Global near real‐time daily apparent temperature and heat wave dataset

Temperature is the most concerned factor in the human–environment interactions. Apparent temperature accounting for other meteorological variables, such as humidity, wind speed and solar radiation, is the equivalent temperature, and it is a more accurate indicator to reflect human's environmental temperature perception. High‐quality apparent temperature data are urgently needed for the further research on human–environment interactions. At the same time, as global heat waves continue to increase in frequency, duration and intensity, understanding the impact of heat waves on human health needs human perception‐based heat wave data. Using ERA5 hourly data on single levels of 2 m temperature, wind speed, dewpoint temperature and solar radiation, this study developed a global apparent temperature and heat wave (GATHW) toolbox based on the Climate Data Store (CDS) online platform. This toolbox allows using three methods to calculate daily apparent temperature and heat wave at three spatial resolutions of 0.25°, 0.5° and 1° respectively. It can realize online calculation, display and real‐time download and is updated in near real‐time. The global daily apparent temperature and annual heat wave dataset from 2006 to 2020 calculated by the toolbox can be obtained from https://www.doi.org/10.5281/zenodo.4764325. After evaluation, this dataset can well reflect the typical extreme temperatures and heat wave events, and is more accurate, with higher resolution and faster update frequency than similar data products, which can provide data support for the study of human–environmental interactions and extreme climate events.


| INTRODUCTION
Due to the important effects of temperature on human activity and health (Achebak et al., 2020;Rodo et al., 2021;Xu et al., 2020;Yang et al., 2021), temperature has always been the most concerned factor in the humanenvironment interactions (Watts et al., 2018;Xu et al., 2020). Numerous studies have analysed the relationship between temperature and disease and injury, and ambient temperature is usually used to calculate the comfort temperature index (Achebak et al., 2020;Gosling et al., 2009;Li et al., 2013;Yang et al., 2021). However, human's environmental temperature perception is comprehensively affected by ambient temperature, wind speed, humidity and solar radiation (Liu et al., 2020;Steadman, 1979;Yi et al., 2019b), apparent temperature calculated based on these factors can better reflect human's temperature perception of the environment than ambient temperature only (Kovats & Hajat, 2008;Niu et al., 2020;Steadman, 1984). Recent studies have analysed the relationship between apparent temperature and human disease and health (Aguglia et al., 2021;Liu et al., 2020). High apparent temperature could greatly increase the emergency ambulance dispatches (Alessandrini et al., 2011), as well as the number of acute excessive drinkers and the risk of acute excessive drinking (Liu et al., 2020). Extreme apparent temperature may also be a threat for psychiatric disorders and human health (Niu et al., 2020). The importance of apparent temperate has, therefore, been recognized in recent studies. Several comfort temperature indices based on multiple meteorological variables have been developed, which can be mainly divided into empirical indices and human thermal stress indices (Yan et al., 2021). Empirical indices, such as Heat Index (HI) and Humidity Index (Humidex), are widely used in meteorological service departments because they are easy to calculate and understand (Giannaros et al., 2018). Human thermal stress indices, such as Universal Thermal Climate Index (UTCI) and Standard Effective Temperature (SET), have better performance in applicable scenarios and accuracy. For example, except for meteorological variables, UTCI takes personal factors, such as physical activity and adaptive clothing, into account (Jendritzky et al., 2012). Based on these indices, some heat stress datasets have been generated (Mistry, 2020;Di Napoli et al., 2021;Yan et al., 2021). Further studies are in urgent need of large spatial scale, high spatiotemporal resolution and highquality apparent temperature data. In this study, we used multiple indices to generate a long-term high-resolution global apparent temperature dataset, and this dataset is then used to produce global heat wave dataset.
Global warming may continue to increase the frequency, duration and intensity of heat waves (Meehl & Tebaldi, 2004;Perkins-Kirkpatrick et al., 2012). In 2003, an intense heat wave occurred in Western Europe, with temperatures reaching the highest level in 1500 years (Luterbacher et al., 2004), resulting in the deaths of about 70,000 people (Coumou & Rahmstorf, 2012). In 2009, a heat wave in southeast Australia resulted in 374 deaths (Carnie, 2009). In 2010, a heat wave in Russia claimed an estimated 54,000 lives (Gutterman, 2010;McMichael & Lindgren, 2011). Several scholars have developed heat wave or warm-spell datasets. Raei et al. proposed a multimethod global heat wave and warm-spell record and toolbox (GHWR), which is based on gridded temperature data on the resolution of 0.5° × 0.5° from 1979 to 2017 (Raei et al., 2018). Malcolm N presented a high-resolution global Climate Extreme Indices (CEIs) dataset based on daily temperature and precipitation data from the Global Land Data Assimilation System (GLDAS). The dataset includes 71 annual (and in some cases monthly) CEIs with a grid resolution of 0.25° × 0.25°, covering 47 years from 1970 to 2016 (Mistry, 2019). By simulating daily maximum and minimum temperature and precipitation, Defrance conducted a prediction study and produced a global annual 0.5° × 0.5° global CEI dataset for 1951-2099 (Defrance, 2019). However, current heat wave or warm-spell datasets are calculated based on ambient temperature, which is different from people's temperature perception of the environment. Heat wave based on apparent temperature can effectively reduce this difference and reflect the severity of heat waves more accurately.
In order to compensate for the lack of apparent temperature and heat wave data, this study used ERA5 hourly The global daily apparent temperature and annual heat wave dataset from 2006 to 2020 calculated by the toolbox can be obtained from https://www.doi. org/10.5281/zenodo.4764325. After evaluation, this dataset can well reflect the typical extreme temperatures and heat wave events, and is more accurate, with higher resolution and faster update frequency than similar data products, which can provide data support for the study of human-environmental interactions and extreme climate events. data on single levels of 2 m temperature, wind speed, dewpoint temperature and solar radiation, and developed global apparent temperature and heat wave (GATHW) toolbox based on the Climate Data Store (CDS) online platform. This toolbox allows using multiple methods to calculate hourly apparent temperature and yearly heat wave event records from 1979 to present in different spatial resolutions, which can realize online calculation, display and real-time download, and is updated in near real-time. GATHW can provide not only data support for the further study of human-environment interactions but also greatly improve the user's freedom in choosing the time, region, spatial resolution and method of interest, and reduce the requirements of client computing and storage power. Users can also further process and analyse dataset through the editing of fully open-source code.

| DATA SOURCE AND METHODS
Data requirements, methods and workflow of the production of apparent temperature and heat wave dataset are shown in Figure 1. Firstly, apparent temperature data are produced based on ERA5 data using three methods, and apparent temperature is then used to generate heat wave events record.

| ERA5 hourly data on single levels
We chose ERA5 hourly data not only because of its high spatiotemporal resolution and strict quality control measures, but also because it can be easily accessed online. ERA5 hourly data use data assimilation to combine model data with global observation data to form a globally complete and consistent dataset. It provides hourly estimates for numerous atmospheric, ocean-wave and land-surface quantities. All datasets submitted to CDS are assessed by the Evaluation and Quality Control (EQC) function of C3S independently of the data supplier. EQC encompasses a framework of processes aimed to assure technical and scientific quality harmonized across all dataset types available through the CDS. During the EQC process, the documentation provided with the dataset is scrutinized and data are checked for usability and reliability. An uncertainty estimate is sampled by an underlying 10-member ensemble at three-hourly intervals (Hersbach, 2018). ERA5 hourly data can be easily obtained in two ways: one is in the CDS direct download (https://cds.clima te.coper nicus. eu/), the other is based on CDS Toolbox (https://cds.clima te.coper nicus.eu/cdsap p#!/toolbox). For the purpose of online calculation, this study obtained ERA5 hourly data of 2 m temperature, wind speed, dewpoint temperature and solar radiation based on the CDS Toolbox. The spatial resolution of the data is 0.25°, and the time range was 1979 to the present. We resampled the hourly data to daily and carried out further processing.

| Calculation of apparent temperature
Apparent temperature refers to the equivalent temperature perceived by human body to the environment. It considers multiple meteorological variables, such as 2 m temperature, wind speed, relative humidity and solar radiation, which can better reflect human's temperature perception of the environment than ambient temperature only (Liu et al., 2020;Yi et al., 2019a). In 1979, Masterton quantified the discomfort of human body due to excessive heat and humidity, and proposed the Humidex (Masterton et al., 1979). Humidex is increasingly widely used in indoor and outdoor thermal environment evaluation and human comfort evaluation due to its simple calculation and strong interpretation ability. According to the current international standards, Humidex has a good evaluation effect on thermal comfort under hightemperature conditions in summer (d'Ambrosio Alfano et al., 2011). Humidex considers ambient temperature (T) and dewpoint temperature (D). Steadman proposed a general formula for calculating apparent temperature in 1984, which is the most widely used formula for calculating apparent temperature at present (Niu et al., 2020;Roye et al., 2019). The Steadman formula takes ambient temperature (T), 10-m wind speed (W), relative humidity (Rh, in some cases using vapor pressure (P) instead) and solar radiation (R) into account. As solar radiation data are relatively difficult to obtain, this variable is sometimes ignored (Steadman, 1984). GATHW provided three methods to calculate apparent temperature, and the calculation formulas of each method are as follows: where AT is apparent temperature (°C), T a is 2 m temperature (°C), T d is dewpoint temperature (°C), P is vapor pressure (Pa), W is wind speed (m/s), R h is relative humidity (%) and R is net radiation (w/m 2 ).

| Definition of heat wave
Generally, the weather process that exceeds the hightemperature threshold for several consecutive days is defined as a heat wave. Thus, the definition of heat wave involves two main parameters, one is the high-temperature threshold and the other is the duration threshold (Perkins, 2015). Currently, there is no unified standard for heat waves ( Workflow of the production of apparent temperature and heat wave dataset The World Meteorological Organization (WMO) suggests that a high-temperature process with temperature over 32°C for more than three consecutive days is a heat wave (Wang et al., 2018). In most areas of China, the standard of high temperature is 35°C (Chen & Li, 2017). Methods to determine high-temperature threshold mainly include constant threshold and percentile threshold (Huang et al., 2010;Stefanon et al., 2012;Vautard et al., 2013;Williams et al., 2012). Some scholars proposed a heat wave index (Perkins & Alexander, 2013;Raei et al., 2018). In view of the performance and practicability of the method, GATHW provides constant threshold, percentile threshold and combined threshold methods, and the output includes 7 heat wave characteristics (Table 1), which can be directly downloaded.
• Constant threshold: constant threshold determines a fixed high-temperature threshold over the entire area and time period. This approach defines a heat wave event in which temperatures exceed a constant threshold for several consecutive days. In the long-term adaptation process, the suitable temperature range of residents in different regions may change (Xu et al., 2020;Yin et al., 2019), so we suggest using constant threshold in a small spatial range. GATHW sets an optional constant threshold between 25 and 40°C. Constant threshold is more efficient due to the simplicity of calculation. • Percentile threshold: percentile threshold determines high-temperature thresholds based on local climatology, which is localized in both time and space. More specifically, the temperature threshold for each calendar day and each grid is specified as an upper-tail percentile threshold for the Probability Distribution Function (PDF), which is constructed from long-term daily temperature records on the calendar day of interest (Panda et al., 2017;Vautard et al., 2013). GATHW built the PDF for each calendar day from 1991 to 2020 and provided optional percentile threshold between 80 and 95. The percentile threshold can detect a warm-spell that is significantly higher than the regional climatology, but it is not suitable for detecting large area heat waves. This is because the percentile threshold is a relative value, and the detected warm-spell may appear in the time or region with low temperature. • Combined threshold: considering the characteristics of both constant and percentile threshold, we proposed a combined threshold. This approach defines a heat wave event in which temperatures are both above constant and percentile threshold for several consecutive days. The combined threshold not only considers the spatiotemporal variation of a high-temperature threshold but also avoids the detection of warm-spells with low temperature as heat waves, so it can be used to detect heat wave on a global scale (Yin et al., 2020). GATHW allows users to input custom combinations of constant thresholds and percentile thresholds in a fixed format. Certainly, each threshold should refer to the value range, that is, the constant threshold is between 25 and 40°C and the percentile threshold is between 80 and 95.
3 | DATA RESULTS AND EVALUATION 3.1 | Evaluation of apparent temperature data 3.1.1 | Comparison of different methods Figure 2 shows the mean ambient temperature (2 m air temperature) and apparent temperature calculated based on the three methods in the boreal summer of 2020 (June 1-August 31). As ambient temperature is the main factor controlling apparent temperature, the macro distribution of apparent temperature is consistent with ambient  (Figure 2a), but the maximum apparent temperature is not detected in these two regions. This is because the Sahara Desert and Arabian Peninsula are very dry, which reduces the apparent temperature. In Figure 2b, as Humidex TD considers temperature and relative humidity, the maximum apparent temperature appears in coastal areas with high ambient temperature, such as Arabian Peninsula, India, and Southeast China. Due to the influence of relative humidity, the apparent temperature of North Africa, West Asia, Central Asia, and Siberia with low water vapor content is lower than the ambient temperature, while the apparent temperature of tropical rainforest, Southeast Asia and coastal areas with high water vapor content is higher than the ambient temperature ( Figure 2b). Steadman TWP further considers the wind speed, and the region with higher wind speed has lower apparent temperature. The representative region is Antarctica, where the apparent temperature is more than 5°C lower than ambient temperature due to the high wind speed (Figure 2c). Steadman TWPR increases with the inclusion of the solar radiation factor. In the Tibetan Plateau, the strong solar radiation makes the apparent temperature 10°C higher than ambient temperature (Figure 2d). The comparison results of apparent temperature and ambient temperature show that the calculation results based on Humidex TD are generally higher than ambient temperature, while the results based on Steadman TWP are generally lower than ambient temperature, and Steadman TWPR has a relatively balanced performance. In addition, apparent temperature and ambient temperature differ greatly at high and low latitudes, but little at middle latitudes ( Figure 3). Figure 4 shows ambient temperature and apparent temperature time series of New York City in July 2019 and Beijing in January 2021. In July 2019, many cities in central and eastern America experienced heat waves. On July 21, the highest temperature in New York City reached 36.1°C and the Humidex TD apparent temperature reached 46°C (https://www.weath er.gov/). In January 2021, Beijing experienced the coldest month in the 21st century. On January 6, the temperature reached −17.1°C, and the wind reached level 8 (http://www.weath er.com. cn/). Under the dual effects of low temperature and strong wind, the apparent temperature measured by meteorological department reached −29°C, even lower than −24°C of the Arctic in the same period (https://weath er.com/). Figure 4 shows the two extreme temperatures well. In Figure 4a, the maximum ambient temperature on July 21 is 35.6°C, apparent temperature based on Humidex TD, Steadman TWP and Steadman TWPR reached to 48.9, 39.8 and 41°C respectively. In Figure 4b, the minimum ambient temperature on January 6 is −19.2°C, Humidex TD F I G U R E 2 Mean ambient temperature and mean apparent temperature based on three methods in the summer of 2020: (a) ambient temperature, (b) Humidex TD, (c) Steadman TWP and (d) Steadman TWPR. Mean temperature is calculated based on daily mean temperature resampled from hourly data apparent temperature is −24.6°C, Steadman TWP apparent temperature is −27.9°C, and Steadman TWPR apparent temperature reaches −28.1°C. Apparent temperature varies based on different calculation methods. We suggest that the most appropriate method should be selected based on regional climatology and data availability. For example, ERA5 may have different performance in different regions.

| Overview of heat wave in 2019
In 2019, heat waves have been reported in many places around the world, including Eastern America, Western Europe, Japan, India and Pakistan, and the temperature in many places reached the highest level in history

F I G U R E 4 Time series of apparent temperature of (a) New York City and (b)
Beijing based on Humidex TD, Steadman TWP and Steadman TWPR. Temperature of New York City is daily maximum temperature resampled from hourly data and minimum for Beijing (https://en.wikip edia.org/wiki/List_of_heat_waves). Based on daily mean apparent temperature and combined threshold (constant threshold = 29°C, percentile threshold = 85%), Figure 5 reflects the heat wave event in the summer of 2019. The number of heat wave events in Eastern America, Western Europe and Japan are less than 2. In central America, Southeast China, Brazil, India and Pakistan, the heat wave frequency is more than 4. Heat waves are also detected in western Siberia of Russia (Figure 5a). The total duration of heat wave is less than 20 days in Europe and Japan, 30 days for America and China and more than 50 days in the Arabian Peninsula and India (Figure 5b). Globally, the most persistent heat wave is generally less than 10 days ( Figure 5c). India, Pakistan and China have higher heat wave intensity (Figure 5d). The first heat wave in majority of the lower latitude regions (30°S-30°N) began before June, in Europe and America later than June, in northern China and Japan later than August (Figure 5e), while in some parts of the world, it lasted until the end of August (Figure 5f).

| European heatwave in summer, 2003
In the summer of 2003, Europe experienced the hottest summer since 1540 (Stott et al., 2005). Driven by anticyclonic (high pressure) conditions and unusual soil water deficit (Garcia-Herrera et al., 2010), severe heat waves hit Europe, causing more than 70,000 deaths (Robine et al., 2008), and France was the country with the most serious losses (Coumou & Robinson, 2013). Based on daily mean apparent temperature and combined threshold (constant threshold = 29°C, percentile threshold = 85%), Figure 6 shows Heat wave record is calculated based on daily mean apparent temperature using combined threshold (constant threshold = 29°C, percentile threshold = 85%), the duration threshold is 3 days. Daily apparent temperature is calculated using Humidex TD lasts more than 30 days, while in most other European regions, the heat wave frequency is less than 4 and lasts less than 20 days (Figure 6a and b). The longest duration of a single heat wave is generally less than 10 days ( Figure 6c). Interestingly, heat waves were detected in several capital cities and nearby areas, such as London, Berlin, Prague and Warsaw, which may be related to the close relationship between heat waves and urban heat island (Jiang et al., 2019;Li et al., 2015;Ward et al., 2016) ( Figure 6a-c). France, Italy and Portugal have the highest heat wave intensity, and the extreme apparent temperature is above 35°C (Figure 6d). In southern Europe, the first heat wave started earlier than June, and the last heat wave ended later than August. The heat wave in Central Europe began in early August and ended in middle August (Figure 6e and f). We find that these results are consistent with reported news. For example, France is the country with the highest heat wave intensity, while Italy has the longest heat wave duration (https://en.wikip edia.org/wiki/2003_Europ eanhe at_wave).

| Russian heatwave in summer, 2010
In the summer of 2010, western Russia suffered a severe heat wave, and the temperature in July broke the historical record, which may be the hottest summer in recent 500 years (Otto et al., 2012). The heat wave has had a serious socio-economic impact, resulting in 56,000 deaths, more than the European heat wave in 2003 (Barriopedro et al., 2011). This heat wave is mainly affected by anthropogenic climate change. In addition, soil moisture temperature feedback is considered as an important factor in the formation of abnormal high temperature (Hauser et al., 2016). Based on daily mean apparent temperature and combined threshold (constant threshold = 29°C, percentile threshold = 85%), Figure 7 reflects the Russian heat wave in summer, 2010. Southern Europe and western Russia experienced heat waves, lasting for over 5 days. The capital Moscow is the core heat wave area in Russia, where heat wave began in late June and ended in early August. As a middle-and high-latitude country, Russia has repeatedly experienced heat waves in recent years, which indicates that human factors are leading to severe climate change (Hauser et al., 2016).

| Australian heatwave in winter, 2019
As a southern hemisphere country, Australia's summer is defined as December, January and February. Wildfires in eastern Australia from 2019 to 2020 have attracted considerable international attention (Nolan et al., 2020). From November 2019, wildfires in Australia continued to worsen. In early February 2020, drought and strong winds intensified wildfires again, Heat wave record is calculated based on daily mean apparent temperature using combined threshold (constant threshold = 29°C, percentile threshold = 85%), the duration threshold is 3 days. Daily apparent temperature is calculated using Humidex TD and the fire was not controlled until early March 2020 (https://disas terph ilant hropy.org/disas ter/2019-austr alian -wildf ires/). Heat waves caused hundreds of wildfires in New South Wales, Victoria and Queensland (Arriagada et al., 2020), and millions of Australians experienced record breaking heat waves (https://www. washi ngton post.com/weath er/2020/11/30/austr aliaheat-wave-fire-risk/). Based on daily mean apparent temperature and combined threshold (constant threshold = 29°C, percentile threshold = 85%), Figure 8 maps the Australian heat wave in the summer of 2019. Except for the southeast coastal areas and central part of the country, heat waves were detected in almost the whole country, and the frequency of heat waves was generally higher than two times, reaching more than five times in some areas (Figure 8a). The duration of heat wave in most regions is less than 20 days, and the longest single heat wave is generally less than 10 days, indicating that heat waves in Australia are frequent and short (Figure 8b and c). The heat wave intensity gradually decreases from north to south, and the extreme apparent temperature in the northern coastal area reaches above 40°C (Figure 8d). In most parts of Australia, the first heat wave started in December and the last heat wave ended in early February of the next year (Figure 8e and f).

| IMPROVEMENTS AND LIMITATIONS
The global apparent temperature and heat wave online toolbox developed in this study have the following innovations: (1) The research on human and environmental interactions and extreme climate events urgently needs high-quality apparent temperature and heat wave datasets. This dataset and toolbox can provide data support for relevant research; (2) GATHW can meet the needs of users for different time, region, spatial resolution and method of interest. It can display and download data online, which greatly reduces the requirements for client computing and storage performance. Online computing is fast becoming a popular approach within the research community; (3) Compared with the similar dataset product, the spatial resolution and accuracy of this dataset are greatly improved; (4) The publicly available CDS toolbox and the open-source code makes it convenient for users to further analyse and process data. The main limitations are: (1) Although the data quality of ERA5 has been fully evaluated, due to the limitation of CDS toolbox platform, data source selection is restricted; (2) Because online computing needs to connect to CDS remote server, computing efficiency may be limited by network and concurrency, and sometimes queuing or even connection failure may occur. The CDS Toolbox provides a platform for a wide range of users from climate enthusiasts to researchers to obtain, calculate and display climate data online. It connects data with online computing capability through a programming interface. The CDS Toolbox provides a rich and easy-touse Application Programming Interface (API), and users can easily call the API and write custom Python applications. Because the complex calculation and drawing are carried out in the CDS server, the CDS toolbox greatly reduces the requirements of client computing and storage performance. Based on CDS toolbox, a multi-method apparent temperature and heat wave online calculation toolbox is developed, which realizes online calculation, display and real-time download of apparent temperature and heat wave data. Users can also edit the fully open code for further processing and analysis. GATHW includes two applications, global apparent temperature toolbox (GAT) and global heat wave toolbox (GHW), which are used to calculate daily apparent temperature and heat wave records respectively. Figure 9a shows the GUI (Graphical User Interface) of GAT. The three data sources can be selected via 'Data source' button to drive the toolbox. Note that the data have different available time ranges and spatial resolutions, which should be considered in the following toolbox settings. Specifically, the spatial resolution of ERA5  and ERA5 (1979 to present) is 0.25°, and 0.1° for ERA5-Land (1950 to present). Users can select the interested date range by setting 'Start date' and 'End date'. 'Grid' and 'Unit' determine the spatial resolution and unit of output data. 'Frequency' refers to the frequency of extracting hourly data to daily data. For example, Frequency = 3 means extracting data every three hours, which can reduce the burden of calculation and storage, but may lose accuracy. 'Statistic' determines the way of resampling hourly data to daily data, including average, minimum and maximum. 'Method' refers to three methods for calculating apparent temperature, corresponding to Section 2.2. "Extent" is the region of interest determined by latitude and longitude. "City" refers to the city of interest, and the time series of apparent temperature between the "start date" and "end date" of the selected city will be shown as a graph (Figure 9c). GAT allows real-time download of the calculated apparent temperature data in NetCDF format and can display the global apparent temperature of the first month online (Figure 9b). Figure 10a shows the GUI of the global heat wave toolbox. GHW and GAT have many similarities in interface, but heat wave calculation module is added. Heat wave record is calculated based on daily mean apparent temperature using combined threshold (constant threshold = 29°C, percentile threshold = 85%), the duration threshold is 3 days. Daily apparent temperature is calculated using Steadman TWPR "Heat wave method" refers to the heat wave calculation method, corresponding to Section 2.3. "Constant/ Percentile/Combined threshold" determines the hightemperature threshold of different heat wave calculation methods. "Duration threshold" determines another important heat wave threshold. GHW has 8 output files, which are apparent temperature and 7 heat wave characteristics, and can be downloaded instantly in NetCDF format. GHW supports online display of heat wave frequency ( Figure 10b) and heat wave total duration (Figure 10c).

| Code availability
Online computing and fully open code are important improvements of this study. Users can easily download the source code files (GAT.py and GHW.py) via https:// github.com/Eraer Yean/Globa l-Appar ent-Tempe ratur eand-Heat-Wave-GATHW -Toolbox and copy to the CDS online development platform (https://cds.clima te.coper nicus.eu/cdsap p#!/Toolbox) to run GATHW toolbox, online computing avoids the complex environment configuration program in local operation. In addition, users with F I G U R E 1 0 GUI and output of the global heat wave toolbox: (a) interface that allows users to customize parameters; (b) the graph of heat wave frequency; (c) the graph of heat wave total duration F I G U R E 9 GUI and output of the global apparent temperature toolbox: (a) interface that allows users to customize parameters; (b) the graph of monthly apparent temperature; (c) the graph of apparent temperature time series Python development foundation can edit the code for further analysis and processing.

| Data availability
As GATHW is an online computing toolbox, users can freely choose the time period, region, method, spatial resolution and even unit of interest to calculate and download the data instantly, so we only provide the global daily apparent temperature and annual summer heat wave data for 2006-2020 based on commonly used settings. Among them, the apparent temperature data were resampled from hourly data to daily by the maximum, and the apparent temperature was calculated based on Humidex TD, with a spatial resolution of 0.25° and a unit of °C. Heat wave data are calculated based on the apparent temperature data using combined threshold (constant threshold = 29°C, percentile threshold = 85%), and the duration threshold is 3 days. Data are stored in NetCDF and TIFF format, which can be accessed at https://www.doi.org/10.5281/ zenodo.4764325 (Yin & Yang, 2021).

| CONCLUSION
In this study, we developed a global apparent temperature and heat wave online toolbox and developed the gridded apparent temperature and heat wave dataset based on the toolbox. The online toolbox greatly improves the user's freedom in selecting time period, region, spatial resolution and method of interest, and supports online display and instant download of data, and greatly reducing the requirements of client computing and storage performance. After evaluation, this dataset can well reflect the typical extreme temperatures and heat wave events, and is more accurate, with higher resolution and faster update frequency than similar data products, which can provide data support for the study of human-environmental ecological processes and extreme climate events.