Analytical Study of the Changes in Brightness Temperature Based on the Tectonic Field Associated With Three Earthquakes in the Eastern Tibetan Plateau

The thermal infrared brightness temperature (BT) of the eastern Tibetan Plateau (TP) was retrieved from the Moderate Resolution Imaging Spectroradiometer (MODIS) level-1B data. The multiyear averaged BT background field was subtracted from the punctual BT data to yield monthly BT spatial anomaly, and calculated time series of BT for the secondary blocks. Then, the spatial and temporal changes in the BT of the study area before the Menyuan M6.4, Zaduo M6.2, and Jiuzhaigou M7.0 earthquakes were investigated and analyzed based on the tectonic setting. The results show the following. The spatial BT radiation enhancement frequency rose remarkably before strong earthquakes; each of the three earthquakes was preceded by marked spatiotemporal continuous BT anomalies. The tectonic setting significantly influences the BT anomaly feature. The spatial BT anomaly was not notable in the Qaidam and Qilian block before the Menyuan earthquake; the spatial BT anomaly mainly appeared in the Qiangtang and Bayan Har blocks before the Zaduo and Jiuzhaigou earthquakes. The Qiangtang and Bayan Har block’s BT time series curves have similar features. The Qaidam and Qilian block’s BT time series curves have analogous shapes. The three earthquakes may be regarded as one seismic event induced by a stage of tectonic stress enhancement rather than three independent occasions. The spatial BT anomalous behavior before earthquakes is, to a great extent, like the result of the rock stress loading experiment; the rock compression and the lithosphere–atmosphere–ionosphere coupling (LAIC) may be the main reasons for the intensification of the BT radiation.

Abstract-The thermal infrared brightness temperature (BT) of the eastern Tibetan Plateau (TP) was retrieved from the Moderate Resolution Imaging Spectroradiometer (MODIS) level-1B data. The multiyear averaged BT background field was subtracted from the punctual BT data to yield monthly BT spatial anomaly, and calculated time series of BT for the secondary blocks. Then, the spatial and temporal changes in the BT of the study area before the Menyuan M6.4, Zaduo M6.2, and Jiuzhaigou M7.0 earthquakes were investigated and analyzed based on the tectonic setting. The results show the following. The spatial BT radiation enhancement frequency rose remarkably before strong earthquakes; each of the three earthquakes was preceded by marked spatiotemporal continuous BT anomalies. The tectonic setting significantly influences the BT anomaly feature. The spatial BT anomaly was not notable in the Qaidam and Qilian block before the Menyuan earthquake; the spatial BT anomaly mainly appeared in the Qiangtang and Bayan Har blocks before the Zaduo and Jiuzhaigou earthquakes. The Qiangtang and Bayan Har block's BT time series curves have similar features. The Qaidam and Qilian block's BT time series curves have analogous shapes. The three earthquakes may be regarded as one seismic event induced by a stage of tectonic stress enhancement rather than three independent occasions. The spatial BT anomalous behavior before earthquakes is, to a great extent, like the result of the rock stress loading experiment; the rock compression and the lithosphere-atmosphere-ionosphere coupling (LAIC) may be the main reasons for the intensification of the BT radiation.
In contrast with the ground-based observation stations, spaceborne observation is immune to environmental and human interference and is unconstrained by terrain and observation conditions, so it can be utilized to monitor earthquake-prone zones continuously, accurately, and efficiently [5], [6], [7], [8].
However, how to retrieve the earthquake-related anomaly based on distinct parameters and strategies is a topic of general interest in seismological research. Moreover, how to explain these thermal radiation anomalies before earthquakes, especially under specific tectonic field, also play an essential role in earthquake precursor analysis.
From the perspective of data and method, it is seldomly found that thermal infrared brightness temperature (BT) and the anomaly method are applied in earthquake-related precursory research. BT is the temperature that a blackbody in thermal equilibrium with its surroundings would have to be at in order to duplicate the observed intensity of a gray body object at a frequency. Although BT data has been involved in seismic singularity research before, the products dominate the research data. No research has yet used the BT data retrieved from Moderate Resolution Imaging Spectroradiometer (MODIS) L1B product in earthquake-related information This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ study. In that it needs researchers to retrieve the BT data themselves, the process needs large storage equipment and is relatively complicated.
But on the other hand, the retrieved BT data has its advantages. It can directly reflect the radiation change of the land surface and is less interfered with by the atmospheric condition. Therefore, we used the MODIS Level-1B (MOD021KM) data to retrieve the BT data and try to find the thermal radiation anomaly before the target earthquakes. The anomaly method is the primary strategy for finding earthquakerelated singularity by deducting the background field. This strategy could eliminate the short-time disturbance from the atmosphere and the underlying surface. But this method also asks for a large quantity of data to build the background field, which also requires longtime work. The former research that used the anomaly method usually made the background field based on a few data. The research team of this study has been engaged in the thermal infrared anomaly research of the Tibetan Plateau (TP) for more than ten years, and there are adequate data to support the anomaly method research.
The consensus is that the following three mechanisms have been derived from laboratory and field observations to explain the relationship between the BT radiation anomaly and the earthquake.
1) Rock Compression Theory: Compressive stress often presents in a seismogenic zone during the seismogenic phase of an earthquake, leading to rock fracture and friction between rocks. Under compression or tension, rocks may produce heat that transfers outward from the lithosphere to the surface, affecting the atmosphere. Because this process generally occurs over a large area, the phenomenon can be recorded using techniques such as satellite imaging and is thus manifested as pre-earthquake TIR anomalies [27], [28]. 2) Land-Surface Degassing Theory: During the seismogenic phase, the compression and extension of the lithospheric medium lead to the upwelling of the material from subsurface heat sources. For example, geothermal springs can ascend to the land surface through rock gaps and create a cover layer on the land surface in the form of water vapor. Moreover, subsurface gases (e.g., greenhouse gases, including CO 2 and CH 4 , etc.) can rise to the land surface under compressive stress and produce a regional greenhouse effect in the vicinity of the seismogenic zone, thereby enhancing the radiation [5], [16]. 3) In addition, many studies have identified a lithosphereatmosphere-ionosphere coupling (LAIC) mechanism, according to which the lithosphere within a seismogenic zone is in a high-temperature, high-pressure state during the seismogenic process.
It can make the rock masses fracture or melt. As a result of the dissolution or phase transformation of the minerals, the daughter isotopes produced from the decay of the radioactive parent isotopes contained in the minerals or rocks are released in large quantities. Here, the release of radon is used as an example. The many α particles formed from the decay of radon ionize the near-surface atmosphere, leading to the formation of many charged ions. By inducing hydration in the atmosphere and reducing the relative humidity of the atmosphere, these ions create a local high-temperature, low-pressure environment and, on this basis, a local general environment with an increased total level of radiation [12], [19], [29], [30], [31], [32]. Even these mechanisms have explained the relationship between thermal radiation and earthquakes reasonably. Still, the former studies frequently list these mechanisms without carefully discussing which mechanism is the most probable reason. Consequently, we will try to ascertain the mechanism by referring to the former studies and the experimental result and give the possible explanation for the spatiotemporal anomalies before earthquakes. Therefore, we will use the retrieved BT data to find earthquake-related thermal radiation singularity before three contiguous earthquakes from January 2016 to August 2017 in the eastern TP and try to talk about the relationship between BT anomaly and the earthquakes based on the tectonic stress background field. In the meantime, we are giving our interpretation and inference of the relationship between these three earthquakes.

II. TECTONIC BACKGROUND AND THE SEISMIC EVENTS A. Tectonic Background Information of the Eastern Qinghai-Tibetan Plateau
The eastern margin of the TP (90-110 • E, 20-40 • N) was examined in this study. The TP uplifts under the collision and compression from the Indian subcontinent and the Eurasian continent, and the material in the TP eastward migration at a rate of approximately 45 mm/year [33]. This geodynamic setting has resulted in the formation of many secondary blocks (e.g., the Qiangtang and Bayan Har blocks) [ Fig. 1(a)] [33], [34], [35], [36], [37]. Due to the secondary block's different locations, the blocks within the TP differ in their geodynamic setting. Even for the same secondary block, its properties vary between locations. Fig. 1(a) depicts the displacement velocity field in the whole TP (the blue arrows). The directions of the field are different in the TP, it originates from the compression of the Indian plate [as indicated by ① in Fig. 1(a)], and the Eurasian continent [as shown by ② in Fig. 1(a)], the obstruction by the South China block (or the Yangtze block) in the east have collectively complicated the distribution of the displacement velocity field within the TP. The compression under this tectonic background field also makes unique tectonic background, geodynamic mechanism, and complex, large-scale fractures in the TP. Fig. 1(a) also reveals that the distribution pattern of the displacement velocity field can be divided along the 90 • E meridian [as indicated by the black dotted line in Fig. 1(a)]. The velocity direction in the western part of the TP (i.e., west of the 90 • E meridian) develops northwestward [as indicated by ③ in Fig. 1(a)]. In contrast, the direction in the eastern part of the TP (i.e., east of the 90 • E meridian) develops eastward and northeastward [as indicated by ④, ⑤, and ⑥ in Fig. 1(a)]. Meanwhile, the velocity direction that propagates eastward turns to the right and continues to extend southeastward while obstructed by the South China block, as indicated by the red arrows ⑦ in Fig. 1(a).  The above analysis shows that the eastern and western parts of the TP differ appreciably in stress properties. Therefore, it could not include the whole TP as the research region, and since the earthquakes in this research are in the eastern part of the TP [ Fig. 1(b)], these factors motivated us to set the eastern TP as our research region.

B. Seismic Events of This Work
The details of the earthquakes are in Table I. The study area experienced three consecutive major seismic events-a magnitude(M)-6.4 earthquake in Menyuan, an M6.2 earthquake in Zaduo, and an M7.0 earthquake in Jiuzhaigoubetween January 2016 and August 2017 [ Fig. 1(b)]. The Menyuan M6.4 earthquake occurred on the margin of the Qilian block, and the focal mechanism solution is Thrust type; the Zaduo M6.2 earthquake occurred within the Qiangtang block in the interior of the TP, and the focal mechanism solution is Normal fault type; and the Jiuzhaigou M7.0 earthquake was on the margin of the Bayan Har block, which is in the eastern part of the TP, and the focal mechanism solution is strike-slip type [39].
The occurrence of several moderate to strong earthquakes over a relatively short time may reflect the rapid change in the regional tectonic stress, and such continuous moderate to solid seismic events in a short time may also show that there may have a specific correlation between them. Even many studies have focused on the Jiuzhaigou M7.0 earthquake before. Nevertheless, it seldom finds works associated with these three seismic events together. Still, the seismogenic process always demands a comparatively long time, and much process information may be neglected if only put stress on singularity in a short time or before one specific earthquake. Consequently, in this work, we conduct research in combination with the three earthquakes and find the relations between these three earthquakes on account of the spatiotemporal BT development in a relatively long time.

A. Dataset and Processing
The data used in this study were captured by the MODIS instruments onboard the Terra satellites that fly over the study area at 10:00 A.M. (local time). The satellite operates in a Sun-synchronous orbit at an altitude of 705 km. The MODIS scans a swath of 2330 km with detection bands of 0.405-14.4 µm (36 bands, including the visible and TIR bands) and acquires data at three spatial resolutions (250, 500, and 1000 m). In this study, the BT was retrieved from MODIS data in band 21 (MOD021KM, wavelength: 3.93-3.99 µm; and spatial resolution: 1000 m). This band can yield surface and cloud temperatures at a high accuracy of 0.5 K, which is conducive to cloud removal.
The MOD021KM data selected in this study were received over the period 2004-2021 by a ground-based station that had operated continuously for more than 18 years since the completion of its construction in 2004. These data were processed using a MODIS processing software program developed in-house. Specifically, the MODIS data were atmospherically and geometrically corrected and reprojected. Images of each swath were then cropped based on the size of the study area, and the BT was retrieved simultaneously. The cropped images were combined into images with a temporal resolution of 1 day [40]. Since cloud cover is detrimental to the retrieval of the BT, a cloud removal procedure is also required to process the MODIS BT data. This study used the method used by the United States National Aeronautics and Space Administration (NASA) to produce MODIS products, as demonstrated by the following example.
The maximum value at each pixel over the preceding 8 days period is extracted to yield a product with a temporal resolution of 8 days. The land-surface parameters do not change abruptly. In addition, thermal radiation caused by solid earthquakes does not appear suddenly one day; instead, there is often an evolutional process. Hence, this study combined consecutive single-day data for a certain time to form multiday data. The maximum value at each pixel within this time was extracted as the BT value at the corresponding location. Experimental trials revealed that a 4 days period was most suitable for combining single-day data for the study area, as it allowed the effective elimination of the effects of cloud cover and ensured data validity.

1) Tracking of Spatial Changes in the BT:
The spatial changes in the BT were tracked by identifying BT anomalies from the data for the 18-year period from 2004 to 2021. Through repeated trials, we found that spatial BT anomalies calculated at time intervals of calendar months could reflect abnormal information to the greatest extent. Therefore, a background BT field was calculated for each of the 12 calendar months based on the data for the 18-year period [40]. Precisely, the spatial BT values for a particular month over the 18-year period were summed and then averaged to produce the background field for the month. Calculations over an extended time can effectively remove short-term or abrupt local changes and yield a background field that can objectively reflect the BT information of the investigated area. The obtained background field for each month was then subtracted from the punctual spatial BT data to yield spatial BT anomalies, as shown in (1). In the equation, S(x, y, t) is the punctual spatial BT data in the tth month, S(x, y, t i ) is the BT in the tth month of the ith year, and n is the total number of years (n = 18). Our findings from long-term tracking examining the eastern TP show that setting a BT threshold of 4 K can best accentuate spatial anomalies. The occurrence of an anomaly exceeding the threshold in tandem with sustained spatial evolution and the existence of the BT enhancement can be viewed as an abnormal phenomenon portending a future earthquake, and the abnormal phenomena often gradually disappear after earthquakes At the same time, it must be emphasized that there may be random spatial BT enhancement not caused by earthquakes, so it should identify the BT anomaly that may connect with the corresponding earthquakes. Considering many researchers have found TIR enhancement before earthquakes, in other words, there should have more spatial TIR enhancement anomalies before earthquakes. Based on this, we calculated the spatial BT radiation enhancement frequency in the eastern TP under the different fixed sliding windows. The result shows that when setting eight months as the sliding window, and if there are more than three months of TIR enhancement under the sliding window, there always have corresponding strong earthquakes. This method was established around 2014 and has withstood the test of implementation. Fig. 2 indicates the TIR enhancement frequency of the eight-month sliding window; it could be seen that the strong earthquakes always occurred during a high-frequency period, the Wenchuan M8.0 earthquake outbroke after continuous long time high frequency, the frequency before Yushu M7.1 earthquake is five, the frequency before Lushan M7.0 and Minxian M6.6 earthquakes are, namely, five and six times, Ludian M6.5 earthquake and Jinggu M6.6 earthquakes are five times. The enhancement frequency of the earthquakes in this research is six, six, and five times separately. Although there is more spatial TIR enhancement before solid earthquakes, it only documents the possibility of the corresponding earthquakes in the eastern TP. Finding the quantitative anomaly result also requires a more precise method.
2) Tracking of BT Time Series: Time series can quantitatively reflect BT evolution. Determining the spatial scale at which time series should be tracked has always been challenging. Some earlier studies included the entire study area in time-series tracking analysis. This approach often incorporates useless information into the calculation, preventing useful information from escaping the noise. Furthermore, some studies only included the area surrounding the epicenter in their time-series tracking analysis. This approach could sometimes find the singularities, but the seismogenic region often connects with the nearby tectonic zone so this method may miss some useful, abnormal information. Consequently, the two approaches described above are unsuitable for this study.
As mentioned earlier, the large secondary blocks have formed on the TP due to tectonic action; these blocks also have different tectonic background behavior from each other. The tectonic origin allows these blocks to reflect the stress evolution. Therefore, we undertook an analysis with the secondary blocks as the units. Specifically, punctual BT timeseries data were first obtained for the secondary blocks. Then, a background BT field was calculated for each secondary block, which was subtracted from the punctual BT time-series data to yield its BT anomaly time-series curve. The threshold for determining anomalies in time-series curves is generally set to 2δ or (mean + δ). Because a relatively long time series has a near-zero mean, setting the threshold to 2δ is a better choice, which we did.

IV. RESULTS
A. Analysis of Spatial BT Anomaly Before and After the Earthquakes 1) Spatial BT Anomaly Before and After the Menyuan M6.4 Earthquake: Fig. 3(a)-(i) shows the BT anomaly on the eastern margin of the TP over the period June 2015-February 2016. Fig. 3(a) reveals that the BT radiation of the Sichuan-Yunnan block increased, then starting in July 2015 [ Fig. 3(b)], the radiation from the eastern Qiangtang block began to intensify, with more than 21% of the area experiencing an increase in BT above the threshold (4 K). This trend continued through August 2015 [ Fig. 3(c)], and the enhancement of the thermal radiation extended from the eastern Qiangtang block to the Qilian block. The radiation enhancement became more pronounced in September 2015 [ Fig. 3(d)], particularly in the eastern Qiangtang block, with more than 33.4% of the area experiencing an increase in BT above the threshold. The region with high-level radiation intensification continued to expand from October to November 2015 [ Fig. 3(c) and (f)] and essentially covered the eastern Qiangtang and eastern Bayan Har blocks. BT rose above the threshold in 50% and 48% of the two blocks, respectively, in October. These percentages increased to 55% and 60.5% in November. Later in December, the radiation's spatial enhancement disappeared on a large scale [ Fig. 3(g)]. In the month when the earthquake occurred [January 2016, Fig. 3(h)], there was no marked abnormal increase in BT. The location of the radiation enhancement area develops from the west to the east and then expands to the whole eastern TP. These phenomena deserve special attention; if the earthquake-related TIR anomaly exists, its evolution should be primarily consistent with the TP's stress background field, which should have a development trend from the west to the east. Qu et al. [41] also emphasized that the seasonal TIR enhancement and the region-fixed enhancement at different times cannot be counted as the anomaly, as they may be caused by other factors. However, it could be figured out that there is no seasonal change in the BT enhancement from Fig. 3. Meanwhile, the region of the TIR enhancement is also not changeless in Figs. 3-5; they have the feature of gradual transformation, from one direction to another, from small to large, which we will discuss in detail in the following content.
However, after the earthquake, in February 2016 [ Fig. 3(i)], the BT radiation intensified again, predominantly in the eastern Qiangtang and eastern Bayan Har blocks, approximately 32% and 14% of which experienced an increase in BT above the threshold, respectively. Available research findings suggest that spatial radiation intensification gradually disappears after an earthquake [40]. However, as seen in Figs. 3 and 4, the abnormal enhancement of the BT radiation persisted after the earthquake. This finding suggests that the Menyuan M6.4 earthquake did not terminate the TIR anomaly event within the TP but that the anomaly continued to evolve after the earthquake.
2) Spatial BT Anomaly Before and After the Zaduo M6.2 Earthquake: Fig. 4(a) Fig. 4(a)]. Analysis of the spatial distribution of the increase in BT reveals that the region with an increased level of radiation tended to expand toward the northeastern part of the TP, including the eastern Bayan Har, eastern Qaidam, and Qilian blocks. These three blocks experienced an increase in BT above the threshold in 43%, 15%, and 14% of their areas, respectively. In the following April, the abnormal enhancement of the radiation essentially disappeared in the eastern part of the TP [ Fig. 4(b)]. However, in May 2016, the radiation emitted from the eastern Qiangtang block intensified again [ Fig. 4(c)], with an increase in BT above the threshold of over 32% of the area. Then, the spatial increase in the BT radiation continued and extended successively to the eastern Bayan Har, eastern Qaidam, and Qilian blocks [ Fig. 4(d)-(e)]. The proportions of these three blocks with an abnormal increase in BT above the threshold peaked at 47%, 10%, and 8%, respectively, in August 2016 [ Fig. 4(f)]. In the following September and October (when the Zaduo earthquake occurred), the radiation did not intensify, as evidenced by an absence of BT enhancement over a large area [ Fig. 4(g) and (h)]. It is not hard to find that the BT radiation feature before the Zaduo M6.2 earthquake is like the feature before the Menyuan M6.4 earthquake; that is, the area of the BT enhancement grew larger before the earthquake, and the location of the area initiated from the west and expanded to the east later, then when the radiation starts decreasing, the earthquake occurred.
However, in November 2016 [ Fig. 4(i), after the Zaduo earthquake], the radiation from most parts of the eastern Qiangtang and eastern Bayan Har blocks intensified, with an increase in BT above the threshold of 22% and 30% of their areas, respectively. Moreover, the BT of the northern part of the Sichuan-Yunnan block increased; meanwhile, according to the result from Fig. 5, it is not hard to figure out that the spatial anomaly still exists.
3) Spatial BT Anomaly Before and After the Jiuzhaigou M7.0 Earthquake: Fig. 5(a)-(l) shows the BT anomaly on the eastern margin of the TP over the period December 2016-November 2017. As indicated in Fig. 5(a), after November 2016 [ Fig. 4(i)], the BT remained at abnormal levels in December 2016. In addition, the region with an increased level of radiation extended once again to the eastern Qaidam and Qilian blocks, that is, across the TP to its northeastern part. This trend continued in January 2017 [ Fig. 5(b)]. In February 2017 [ Fig. 5(c)], the intensification of the radiation began to decrease in both range and area. Nevertheless, the BT of most parts of the TP remained at abnormal levels. After that, the BT of the TP fluctuated over a large area. Specifically, in March 2017 [ Fig. 5(d)], the BT decreased to low levels over a large area. The low values of the BT of 40%, 62%, 35%, and 27% of the eastern Qiangtang, eastern Bayan Har, eastern Qaidam, and Qilian blocks, respectively, were below −4 K. In April, the BT was low in the southern and northern parts and high in its central part The intermittent increases in the BT may have been related to post-earthquake stress adjustments or local changes in weather conditions [6], [10].
The spatial radiation behavior before the Jiuzhaigou M7.0 earthquake differs from the former two earthquakes. Even though there was a migration of radiation from east to west from November 2016 to January 2017, the radiation area showed a substantial increase and slump before the earthquakes.

B. Analysis of the Changes in the BT Anomaly Time-Series Curves Before and After the Earthquakes
Section IV characterizes the abnormal changes of the spatial BT in the TP before and after the Menyuan M6.4, Zaduo M6.2, and Jiuzhaigou M7.0 earthquakes. However, a more qualitative analysis is required to track BT anomalies. We drew time-series curves to track the characteristics of the changes in the BT. Fig. 6(a)-(d) depict the BT anomaly time-series curves for the eastern Qiangtang, eastern Bayan Har, eastern Qaidam, and Qilian blocks, respectively, over the period 2014-2017. To analyze the characteristics of the changes in each curve, we drew a zero-value line (red dotted line) and lines indicating ±1.5δ and ±2δ (black dotted line). Fig. 6 shows similarities in pattern between the curves for the eastern Qiangtang and eastern Bayan Har blocks and between the curves for the eastern Qaidam and Qilian blocks. Tectonically, the eastern Qiangtang and eastern Bayan Har blocks belong to the paleo-Tethyan orogenic belt located at the center of the TP. This factor is responsible for the affinity between the time-series curves for these two blocks. In contrast, the eastern Qaidam and Qilian blocks belong to the proto-Tethyan orogenic belt located at the terminal end of the TP, which explains the similarities in pattern between their time-series curves [39].
The pronounced increase and decrease changes can be observed in the BT anomaly time-series curve before these three earthquakes. The analysis of the four-year period data in Fig. 6  ), but which also approach the threshold line. Later the curves rapidly decreased to the trough and then quickly increased to another peak; the Jiuzhaigou M7.0 earthquake occurred and coincided with the later upward trend.
The fluctuation of the time series curves associated with the spatial BT radiation change presents a roller-coaster change before the Jiuzhaigou M7.0 earthquake. Meanwhile, after the earthquakes, despite fluctuations, the BT anomaly fluctuated within the normal range. In addition, it also should be noticed that, albeit the time series curves present substantial change, the curve is hard to surpass the +2δ threshold line. As seen in the 4-year-long curves, the curves constantly overpassed the threshold line around the time node before the earthquakes. If it exceeds the +2δ line, there must be a succeeding earthquake. It indicates that the stated threshold line is applicable in the research and reflects that the BT time series curves could lay notable anomalies before the earthquakes.

A. Relationship Between Spatial BT Anomalies and Tectonic Stress Development
It can draw from Figs. 3-5 that the ascending and descending of the BT radiation in different blocks. However, a phenomenon of interest occurred before the Meyuan M6.4 earthquake. The times of BT radiation intensification in the eastern Qiangtang block were five (July-November 2015) and three times in the eastern Bayan Har block (August and October-November 2015). However, the earthquake occurred in the Qilian block, which only has two times the radiation enhancement phenomenon (August and October 2015). Besides this situation, the Sichuan-Yunnan block endured  Fig. 4(i)], but the Sichuan-Yunnan block is not bound up with the earthquakes, considering that there must have seismogenic stress before strong earthquakes, the seismogenic stress is often connected to the blocks which will proceed with the earthquakes. Therefore, the radiation enhancement in the Sichuan-Yunnan block is mainly influenced by the intensification of other blocks. These occasional increases do not represent it's the main stress accumulated area. Except for the occasional intensification area, the repeated BT radiation enhancement region could better reflect the effect of stress accumulation in that the anomaly induced by the seismogenic stress should be persistent in space and should generate in a relatively fixed place [43], the fixed place of which has continuous stress loading or evolution and corresponds to the later earthquakes.
Consequently, for a more visual illustration, we averaged the BT anomaly (only the one which has enhancement anomaly) over June-December 2015 [ Fig. 7(a)] to accentuate the region with repeated increases in radiation intensity during this period and eliminate the occasional non-abnormal information while preserving the abnormal information. The repeated increase region could be called the "heating core" [43]. Furthermore, we averaged the BT anomalies before the Zaduo M6.2 and Jiuzhaigou M7.0 earthquakes [ Fig. 7(b) and (c)] and over the entire period examined in this study [ Fig. 7(d)].
Observation of Fig. 7(a) illustrates the following: Before the Menyuan M6.4 earthquake, the "heating core" occurred primarily in the Qiangtang and Bayan Har blocks and, to a minimal extent, in the Qaidam and Qilian blocks, but the epicenter is in the margin of Qilian block. Three factors may be responsible for this phenomenon: 1) the Qilian and Qaidam blocks are located at the northernmost end of the propagation path of the TP stress; 2) the GPS measurements show this region's relatively lowest strike-slip rate of approximately 1-2 mm/year; and 3) stress accumulation has formed a series of folded tectonic zones in this region; in contrast, the thrust fault zones have developed a series of basin and range structures, reflecting the upper crust's continuous deformation (compression, shortening, and thickening) [35], [39]. Therefore, combined with the above reasons, the "heating core" did not occupy the Qaidam and Qilian blocks before the Menyuan M6.4 earthquake may be mainly due to the that these two blocks have relatively slow strike-slip rate and are on the end of the stress passageway, that gives rise to the slow accumulation of seismogenic stress in this region, which together with the region's folded topography, collectively cause most of the energy originated from the stress be absorbed and adjusted other than outward emission, and finally lead to the low BT radiation in these two blocks [36], [43], [44]. Meanwhile, the focal mechanism solution of the Menyuan M6.4 earthquake is thrust type [ Fig. 1(b)]. The thrust type means the earthquake occurred under stress extrusion; if the high value of BT time series represents the stress loading, the focal mechanism solution of this earthquake could be the confirmation of the result of Fig. 6, which shows Menyuan M6.4 earthquake took place during the BT high-value stage. In addition, the co-seismic deformation field from INSAR found the vertical deformation is uplifting, indicating there was stress extrusion before the Menyuan M6.4 earthquake [45].
The focal mechanism solution of the Zaduo M6.2 earthquake is normal fault type and with the left-handed strike-slip component, indicating that the seismogenic fault was in a tense state when the earthquake occurred. Fig. 6 could supplement this result; the earthquake is outbroken when the time series is during the trough stage. Meanwhile, considering the outcome of the rock compression experiment, which shows the TIR's low value may present the stress tension or unloading status [46], [47], [48]. Fig. 7(b) depicts the average BT anomaly from February to September 2016. It is obvious that the BT's "heating core" occurred predominantly in the eastern Bayan Har and Qiangtang blocks. Combining the result of focal mechanism resolution and the BT spatiotemporal development result, there must have been a stress compression or tensioning situation; the shape of the time series curves before the Zaduo M6.2 earthquake also shows that the stress in the eastern TP experienced loading and rapidly unloading before Zaduo M6.2 earthquake, meanwhile the "heating core" also suggesting a continuous increase in the stress may be in the "heating core" region. Compared to the Menyuan M6.4 earthquake, the "heating core" location is more eastward, indicating that the tectonic stress has an eastward movement trend. The Zaduo M6.2 earthquake occurred on the back of the "heating core." The same phenomenon could also be witnessed in Fig. 4. Considering the tectonic stress development direction is from west to east in the TP, the Zaduo M6.2 earthquake may be a triggered seismic event on the back of the stressed channel during the stress's eastward development. Yang et al. [40] found that middle-magnitude earthquakes may have occurred before the final strong earthquake, and the final earthquake, more often than not, took place on the frontier of the "heating core." Fig. 7(c) indicates the average spatial BT anomaly in November 2016-July 2017 before the Jiuzhaigou M7.0 earthquake. It could be seen that the region of the BT "heating core" extended along the West-East direction, from the eastern Qiangtang block to the eastern Bayan Har block and eventually out of the TP, compared with Fig. 7(a) and (b), the "heating core" more centered at the eastern TP. On August 8, 2017, an M7.0 earthquake occurred in Jiuzhaigou on the eastern end of the radiation intensification region. The focal mechanism solution of the Jiuzhaigou M7.0 earthquake is strike-slip type. The main stress compression direction is the West-East orientation and the main stress tensioning direction is the South-North orientation, which demonstrates that this earthquake occurred under the West-East orientation stress squeeze movement of the Bayan Har block [49], [50], [51], the expansion feature of the "heating core" is West-East orientation, if the "heating core" could reflect the seismogenic stress, it is expansion is the same as the focal mechanism solution. Chong et al. [51], [52] took advantage of the strain observation result (including data from the cave strainmeter stations, the four-gauge borehole strain meter stations, and the volumetric borehole strain meter stations) around the epicenter of Jiuzhaigou earthquake in 300 km, using the LURR (load/unload response ratio) method found that there is stress accumulation in the northeastern TP before the Jiuzhaigou earthquake, the gradually stronger accumulation of stress finally gave rise to the continuous damage and weakening of rocks, and caused the Jiuzhaigou earthquake eventually. The strain observation result starts an anomalous transition from far to near the epicenter in 1-2 years before the Jiuzhaigou earthquake and an obvious LURR anomaly. The rock stress experiment also indicates that the TIR would expand from one side to another under the stress loading situation, just like the evolution of the BT development in the research [47]. The spatial BT evolution and the strain observation station's result are consistent. One more point, the strain observation stations are established on the bedrock or embedded in the bedrock, and the results can reflect the rock's stress change. The BT's spatiotemporal result has the same change feature as the strain observation, which could indicate that stress change exists before earthquakes [51], [52].
It can also be seen from the average BT anomaly of June 2015-July 2017 in Fig. 7(d) that the "heating core" occurred primarily within the Qiangtang and Bayan Har blocks, through the previous analysis combined with other data. This phenomenon suggests that these blocks constituted the central region under compressive stress during the research time. The BT "heating core" location also makes clear that the stress keeps eastward development. Considering that there are three directions of the stress in the eastern TP, the northeastward, the eastward, and the southeastward [ Fig. 1(a)]. From the distribution feature of the "heating core," the principal direction of the stress during the research period is from west to east [ Fig. 7(c) and (d)] and accompanied by the northeastward stress component [ Fig. 7(a) and (b)], these earthquakes outbroke under this stress development background.

B. Relationship of the BT Anomaly Time-Series Curves and the Tectonic Stress Background Field
The aforementioned BT fluctuations in Fig. 6, particularly marked pre-earthquake fluctuations, in the time-series curves reflect the changes in the regional stress field. These phenomena suggest a general and repeated alternation between compressive and tensile stresses during the seismogenic phase.
Further in-depth analyze the BT time curves' different features in distinct secondary blocks and the blocks' response during the seismogenic process requires taking the tectonic background field into account. The terrain, part of the main faults, and the distribution of the secondary blocks within the TP are shown in Fig. 8(a). The Qiangtang block features an undulating terrain dominated by normal and conjugate fault zones on its west side and is home to large fault zones that primarily strike from the east to the west on its east side [36]. The anomaly of the BT time series curve of the eastern Qiangtang block did not exceed the threshold before either the Menyuan M6.4 or the Zaduo M6.2 earthquake but surpassed +2δ before the Jiuzhaigou M7.0 earthquake [ Fig. 6(a)], which may be attributed to the following factor. The Qiangtang block is in the middle position of the TP. The stress of eastward or northeastward immigration will always pass through it [ Fig. 1(a)], which makes the block obtain the feature of the stress transition zone. This behavior makes the Qiangtang block has less stress accumulation and more stress transition. This situation resulted in even the BT time series reaching a high value but did not overpass the threshold line before these two earthquakes. But the time series overpast the threshold line before the Jiuzhaigou M7.0 earthquake, which probably implies that the seismogenic stress is remarkably stronger than the two former earthquakes. Of the Bayan Har block, it has had frequent strong (M ≥ 7.0) earthquakes over the last two decades, with epicenters mainly located on its margin [ Fig. 8(b)]. Considering the distribution pattern of the strong earthquakes around the Bayan Har block, Li concluded that the Bayan Har block undergoes deformation on the boundaries but little or no deformation in the interior [35]. That is, the Bayan Har block exhibits the properties of a rigid block. The rigidity of the Bayan Har block makes it more effective and sensitive at reflecting seismic events than the other three blocks. A rigid nature reduces the block's energy dissipation and highlights the stress state across the block. This may be why the BT curve in the Bayan Har block surpasses the threshold line before the three earthquakes [ Fig. 6(b)].
The spatial BT radiation enhancement was not prominent before the Menyuan M6.4 earthquake in the eastern Qaidam and Qilian blocks. Still, the time series curves could indicate that the BT value of these two blocks sustained a relatively high-value stage. This phenomenon demonstrates that these two blocks have endured continuous, slow, and relatively uniform stress loading before the earthquakes, which is consistent with the tectonic stress feature discussed above. The BT time series anomaly in the eastern Qaidam block exceeded the threshold before the Zaduo M6.2 earthquake [ Fig. 6(c)]. It could also be seen in Fig. 4 that there were three months endured spatial BT radiation enhancement (March, June to August 2016) before the Zaduo earthquake. Meanwhile, the time series curve approached the threshold line before the Jiuzhaigou M7.0 earthquake in the eastern Qaidam block. The Qilian block also approached the threshold line before Zaduo M6.2 and exceeded the threshold line before the Jiuzhaigou M7.0 earthquakes. These phenomena in the two blocks illustrate that stress accumulation is relatively complex during the seismogenic process. On the one hand, the stress accumulation has continuous influence in the Qaidam and Qilian blocks; on the other hand, the BT radiation frequently intensified in these two blocks before the Zaduo earthquake, but after the Zaduo earthquake, they are not the "heating core" dominated region. Consequently, it infers that the seismogenic stress intensified northeastward and eastward in the TP before the Zaduo earthquake but principally developed eastward after the Zaduo earthquake and has the northeastward stress component at the same time.

C. Connection of These Three Continuous Earthquakes Based on the 18 Years' Time Series Result
Three continuously strong earthquakes happened in the eastern TP from January 2016 to August 2017. Meanwhile, we can draw that the spatial BT radiation keeps intensified since June 2015, and the time series also fluctuated in the high value. Still, after the Jiuzhaigou M7.0 earthquake, the spatiotemporal anomaly gradually disappeared, so these phenomena may throw light on that these three earthquakes could be caused by one-time continuous stress intensification from 2015 to 2017 in the TP. To better demonstrate this viewpoint, Fig. 9 reveals the BT anomaly time-series curve for the eastern Bayan Har block over 2004-2021. The reason why this block is selected is for its "rigid" feature, which makes it could obtain the changes of stress development in the TP better. This curve can be divided into five stages ( Fig. 9). After the Wenchuan M8.0 earthquake, the BT anomaly exceeded the threshold in 2009 for another time. This situation connects to the energy trimming after the Wenchuan earthquake; the multiple pre-and post-earthquake stress increases and adjustments suggest that a large amount of energy converged throughout the Wenchuan M8.0 earthquake. Stage [B] that spanned the period 2009-2012 was a relatively quiet period in terms of BT radiation (Fig. 9 [B]) that witnessed no marked seismic events. Stage [C] began before the Lushan M7.0 earthquake that occurred on April 20, 2013, and lasted until after the Jiuzhaigou M7.0 earthquake on August 8, 2017 ( Fig. 9 [C]). During this stage, the anomaly of the BT exceeded the threshold on multiple occasions that corresponded well to the subsequent earthquakes. Then, 2018 and 2019 constituted another quiet period regarding BT radiation ( Fig. 9 [D]). It was not until after 2020 that the radiation intensified, and the BT anomaly exceeded the threshold (Fig. 9 [E]), after which an M7.4 earthquake occurred in Madoi on May 22, 2021.
Observation of Fig. 9 distinctly shows a staged cycle of increases and decreases in the TIR of the TP. The TP has regular stress development that originates from the compression of the Eurasian Continent [as shown in Fig. 1(a)]. Still, there may have been staged stress enhancement overlayed on the regular stress. This factor causes the TIR anomaly, as shown in Figs. 2 and 9 has the same feature. The time span of the high frequency of TIR enhancement corresponds to the [A]/[C] stage of the Bayan Har block. This staged stress development of the TP triggered a series of high-magnitude earthquakes. The three seismic events examined in this study took place in this general setting. In stage [C], the three seismic events, starting with the Menyuan M6.4 earthquake and ending with the Jiuzhaigou M7.0 earthquake, were followed by a quiet period in terms of radiation. Therefore, combining the BT spatiotemporal anomaly of the eastern TP and the time curve series in the Bayan Har block from 2004 to 2021. These three consecutive earthquakes are one seismic event caused by one stage of the tectonic stress enhancement in the TP rather than three independent earthquakes.
Combining the former discussion and the outcomes from the focal mechanism solution and the strain observation, we may boldly speculate that there are two development passageways of stress in this stage. One of them spread to northeastern TP through Qiangtang Block [refer to Fig. 1 Fig. 1(a) ⑥], the Zaduo M6.2 and Jiuzhaigou M7.0 earthquakes occurred in this period. In addition, this stage of stress intensification was quietly released after the Jiuzhaigou earthquake, as the time series also turns to the earthquake quiet period (Fig. 9 [D]).

D. Mechanism Behind Pre-Earthquake BT Radiation Anomalies
Based on the three main mechanisms mentioned before, these three earthquakes are closely related to rock compression. Since the BT radiation repeatedly fluctuates before earthquakes is similar to the rock fracture stress loading experiment result. The result illustrates that the rock's volume will reduce and the temperature will increase when loading the stress on the rock, and the result also reverses when unloading the stress. Meanwhile, the experiment also demonstrates that the radiation increases slowly at the first phase of stress loading. The radiation will start decreasing when the rock produces microfracture at the elastic deformation stage; then, the radiation intensifies again before the rock's total fracture. In addition, the decrease occurred as the rock developed layered fractures, and the later increase occurred as the rock developed surface fractures [27]. It can be seen from Figs. 3-5 that the BT radiation around the seismogenic zone decreased conspicuously from June to July 2015 before the Meyuan M6.4 earthquake. The decrease phenomenon could also be found in the south and the nearby area of the Zaduo M6.2 earthquake's epicenter from June to July and September to October in 2016. The BT radiation's low-value phenomenon is more remarkable from March to June 2017 before the Jiuzhaigou M7.0 earthquake, and the low-value area has almost occupied the whole TP. Just like the experiment, the decrease is not only a single phenomenon; there is a shift from a decrease to an increase of BT radiation before the Menyuan and Jiuzhaigou earthquakes and an increase to a decrease before the Zaduo earthquake. The BT radiation before earthquakes documented the same feature as the laboratory result; the strain observation's result also demonstrates that rock compression and fracture did exist before the Jiuzhaigou earthquake [51], [52]. Therefore, the BT radiation's anomaly before earthquakes in this research mainly results from the rock fracture's different phases under stress compression. It needs to emphasis that the radiation decrease phenomenon is a complex issue, except for the rock fracture mode, which may also be caused by the stress tensile. Taking the BT radiation and the other scholar's results together, the Menyuan and Jiuzhaigou earthquakes occurred under the stress compression stage, the Zaduo earthquake outbroke under the tensile stage, and the time series quickly changed before Zaduo and Jiuzhaigou earthquakes indicated that the seismogenic stress fluctuated before the earthquakes. Overall, the infrared radiation anomaly exists due to rock fracture caused by stress compression.
At the same time, research has found spatial anomalies in total electron content (TEC) before the Jiuzhaigou M7.0 earthquake [23]. The rock's melting will create many P-holes (positive hole). The P-hole is the electric charge's carrier and generates electric current along the stress gradient direction. If there are adequate P-hole delivered to the land surface, it may stimulate the ionization of air molecules and finally increase the TEC in the lower ionosphere; additionally, the P-hole may also recombine to form peroxy bonds and transform into a vibrationally excited state on the land surface, particles in this state are most likely to release more energy by releasing infrared photons, thus generating thermal infrared anomalies [53], [54], [55], [56], [57]. Hence, we consider that the rock melting is one of the main reasons causes the BT anomaly. The LAIC would activate this process, so the LAIC probably be another main reason leading to the BT radiation anomaly. As for whether there was a "degassing" effect, we have not found relevant research results. Consequently, to sum up, the BT radiation anomaly in this research relates to the rock compression and LAIC action, and the "degassing" effect may play a role in the seismogenic process.

VI. CONCLUSION
In this study, three marked moderate to strong earthquakes that occurred on the eastern margin of the TP between January 2016 and August 2017 were analyzed based on the MODIS BT data for the 18-year period from 2004 to 2021. The spatial anomaly has been used to track the TIR enhancement information; the secondary block's BT time series has been applied to determine quantitatively whether a BT anomaly exists. The spatiotemporal features of BT anomaly and the connection between these three earthquakes are discussed based on the tectonic stress background field. Utilizing the focal mechanism solution and the strain observation result to assistant explain the stress change before the earthquakes. Meanwhile, from the standpoint of the seismogenic process, combined with the rock stress loading experiment, the possible reasons for the BT radiation anomaly also are discussed. The conclusion is as follows.
1) The spatial and temporal BT anomalies coexisted before earthquakes and fit together on changes. The frequency of spatial BT enhancement significantly rose before strong earthquakes; there exists obvious precursor singularity in spatial and temporal BT anomaly; combining with the "heating core," the spatial BT anomaly could indicate the TIR enhancement region, the time series curves could show the quantitative change in the BT. 2) The BT radiation anomaly is influenced by the tectonic background. Qiangtang and Bayan Har blocks have the same time series curve shape, and the Qaidam and Qilian blocks share the same pattern in time series. This phenomenon relates to the tectonic location of these blocks. Taking the focal mechanism solution and the anomaly from the strain observation into account, seismogenic stress change exists before earthquakes. The BT has the ability to capture this kind of stress change, but no matter the BT spatial change or the time series variety, it is all influenced by the tectonic background field, and the influence is noticeable and remarkable. 4) The BT anomaly may closely connect with the rock compression and the LAIC effect. The spatial BT anomaly and the secondary block's BT time series anomaly's variation characteristics are similar to the rock stress loading experiment. The strain observation also demonstrates that the rock fracture did exist before the earthquakes. The LAIC effect may also cause radiation enhancement and TEC anomaly. There may exist the "degassing" effect.