar panel charged. For that study, we were measuring CO and PM2.5 exposure to the cook and other residents in a subset of the homes. We were also surveying the participants to understand behavior. For the EPA project, we added 50 homes in the urban area of Navrongo, but without stove intervention. We added ambient monitors in the urban area. To further our assessment of emissions, we increased our field measurement scope to more samples numbers for each type of stove, added coal stoves to the measurement plan, upgraded our stove use monitoring, and added emissions measurements of other prominent sources including lighting. New surveys were developed to understand behavior in the urban areas as well. At the close of Yr 1, we had completed the bulk of our field measurements. Additionally, to meet our proposed objectives we have added components that allow us to undertake source apportionment of both CO exposures as well as all PM samples. Why? Our prelim exposure data for the intervention study showed no differences in exposures between the groups. We thought that one explanation might be that study participants were being exposed significantly to other sources. So, we added this source apportionment activity and are finding this aspect to be very important.
Field Measurement Overview:
Several data collection methods are being used to measure cooking behavior, emissions, exposure, and air quality in urban and rural environments. These field measurements will be used to improve emission inventories which will then be used to model regional air quality under various scenarios as well as to drive source apportionment analysis of exposure CO and PM as well as ambient PM.
Cooking behavior assessments:
Understanding how the introduction of new cooking technologies affects cooking behaviors is essential in order to understand whether and how these technologies subsequently affect other key outcomes like emissions, exposure, and health. To assess stove use, we used both stove use monitors (SUMs) and quarterly household surveys. The surveys reached every household while SUMs were deployed in a subset, 10-12 household per intervention arm and urban area, due to resource constraints. We used both of these instruments to determine the stove use over time, their patterns of use by intervention study arm and differences based on urban/rural settings. We have also compared the SUM results with the cross-sectional data provided by quarterly surveying to evaluate each instrument from a methodological standpoint. Analysis of these trends, comparison of data collection methods, and recommendations on stove adoption in the region for rural were presented in Assessment of cookstove stacking in Northern Ghana using surveys and stove use monitors (Piedrahita et al., 2016).
In addition to the stove use results described in the Key Findings above, we were able to use the quarterly surveys to explore how stove use was linked with meal cooked, see the figure below. In this region, by far the most commonly eaten meal is tuo zaafi (or TZ), which is a millet porridge. Many of the rural households grow millet. The second most commonly cooked starch is rice. Both TZ and rice are often paired with a vegetable soup. These three dishes represent over 90% of the meals cooked in the rural households. Of those three dishes, soup cooking was more commonly done over replacement stoves than rice or TZ cooking. Rice cooking saw some replacement of 3-stone stove use with the Gyapa stove, at least more so than for TZ. The cooking of TZ requires vigorous stirring and as such cooks are less likely to move away from the stability of their 3-stone stove for this dish. If the goal of a cookstove program is replacement of traditional stoves, then the ability of the improved stoves to cook traditional meals is essential. Paired with that conclusion, a cookstove replacement program may end up shifting a populations diet to meals that are more easily cooked on the new stoves.
Cookstove. Continuous cooking emissions were measured in different households from all clusters selected in the study, urban and rural. Our team wrapped up the continuous emissions collection field work in the first half of Yr 2. As we began processing the data from that field work, our student lead decided to take a break from the PhD program. We describe this slow down in subsequent sections of this report, but here we note that it took some time to get a new team member up to speed on the data. This team member subsequently reprocessed all the data and is now working on a manuscript which should be part of the Yr 3 report. In our Yr 1 report, we highlighted the preliminary heat transfer efficiency results for the field stove testing. When we reprocessed the emissions data, we still observed the same trends; the stoves tested during real world cooking in the field showed a heat transfer efficiency of about ½ of what has been observed in the lab for a water boiling test. When reprocessing the data, we also observed similar emissions trends, which are shown in the figure below. The figure highlights the trade-offs between CO and PM emissions and all emissions results are similar to those observed in other studies. Fuel choice impact on emissions is the most distinguishing trait of this figure. Charcoal cooking emits more CO and less PM, especially elemental carbon (EC) PM. The intervention stoves result in reduced CO emissions, for both charcoal and wood use. For wood use, the intervention stoves seem to reduce organic carbon (OC) PM emissions but not necessarily EC PM emissions. The EC-OC ratios are different when comparing intervention stoves to traditional stoves; this will be used in subsequent modeling effort to assess climate impacts. We are also going to explore CO and PM emissions per unit energy delivered and per mass of food cooked to further understand cooking activity not just emission factors.
Additionally, in Yr 2, our team initiated organic analysis of the PM collected on the filters. We are exploring the benefit to the field of publishing those detailed organic speciation results.
Other significant sources. To better understand the relative impact of cookstove emissions, our team is also assessing real-world emissions from other significant sources in sub-Saharan Africa. In year 2, we added a few more samples from kerosene lighting and trash burning from northern Ghana. More emission sampling is planned for Yr 3 with some interest in adding LPG stove emissions and including samples collected in a different region in Ghana. These emissions are measured using a small portable emission-pod (dubbed E-Pod), which uses low-cost sensors to measure real-time emissions of carbon monoxide (CO), carbon dioxide (CO2), nitrogen oxide (NO), nitrogen dioxide (NO2), volatile organic compounds (TVOCs), temperature, and humidity. This past year, we have continued to improve our understanding of the performance of this system.
Personal exposure and microenvironment monitoring:
Household sampling visits were conducted in 48-hour sampling periods. Typically, six households were sampled every two weeks (three urban and three rural). Stove use, personal CO and PM2.5 exposures, and microenvironment CO, CO2 and PM2.5 concentrations were measured in households during each deployment. To assess personal exposure to pollution from biomass combustion and other sources, study participants wore real-time CO monitors (EL-USB-CO300, Lascar Electronics) with a one-minute sampling frequency. Personal PM2.5 exposure was collected on quartz filters for EC/OC and organics analysis.
Early on in Yr 1, we observed preliminary CO and PM exposure results where there was little to no difference in exposure based on intervention arm. These initial observations made us want to really understand intervention stove use as well as exposure to other sources. We already had plans in place to dig into stove use so we could think about exposure in relation to how effective a stove is at replacing the traditional stove, this is the effort described above. The question of exposure apportionment, how much of someones exposure is attributable to a specific source, was an interesting new research angle. For the composite exposure PM samples, we could use organic speciation to feed a source apportionment modeling effort which would allow us to determine what proportion of the PM2.5 exposure samples were coming from biomass combustion. This source attribution approach has been used previously for exposure samples, but never in this kind of setting. For the continuously monitored exposure CO concentrations, we would not be able to use speciation as we didnt have that ability, so we instead attempted to use proximity to stove and stove use to attribute a portion of CO exposure to cooking. This approach is novel and implementing and validating it was a significant portion of effort in Yr 2. The results from both of these exposure apportionment paths will result in manuscripts in Yr 3.
Regional air quality monitoring
In order to understand spatial variability of pollution and to help identify pollutant sources, inexpensive air quality sensor packages that we have developed and dubbed G-Pods were deployed throughout the study region, see the map on page 2. The G-Pods were configured to measure O3, CO, CO2, NO, and NO2. The G-Pods were mounted 3-4 meters above ground at 6 clinics that are part of the National Health Service, in Paga, Kandiga, Kologo, Chiana, Northern Navrongo, and South-West Navrongo at the NHRC compound. At the start of the EPA project Yr 1, an additional G-Pod was installed in the Navrongo town market, an urban environment which experiences high vehicle and human traffic especially every third day, a market day, year round. The NHRC also served as the home base of the study and housed reference quality instruments in addition to the low-cost monitors. In general the ambient monitors showed much lower levels of combustion related pollutants than we are observing in the homes, while the market pod exhibited the highest levels on days when the market was active. This data is being incorporated into a manuscript that specifically explores the differences between the rural and urban air quality in the KN District.
Continent scale air quality modeling
Co-PI Wiedinmyer, along with colleague Dr. Eloise Marais (Harvard University), developed the Diffuse and Inefficient Combustion Emissions in Africa (DICE-Africa) emissions inventory. This is a continental-scale emissions inventory from sources that have been poorly characterized (i.e., residential biomass burning from cooking/heating activities) or ignored (i.e., back-up generator use and open waste burning) from existing inventories used for chemical model applications. Our team applied available activity data and published emission factors to drive the emission estimates. The resulting emissions estimates were subsequently used with the GEOS-chem chemical transport model to evaluate the impact of these sources on surface-level ozone and PM2.5 concentrations across the continent. The results were presented in Air quality impact of Diffuse and Inefficient Combustion Emissions in Africa (Marias and Wiedinmyer, 2016).
When available, emission factors determined by the activities of this grant will be used to update the DICE-Africa inventory. Further, this inventory is being used within an adjoint model framework to determine the health and climate impacts from different emission source sectors, including the residential biomass burning.
Future activities are described along with progress in each section above.
Piedrahita, R., K.L. Dickinson, E. Kanyomse, E. Coffey, R. Alirigia, Y. Hagar, I. Rivera, A. Oduro, V. Dukic, C. Wiedinmyer, M. Hannigan. Assessment of cookstove stacking in Northern Ghana using surveys and stove use monitors. Energy for Sustainable Development, 2016, 34:67-76. doi:10.1016/j.esd.2016.07.007
Marais, E and C. Wiedinmyer. Air quality impact of diffuse and inefficient combustion emissions in Africa (DICE-Africa). Environmental Science & Technology, 50(19):10739-10745; DOI: 10.1021/acs.est.6b02602, 2016.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
|Other project views:||All 7 publications||3 publications in selected types||All 3 journal articles|
||Marais EA, Wiedinmyer C. Air quality impact of Diffuse and Inefficient Combustion Emissions in Africa (DICE-Africa). Environmental Science & Technology 2016;50(19):10739-10745.||
||Piedrahita R, Dickinson KL, Kanyomse E, Coffey E, Alirigia R, Hagar Y, Rivera I, Oduro A, Dukic V, Wiedinmyer C, Hannigan M. Assessment of cookstove stacking in Northern Ghana using surveys and stove use monitors. Energy for Sustainable Development 2016;34:67-76.||