Association of land surface temperature anomalies from GOES/ABI, MSG/SEVIRI, and Himawari-8/AHI with land earthquakes between 2010 and 2021

Figures & data

Figure 1. The general architecture of the LST data processing and anomalies calculations.

Table 1. LST data quality products.

Satellite/Sensor	SPR	TMR	Flag	Flag value
MSG/SEVIRI	1 km	15 min	QC	0	1	2	4	8	64
				Result OK	Data unevenly distributed	Diurnal LST variation too small	The data gap is too large	Too few data points	The maximum number of iterations
Himawari-8/AHI	2 km	1 h	Q_Flags	0	1	2	3	4	> 8
				Sea pixel	Cloud free pixel	Pixel contaminated by clouds	Cloudy pixel not produced, poorly calibrated, and processing skipped	Pixel filled by clouds	Unprocessed pixel and pixel out of disk
GOES/ABI	4 km	1 h	DQF	0	2	4	8	16	32
				Good retrieval valid input data, type, and LST Clear conditions	Wrong or missing input data	Cloudy conditions	Degraded due to LZA threshold being exceeded	Invalid due to water surface type	Invalid due to out-of-range LST

Figure 2. Geographic distribution of earthquakes occurring between 2010 and 2021 with M_w ≥ 4.

Table 2. Classification details of magnitude, depth, and altitude.

Magnitude		Depth		Height above sea level
Classes	Range (Mw)	Classes	Range (km)	Classes	Range (m)
Mw = 7	≥ 7	DP5	> 400	AL4	> 4000
Mw = 6	6–6.9	DP4	201–400	AL3	3001–4000
Mw = 5	5–5.9	DP3	101–200	AL2	2001–3000
Mw = 4	4–4.9	DP2	51–100	AL1	1001–2000
		DP1	21–50	AL0	0–1000
		DP0	0–20

Figure 3. Statistics of the 30,248 land earthquakes between 2010 and 2021.

Figure 4. The map location of the 8 case studies presented.

Table 3. Details of the 8 case studies presented.

ID	Date	Latitude	Longitude	Depth	Mw	Elevation	Country	Land cover
ID7347	2010-04-04	32.29	−115.3	9.99	7.2	25	Mexico	Shrub
ID25080	2011-01-01	−26.8	−63.14	576.8	7	187	Argentina	Grassland
ID100966	2016-01-30	53.98	158.55	177	7.2	445	Russia	Shrub
ID103900	2016-04-15	32.79	130.75	10	7	160	Japan	Forest
ID110242	2016-09-10	−1.04	31.62	23	6	1211	Tanzania	Cropland
ID112123	2016-10-30	42.86	13.1	8	6.6	458	Italy	Cropland
ID157456	2020-01-24	38.43	39.06	10	6.7	997	Turki	Grassland
ID161547	2020-05-15	38.17	−117.85	2.7	6.5	1671	USA	Shrub

Figure 5. LST data processing and anomalies calculation flowchart in GIS software.

Figure 6. Visualization of LST maps during GIS processing.

Figure 7. Map visualization of a) pixel percent, b) STD of detrended LST (°C), c) median of detrended LST (°C), and d) mean of detrended LST (°C) between 2010 and 2021.

Table 4. The details of the used raster configuration and grid sizes.

Extent	[−178°, +178°, −60°, +60°]
Pixel size	1°	0.5°	0.1°
LPF scale (pixels)	1	2	5
X-axis (Longitude pixels)	346	712	3560
Y-axis (Latitude pixels)	120	240	1200
XY Matrix pixels	42,720	170,880	4,272,000

Table 5. Definition of the confusion matrix elements and derived parameters.

Condition Positive (CP)	TP	FN	True positive rate (TPR) = TP/P	False negative rate (FNR) = FN/P
Condition negative (CN)	FP	TN	False positive rate (FPR) = FP/N	True negative rate (TNR) = TN/N
Prevalence (P) = CP/CP + CN	Accuracy (ACC) = TP + TN/CP + CN		Positive likelihood ratio (LR+) = TPR/FPR	Negative likelihood ratio (LR-) = FNR/TNR
G-mean = $\sqrt{\frac{\mathit{\text{TP*TN}}}{\left(\mathit{\text{TP}}+\mathit{\text{FN}}\right)*\left(\mathit{\text{TN}}+\mathit{\text{FP}}\right)}}$		Diagnostic odds ratio (DOR) = LR +/LR−
D = $\frac{1}{2}$ - $\frac{\mathit{\text{FPR}}}{2}$ + $\frac{\mathit{\text{TPR}}}{2}$

Figure 8. The CMs of different methods, grid sizes, and FSR.

Figure 9. Confusion Matrix parameters, ACC, G-mean, D, AUC, and DOR figures of merits for LST anomalies by earthquake magnitude (from Mw = 4 to 8).

Figure 10. Confusion Matrix parameters, ACC, G-mean, D, AUC, and DOR figures of merits for LST anomalies by hypocenter depths (from DP0 to DP5).

Figure 11. Land composition.

Figure 12. Confusion Matrix parameters, ACC, G-mean, D, AUC, and DOR figures of merits for LST anomalies by land cover type.

Figure 13. Confusion Matrix parameters, ACC, G-mean, D, AUC, and DOR figures of merits for LST anomalies by altitude.

Table 6. The average LST anomalies (°C) classified for M_w > 6, and 8 a) land covers, and b) 3 elevation classes.

(a)
Mw	LC	Mean LST anomalies	Days before	Record
8	Forest	0.56	−6	1
7	Grassland	0.31	−3	4
7	Forest	0.5	−3	9
7	Shrub	0.82	−5	5
7	Sabanas	0.7	−3	1
7	Mosaic	0.32	−3	2
7	Cropland	0.7	−5	1
6	Grassland	0.31	−4	10
6	Forest	0.41	−3	33
6	Shrub	0.42	−3	13
6	Sabanas	0.67	−2	1
6	BSV	0.92	−2	9
6	Mosaic	0.12	−3	2
6	Cropland	0.66	−3	10
(b)
Mw	Depth	Mean LST anomalies	Days before	Record
8	DP3	0.56	−6	1
7	DP5	0.69	−5	4
7	DP3	0.63	−5	4
7	DP1	0.28	−3	5
7	DP0	0.32	−3	9
6	DP5	0.47	−4	6
6	DP4	0.65	−5	2
6	DP3	0.32	−4	10
6	DP2	0.29	−4	4
6	DP1	0.44	−2	22
6	DP0	0.46	−3	34

Table 7. The average LST anomalies (°C) classified over M_w > 6.

Mw	Altitude	Mean LST anomalies	Days before	Record
8	AL0	0.56	−6	1
7	AL0	0.62	−4	16
7	AL1	0.31	−3	4
7	AL2	0.1	−2	2
6	AL0	0.79	−3	61
6	AL1	0.49	−3	13
6	AL2	0.09	−4	4

Table 8. The measured LST anomalies (°C) for all strong studied earthquakes between 2010 and 2021.

ID	Date	Latitude	Longitude	Depth	Mw	Elevation	Country	Land cover	Anomaly 50 km	Anomaly 100 km	Days before
ID7347	2010-04-04	32.29	−115.3	9.99	7.2	25	Mexico	Shrub	0.29	0.33	−1
ID25080	2011-01-01	−26.8	−63.14	576.8	7	187	Argentina	Grassland	0.22	0.23	−3
ID100966	2016-01-30	53.98	158.55	177	7.2	445	Russian Federation	Shrub	0.67	0.75	−5
ID103900	2016-04-15	32.79	130.75	10	7	160	Japan	Forest	0.44	0.31	−1
ID110242	2016-09-10	−1.04	31.62	23	5.9	1211	Tanzania	Cropland	0.12	0.03	−5
ID112123	2016-10-30	42.86	13.1	8	6.6	458	Italy	Cropland	0.53	0.12	−1
ID157456	2020-01-24	38.43	39.06	10	6.7	997	Turkey	Grassland	0.71	0.84	−3
ID161547	2020-05-15	38.17	−117.85	2.7	6.5	1671	United States of America	Shrub	0.09	0.15	−3

Figure 14. LST detrended and LST anomaly maps for the earthquake ID25080.

Figure 15. LST detrended and LST anomaly maps for the earthquake ID100966.

Figure 16. LST detrended and LST anomaly maps for the earthquake ID112123.

Figure 17. LST detrended and LST anomaly maps for the earthquake ID157456.

Figure 18. Mean LST anomalies with a searching radius of 50 km in time series for the 8 studied earthquakes 1 week before and after their occurrence.

Table 9. Comparison table of this study with some related works.

Study/Metric	Data source	Studied period	Studied location	Studied samples	Methods	Tools
This study	GOES/ABI MSG/SEVIRI Himawari-8/AHI	Between 2010 and 2021	Globally	More than 30,000 land earthquakes	- Data ingestion - Data filtering - Time series anomalies - Correlation study	- Big data analytics - GIS processing - Confusion matrix - ROC calculation
Boudriki Semlali et al. (2023)	FENGYUN-2F/VISSR	Between 2014 and 2022	Continental (Asia and western Europe)	More than 14,000 land earthquakes	- Data ingestion - Time series anomalies - Correlation study	- Big data analytics - Confusion matrix - ROC calculation
Boudriki Semlali et al. (2023)	GOES/ABI AQUA/MODIS	2020	Continental (Northern and Southern of Americas)	1350 land earthquakes	- Data ingestion - Time series anomalies - Correlation study	- Big data analytics - Confusion matrix - ROC calculation
Wu et al. (2012)	NOAA/VISSR FENGYUN-2F/VISSR NCEP-NCAR Reanalysis data	2008 2010	Regionally (China)	− 2008 M7.3 Yutian. - 2008 M8.0 Wenchuan - 2010 M7.1 Christchurch	- Data ingestion - Time series anomalies - Correlation study	- Statistical correlation - Image processing
Hassini and Belbachir (2013)	MSG/SEVIRI	2008	Locally (Algeria)	− 2008 M5.5 Oran city (North West of Algeria).	- Data ingestion - Time series anomalies - Correlation study	- Image processing
Jing et al. (2020)	AMSR-E/SSM/I	Between 2008 and 2018	Regionally (Sichuan province, China)	− 2008 M7.8 Wenchuan - 2013 M6.6 Lushan - 2017 M6.5 Jiuzhaigou	- Data ingestion - Time series anomalies - Correlation study	- Anomaly detection - Image processing
Pavlidou et al. (2018)	GOES/ABI MSG/SEVIRI Himawari-8/AHI	Between 2011 and 2016	Globally	20 land earthquakes	- Data ingestion - Time series anomalies - Correlation study	- Anomaly detection - Statistical correlation
Chen et al. (2020)	AQUA/MODIS TERRA/MODIS	2015	Locally (Nepal)	2015 M7.8 Nepal	- Data ingestion - Time series anomalies - Correlation study	- Anomaly detection - Statistical correlation
Akhoondzadeh (2013)	AQUA/MODIS	2012	Locally (Iran)	2012 M6.4 Iran	- Data processing - Time series	ARIMA, ANN, and SVM
Zhu et al. (2019)	MSG/SEVIRI	2017	Locally (Iran)	2017 M7.3 Iran	- Data processing - Time series	- Anomaly detection - Statistical correlation

Boudriki Semlali B-E, Molina C, Park H, Camps A. 2023. First results on the systematic search of land surface temperature anomalies as earthquakes precursors. Remote Sens. 15(4):1110. doi: 10.3390/rs15041110.

 Google Scholar

Wu L, Qin K, Liu S. 2012. GEOSS-Based Thermal Parameters Analysis for Earthquake Anomaly Recognition. Proc IEEE. 100(10):2891–2907. doi: 10.1109/JPROC.2012.2184789.

 Google Scholar

Hassini A, Belbachir AH. 2013. Hardware and software consideration to use near real time MSG-SEVIRI and NOAA-AVHRR images. 2013 11th International Symposium on Programming and Systems (ISPS). Algiers, Algeria: IEEE; p. 12–16. [accessed 2020 Apr 2]. doi: 10.1109/ISPS.2013.6581487.

 Google Scholar

Jing F, Singh RP, Cui Y, Sun K. 2020. Microwave brightness temperature characteristics of three strong earthquakes in Sichuan Province, China. IEEE J Sel Top Appl Earth Observ Remote Sens. 13:513–522. doi: 10.1109/JSTARS.2020.2968568.

 Web of Science ®Google Scholar

Pavlidou E, van der Meijde M, van der Werff H, Hecker C. 2018. Time series analysis of land surface temperatures in 20 earthquake cases worldwide. Remote Sens. 11(1):61. doi: 10.3390/rs11010061.

 Google Scholar

Chen C-H, Lin L-C, Yeh T-K, Wen S, Yu H, Yu C, Gao Y, Han P, Sun Y-Y, Liu J-Y, et al. 2020. Determination of epicenters before earthquakes utilizing far seismic and GNSS data: insights from ground vibrations. Remote Sensing. 12(19):3252. https://doi.org/10.3390/rs12193252.

 Google Scholar

Zhu C, Jiao Z, Shan X, Zhang G, Li Y. 2019. Land surface temperature variation following the 2017 Mw 7.3 Iran earthquake. Remote Sens. 11(20):2411. doi: 10.3390/rs11202411.

 Google Scholar

Data availability statement

The GOES/ABI, MSG/SEVIRI, and Himawari-8/AHI data used in this study comes from the Copernicus global land service https://land.copernicus.eu/global/products/lst and earthquake data comes from the open USGS earthquake database available at https://earthquake.usgs.gov/earthquakes/search/.