Climate-driven changes in CO2 emissions associated with residential heating and cooling demand by end-century in China

Global climate change affects residential heating and cooling demand that further contributes to carbon dioxide (CO2) emissions. The spatio-temporal changes in magnitude and distribution of the demands in China are poorly understood. In addition, few studies have focused on the future impact of climate change on long term residential CO2 emissions in China. Here we investigate regional changes in CO2 emissions calculated from degree-days. Our results show that heating degree-days (HDD), cooling degree-days (CDD) and their associated CO2 emissions all have large spatio-temporal variability. We find that average durations of HDD and CDD are predicted to be 34 days shorter and 63 days longer by the end of century (2071–2100) than history (1976–2005). CO2 emissions from residential cooling and heating are predicted to increase 218% and decrease 30% in China by end-century, respectively. We further examine the CO2 emissions from residential heating and cooling in five cities representative of five contrasting architectural climate zones in China. The CO2 emissions from heating of these cities are projected to decrease by end-century: 26% in Harbin, 32% in Beijing, 43% in Shanghai, 42% in Kunming, and 61% in Shenzhen. The CO2 emissions from cooling of these cities all increase by end-century: 436% in Harbin, 215% in Beijing, 223% in Shanghai, 765% in Kunming, and 149% in Shenzhen.


Introduction
China has experienced a large national increase in energy use for the last seventeen years (British Petroleum 2018). In 2017, China accounted for 23.2% of global energy consumption, and contributed 33.6% of global energy consumption increase (British Petroleum 2018). About 25% of the total national energy used for buildings is dominated by residential use (China Ministry of Construction 2010b). These high levels of energy consumption exacerbate carbon dioxide (CO 2 ) emissions in China, which increase 1.6% in 2017 (British Petroleum 2018). Therefore, China plays a fundamental role in global transition to low-carbon use.
Global climate change affects energy consumption and corresponding CO 2 emissions (Auffhammer and Mansur 2014), thus domestic energy use needs to adapt to the changing climate (Holmes et al 2017). In response to future climate change among different regions of China, it is urgent to understand the spatiotemporal distribution of residential building energy consumption and corresponding CO 2 emissions.
Previous research estimated energy consumption and associated CO 2 emissions by means of degreedays (Moustris et al 2015, Hao et al 2016, Park et al 2018. Heating degree-days (HDDs) and cooling degree-days (CDDs) are accumulated Celsius degree temperature deviations from a defined base temperature within a certain period (CIBSE 2006). The base temperature is the outdoor temperature, above or below which there is no need for domestic heating or cooling (Buyukalaca et al 2001). CDD well reflect the weather dependence of electricity consumption at building scale (Guan et al 2017). As indicators of thermal comfort, CDD were shown to increase, whereas HDD were shown to decrease at regional or national scale due to climate change (Mourshed 2011, Petri andCaldeira 2015).
Most current studies of degree-days in China only focused on the future changes (You et al 2014, Shen and Liu 2016, Shen et al 2017, rather than energy demand or CO 2 emissions. Some studies focused on national future changes in energy consumption, but lacked of comprehensive regional analysis (Shi et al 2016, Gi et al 2018. Furthermore, estimation of energy consumption and CO 2 emissions from the architectural climate zones across China employing various designs of building envelopes to conserve energy is deficient (China Ministry of Construction 2016). Without consideration of architectural climate zones, the influence of different building envelopes on energy demand would be ignored, especially the demand for heating. Ignoring such demand will result in inaccurate estimation of energy consumption.
In this paper, we aim to quantify potential impact of climate change on residential heating and cooling degree-days from five contrasting architectural climate zones across China. We select one city from each contrasting architectural climate zone (figure S1 is available online at stacks.iop.org/ERL/14/084043/ mmedia), comprising Harbin (severe cold), Beijing (cold), Shanghai (hot summer and cold winter), Shenzhen (hot summer and warm winter), and Kunming (moderate climate) for further analysis and comparison (China Ministry of Construction 2016). We also use historical  and future

Methods
A summary of data processing and analyzing steps is presented as figure S3. First, we use a joint bias correction method (Li et al 2014) to correct CMIP5 data. This method considers the relationship between variables, whereas commonly used bias correction methodologies treat each variable independently (Li et al 2014). After bias correction, the CMIP5 simulations of historical HDD and CDD are consistent with the observations (figure S4, R 2 >0.95). Then we calculate degree-days and CO 2 emissions by the methods described below.
2021-2050, 2071-2100) that are further interpolated to a 25 km×25 km grid using universal Kriging (Mosammam 2013). We include latitude and elevation as auxiliary variables in the kriging method.

CO 2 emissions
We develop a standardized residential building for each architectural climate zone to allow comparative analysis of CO 2 emissions that is a 10 m×10 m×3 m single-story building, with a 45% window-to-wall ratio of two of the four walls. The building envelope is based on the local design standard for energy efficiency of residential buildings. We estimate energy demand for heating as (CIBSE 2006): where Q h is the energy demand for heating (kWh), U′ is the overall building heat loss coefficient (kW K −1 ), HDD a is the annual HDD, and η denotes the overall heating system efficiency. Coal-fired boiler plants are used in cities in severe cold and cold zones for space heating. Air conditioners and electric heaters are used for heating in cities in the hot summer and cold winter, hot summer and warm winter, and moderate climate zones. The overall efficiency η of coal-fired boiler plants is 64.4% (i.e. 92% transport efficiency×70% boiler operation efficiency, China Ministry of Construction 2010b). And the overall mixed energy efficiency ratio equal to the overall heating system efficiency (η), which is 1.7 for air conditioners and electric heaters (China Ministry of Construction 2012). We calculate the overall building heat loss coefficient U′ (CIBSE 2006) as: , N is the air infiltration rate per hour (h −1 ), and V is the volume of infiltrated air (i.e. 180 m 3 ) equivalent to 60% of the volume of the standardized building (China Ministry of Construction 2010b). Here, we only consider the heat loss from wall, window, and roof of a residential building. Table 3  Similarly, we use a generalized method to calculate annual energy consumption for cooling demand (CIBSE 2006): where Q c is the energy demand for cooling (kWh), m  denotes the mass flow rate of air cooled per second (kg s −1 ), equivalent to 0.23 kg s −1 for an air conditioner with refrigerator power equal to 3486 W, C p is the specific heat of air equal to 1.005 kJ kg −1 K −1 , CDD a is the annual sum of CDD, and COP, which represents the energy efficiency ratio of the air conditioner for cooling, is equal to 3.0 (China Ministry of Construction 2012). We assume that two air conditioners (2Q c ) are in use at the same time in each household.
Finally, we calculate CO 2 emissions from raw coal as: where M is the mass of CO 2 emissions per square meter (kg m −2 ), Q is the energy demand for heating (Q h ) or cooling (Q c ), m is the net raw coal consumption rate of a thermal power plant or boiler heating system, r is the CO 2 emissions coefficient based on raw coal (1.9003 kg kg −1 ) (Zhang et al 2018), and S is the area of the standardized house (100 m 2 ). The net standard coal consumption rate of a thermal power plant in China in 2016 is 0.312 kg kWh −1 (http://cec. org.cn, accessed in October 2018), equivalent to a net raw coal consumption rate of 0.437 kg kWh −1 (based on the heat value of 1 kg of raw coal is equal to 0.7143 kg of standard coal, Evans and Lin 1997). The heat value of raw coal equals to 20908 kJ kg −1 (Evans and Lin 1997). Therefore, the raw coal consumption rate of boiler heating system is calculated as 3600 s h −1 20908 kJ −1 kg −1 =0.172 kg kWh −1 .

China annual degree-days
There are large spatial variations in HDD andCDD in history (1976-2005), mid-century (2021-2050), and end-century (2071-2100). HDD and CDD generally decrease and increase from the northwest to the southeast, respectively (figures 1 and S6). HDD are much larger than CDD, and also vary more than CDD in most parts of China, which indicates that heating demand is far more than cooling demand.
During the three periods, regions with the largest HDD and smallest CDD are all located in North Daxinganling, Altai and Tianshan Mountains, and Qinghai-Tibetan Plateau. Such result reflects the demand for heating, but not for cooling in residential buildings of these regions. We find that the regions with the largest CDD, such as Hainan Island and Turpan Basin, all have small HDD (figure 1). The HDD of the 828 stations are positively correlated with elevation (r=0.5) and latitude (r=0.8). This means that HDD are greater at lower elevations and at higher latitudes. For example, HDD are smaller in the low-lying Sichuan and Turpan Basins than in the surrounding regions, while CDD are larger than in the surrounding regions. Elevation and latitude influence degree-days through air temperature. As air temperature decreases with increasing elevation and latitude, there is more demand for heating and less demand for cooling. HDD decrease from history to mid-and end-century, and the decreasing magnitude is greater from mid-to late-century than from history to mid-century. This temporal increasing rate of HDD reduction indicates that heating demand is expected to become less and less in the future ( figure 2). Notably, regions with larger HDD in history, such as Qinghai-Tibetan Plateau, experience larger decreases in HDD, compared to the regions with smaller historical HDD. Therefore, the spatial differences in HDD across China are expected to decline in future.
We find that there is a slight increase in overall CDD with a warming climate, except at a few high elevation and high latitude regions (figure 2). Taking Qinghai-Tibetan Plateau with 0 historical CDD for example, there is no change in CDD over the century. Those regions with the largest historical CDD also experience the largest increase. The historical CDD of Hainan Island and Turpan Basin are around 700°C day, and the CDD increase is approximate 300°C day from history to mid-century and 600°C day from mid-to end-century. Our results show that accelerating global warming will lead to increases in magnitude of change in CDD over time.

Regional daily degree-days
To further examine the changes in degree-days over time, we analyze the durations of HDD and CDD of five cities representing five architectural climate zones (figure 3). We define HDD duration and CDD duration as the numbers of consecutive days when  HDD>13°C day and CDD>2°C day, respectively.
In the future, the start day of HDD duration is projected to be later and its end day is projected to be earlier. Compared with the historical period, the start days of HDD duration during mid-century are expected to be delayed by 8 and 11 days in Harbin and Beijing, while the end days advance by up to 3 and 8 days in these two cities ( figure 3). Consequently, the HDD durations are expected to be 11 and 19 days shorter in Harbin and Beijing, respectively. By the end of century, the start days of HDD duration are projected to be delayed by 10 and 16 days in Harbin and Beijing, compared to the mid-century, while end days are projected to advance by up to 9 and 14 days in these two cities. Consequently, the HDD durations are expected to be 19 and 30 days shorter in Harbin and Beijing, respectively. Duration of HDD is 27 days in Shanghai in history, and is 0 during the future periods. HDD durations are 0 in Shenzhen and Kunming for all three time periods. Average duration of HDD across China is predicted to be 34 days shorter by end-century than history.
As expected, future CDD duration is predicted to start earlier and end later due to global warming. Compared with history, the start days of CDD duration during mid-century are projected to 29, 5, and 24 days earlier in Shenzhen, Shanghai, and Beijing, respectively, while end days are projected to be delayed by 18, 21, and 22 days (figure 3). Consequently, the CDD durations are expected to be 47, 26, and 46 days longer in Shenzhen, Shanghai, and Beijing, respectively. From mid-to late-century, the start days of CDD duration are projected to be 29, 27, and 19 days earlier in Shenzhen, Shanghai, and Beijing, respectively, while end days are projected to be delayed by 22, 22, and 20 days. Consequently, the CDD durations in Shenzhen, Shanghai, and Beijing are expected to be 51, 49, and 39 days longer, respectively. The CDD duration in Harbin is predicted to last for 64 days during end-century, and is 0 during history and mid-century. The CDD duration of Kunming is 0 over the three time periods. Average duration of CDD across China is predicted to be 63 days longer by end-century than history.

CO 2 emissions
Consistent with the trends in degree-days, our results show less CO 2 emitted from heating and more CO 2 from cooling over time (figures 4, S8 and S9). In the future, total CO 2 emissions will decrease in most regions of China, except a few southern regions (figures 4(e) and (f)). These results indicate that future reductions in CO 2 emissions from heating are larger than increases in CO 2 emissions from cooling for most regions in China. The spatial distribution of heating CO 2 emissions is different from that of HDD. Along the boundary of architectural climate zones, CO 2 emissions from heating tend to be fewer in the north than in the south (figures 1, S6 and S8). Such result is  (1976-2005, 2021-2050, 2071-2100). The blue and pink dashed lines represent the threshold of HDD duration and CDD duration, respectively. Kunming is not included here, because its HDD and CDD durations are 0 for three periods. In general, the HDD duration is expected to be shorter, while CDD duration is longer over time.  due to the more stringent and more effective insulation building envelope design in the north.
Spatio-temporal changes in CO 2 emissions from heating in the cities representative of the five architectural climate zones show continuous decreases over time (figure 5). Heating CO 2 emissions in Harbin are the most (101.57, 92.10 and 75.47 kg m −2 during history, mid-and end-century, respectively), while those in Shenzhen are the fewest (16.10, 12.14 and 6.32 kg m −2 during the three periods, respectively). CO 2 emissions from heating in the other three cities are of similar magnitude and those in Beijing are a little more than those in Shanghai and Kunming. Conversely, CO 2 emissions from cooling all increase in the five cities. CO 2 emissions from cooling in Shenzhen are the most (14.39, 20.48, and 35.76 kg m −2 during history, mid-and end-century, respectively) with the greatest increase, while those in Kunming are the fewest with the least growth. However, in terms of the growth from history to end-century, the increasing rate is the largest in Kunming (765%) and is the smallest in Shenzhen (149%). Beijing, Shanghai, and Harbin increase 215%, 223% and 436%, respectively. In general, the magnitude of CO 2 emissions from heating in the five architectural climate zones is similar to their representative cities. However, CO 2 emissions from cooling in severe cold zones are the fewest, which is different from the ranking of five representative cities. CO 2 emissions from cooling in moderate climate and cold zones are similar.
Except for Shenzhen, total emissions of CO 2 from heating and cooling decrease over time. Historical total CO 2 emissions in Shenzhen are primarily from heating, whereas those by mid-and end-century primarily derive from cooling. By end-century, total CO 2 emissions in Shenzhen increase by 11.59-42.07 kg m −2 relative to historical levels. Among the other four cities, total CO 2 emissions in Kunming are the fewest, and the reductions are the greatest. Reductions in Harbin are of similar magnitude to those in Kunming, where total CO 2 emissions are the most due to the dominant and high level of heating demand. From history to end-century, the patterns of total CO 2 emissions in Beijing and Shanghai are similar. Specifically, total CO 2 emissions are predicted to decrease by 12.83-57.02 kg m −2 in Shanghai, and to decrease by 10.47-61.30 kg m −2 in Beijing. The uncertainties of CO 2 emissions caused by climate variability tend to be larger in end-century than mid-century (figure 5).

Discussion
The estimated degree-days and associated CO 2 emissions in this work show consistent trend with previous research (Zhou et al 2013, You et al 2014, Hao et al 2016, Shi et al 2016. Our results suggest more accurate spatial variation, because we consider the characteristics of architectural climate zones. Our analyses show that, as temperature increases, CO 2 emissions from CDD increase 218% by the endcentury relative to historical levels. However, we may underestimate such emission increase, because the regions with the largest increase in CDD have the highest the economic development and population density (90% of the Chinese population) in China. There is likely more energy demand on maintaining thermal comfort. The rapid growth in cooling demand, especially in southern China, indicates greater requirements for fuel used for electricity generation. It is important to improve the efficiencies of power generation and energy use of air conditioners, and reform of the power sector to reduce carbon is the most effective measure to decrease future CO 2 emissions (British Petroleum 2018). We also conduct the same analysis under the RCP4.5 emission scenario, the overall outcomes of which make not much difference (figures S5, S7 and S10).
One caveat of this research is the uncertainties to HDD and CDD from the selection of base temperatures (Holmes et al 2017), as they may be spatially different. Estimations of residential energy consumption and CO 2 emissions are based on HDD and CDD, and energy consumption in buildings is affected by various factors, such as the energy from lighting or the Sun (CIBSE 2006). In addition to climate change, heating and cooling demand is also affected by economic development, income growth, population density, and continued improvements in technology (Holmes et al 2017). Further research is needed to analyze the effects of other climate change and socioeconomic factors on energy consumption and CO 2 emissions.
Accounting for these limitations, our research confirms the impact of climate change on HDD and CDD and associated CO 2 emissions in China, which can be used to regulate building design to enhance the ability of conserving energy of residential buildings. It is important for policy makers to allocate energy appropriately. Our research also provides foundation for policy makers to adjust the heating period dates, according to the predicted delay in the start day of heating duration (e.g. 27 days later in Beijing by endcentury than history) and the advance of end day (e.g. 22 days earlier in Beijing by end-century than history). These future spatio-temporal trends in CO 2 emissions can guide mitigation. There may be more and longer summer cooling demand in China except Qinghai-Tibetan Plateau, which will result in more greenhouse gas emissions. Southern China is more likely to experience such increase in CO 2 emissions. climate modeling groups for producing and making their model datasets available (https://esgf-index1.