 Express Letter
 Open access
 Published:
Inverse analysis of seismic activity rate changes for severely incomplete sequences: comparison of aftershock activity patterns immediately following the 2023 M6.5 and 2024 M7.6 Noto Peninsula earthquakes
Earth, Planets and Space volume 76, Article number: 112 (2024)
Abstract
We estimate the true aftershock activity rate from the detected initial aftershocks of the M7.6 Noto Peninsula earthquake on January 1, 2024, in addition to the preceding M6.5 Noto Peninsula earthquake on May 5, 2023. Estimating the initial aftershock activity rates has the serious problem of missing data, but in itself contains important clues to the physical structure associated with aftershocks and other external factors. We first estimate the detection rate models of the two aftershock sequences assuming the Gutenberg–Richter frequency laws, and we have found the contrasting time evolution of bvalues immediately after the mainshocks, in addition to the contrasting detection rate evolutions. Furthermore, by applying the compensated nonstationary ETAS (epidemictype aftershock sequence) model, we have revealed the contrasting time evolution of the background intensity rates, which characterizes the external factor of the original ETAS model besides the static triggering seismic activity. We discuss the implications of these differences of the analyzed outputs.
Graphical Abstract
1 Introduction
In the last century, there has been little seismic activity above M4 in and around the Noto Peninsula, except for the aftershocks of the 1993 M6.6 Noto Peninsula Offshore Earthquake (Ogata 2007, 2011). In 2018, seismic activity began to increase in the northern part of the peninsula, and by the end of 2020, microseisms (M ≥ 1) that had previously been observed in small numbers at shallow depths in the vicinity of Suzu City suddenly shifted to the depths of 14 km or more at the end of November 2020, causing intermittent swarm earthquakes. Then, with a delay of several months, they gradually spread clockwise to the surrounding area (Kumazawa and Ogata 2022a, b; 2023a, b). Later, on May 5, 2023, an M6.5 earthquake occurred in the northern region, which cascaded into the M7.6 earthquake.
This paper compares the statistical characteristics of the aftershocks of the M6.5 Noto Peninsula earthquake of May 5, 2023, and the M7.6 Noto Peninsula earthquake of January 1, 2024. To discuss these in detail, we need to avoid the difficulties for analyzing significant missing data immediately after large earthquakes. In other words, applying the ETAS model to aftershock sequences immediately following large earthquakes is inherently a poor fit because of the severe lack of aftershock data, resulting in biased estimates specific to the aftershock sequences. This is because aftershocks even with relatively large magnitudes are not detected (e.g., Enescu et al. 2009; Sawazaki and Enescu 2014). To avoid estimation bias, setting the lower magnitude limit of the aftershock data used at a high value increases the estimation uncertainty because the sample size becomes smaller. To overcome this difficulty, Ogata (1983, 2001) has conventionally treated a main shock and its immediately following aftershocks as a history, and excluded them from the target interval for fitting the ETAS model.
In this paper, however, all aftershock data from the main shock are fitted to obtain the characteristics of the aftershocks without any delay. For this purpose, we propose different methods from Omi et al. (2014), Zhuang et al. (2017), Morikawa et al. (2021) and Hainzl et al. (2024) to estimate the detection probability of aftershocks and to correctly compensate the ETAS model accordingly.
Furthermore, to test the significance of time variation by external effects such as significant pore pressures or slow slips in addition to the static triggers, the result of the nonstationary ETAS model is compared against the (stationary) ETAS model.
2 Methods
2.1 Magnitude frequency distribution model with missing measurements
In this paper, we assume that earthquake magnitude series always follow the Gutenberg–Richter (GR) frequency law (exponential distribution), even if they have different scales (i.e., bvalues). Large earthquakes are almost certainly detected in the earthquake catalog, while small earthquakes are almost always missed, and intermediate magnitudes are detected or missed by a certain percentage. Therefore, the detection rate of earthquakes is expressed as a cumulative function of the normal distribution of magnitudes, erf {・} (Fig. 1). That is to say, the parameter n represents the magnitude at which the detection rate is 50%, and s represents the range of the magnitudes where partial detection occurs. The GR distribution and the detection rate distribution are then multiplied together. Then, the shape of the density function equation, with the yaxis on a logarithmic scale, is very similar to the empirical histograms of earthquake catalogs, as seen for example in Ogata and Katsura (1993, 2006).
Specifically, the detection distribution density function for each earthquake of magnitude M_{t}, as a function of elapsed time t, is (see Fig. 1a)
From this, the detection rate of earthquakes with M ≥ M_{c} is (see Fig. 1b)
The parameters \(b_{t} ,\,\,\nu_{t} ,\,\,\sigma_{t}\) that depend on the time t are represented by cubic spline functions of equally spaced time intervals. This smoothing is determined by a tradeoff between minimizing the smoothness penalty, which is the integral of the weighted sum of squares of each of the first and secondorder derivatives of the spline function, and maximizing the loglikelihood function. The optimal method for determining the weights (hyperparameters) of the penalties is important and is obtained by minimizing the Akaike Bayesian Information Criterion (ABIC; Akaike 1980).
However, in this paper, in order to automatically relax the smoothing penalty in a short period immediately after a strong triggering effect of the ETAS model, a cubic spline function is defined on the equally spaced intervals in [0, N] where N is the total number of event data on [0, T], and the spline function of n = 1,2, …, N is transformed into the function of real time t_{n} over [0, T]. This provides a more effective representation than penalties applied in real time when earthquake events are densely clustered in real time, such as immediately after large earthquakes or bursting clusters of earthquake swarms.
2.2 ETAS models with compensation
Consider a stationary ETAS model (Ogata 1988),
which is applied to the aftershock data of magnitude M _{c} and greater, where \(H_{t} = \left\{ {(t_{i} ,M_{i} );\,\,\,t_{i} < t,M_{i} \ge M_{c} } \right\}\) is the history of detected aftershocks. The missing data are accounted for by the following compensated conditional intensity function:
using the detection probability function in (2). Then, the following loglikelihood function:
is maximized to obtain the maximum likelihood estimate (MLE) \(\hat{\theta } = (\hat{\mu },\hat{K},\hat{c},\hat{\alpha },\hat{p})\) of the original ETAS model in Eq. (3). The integration function in the second term of Eq. (5) is approximately calculated as a piecewise linear function.
In the nonstationary ETAS model (Kumazawa and Ogata 2013),
Timedependent functions \(\mu (t)\) and \(K(t)\) are assumed to be smooth piecewise linear functions connecting the respective values \(\mu_{i} = \mu (t_{i} )\) and \(K_{i} = K(t_{i} )\) corresponding to earthquake i. That is to say, the loglikelihood in Eq. (5) is adjusted with a smoothing penalty to obtain the log posterior distribution function:
Then, the ABIC is minimized by adjusting the smoothing weights w_{μ}, w_{K}, and the ETAS parameters \((c,\,\,\alpha ,p)\), to obtain the maximum a posteriori (MAP) estimation \(\hat{\theta }_{MAP}\) (see Appendix in Kumazawa and Ogata 2013, for details). Note here that the constraint penalties in Eq. (7) are imposed on the order of the sequence of earthquakes to perform a detailed inverse analysis of the changes immediately after the main shock and major aftershocks. Namely, when the occurrence times are highly clustered or when large discontinuities are emphasized, the penalty applied to the sequence time advancing by one between two consecutive events, as in Eq. (7), may be better than the penalty on the slope of the local linear function applied to the ordinary time (see Kumazawa and Ogata 2013, Appendix B). The superiority of this choice can be confirmed by comparing the ABIC values using a Bayesian method.
In the case of the maximum likelihood method, which maximizes the loglikelihood in (5), the goodnessoffit comparison can be made by calculating the AIC (Akaike 1974) from the loglikelihood in (5) of the compensated stationary ETAS model in (4). However, when this is compared with a nonstationary ETAS model, the AIC values cannot be directly compared with the ABIC value because their respective common abbreviated terms are incompatible.
Instead, we examine whether the MAP solution of the nonstationary Bayesian model in (7) with sufficiently large weights w and w_{K}, which make the functions m(t) and K(t) constant, is equivalent to the stationary ETAS model with the MLE. Then, we compare the ABIC value (= ABIC_{0}) of such a Bayesian model with the ABIC_{opt} values of the optimal weights to test the significance of the fit. That is to say, their difference, ABIC = ABIC_{opt} − ABIC_{0}, the confidence probability of the nonstationary model is exp(−ABIC/2) times large (Akaike 1985; Ogata 2017, see Appendix A; Ogata et al. 2013).
When the superiority of the compensated nonstationary ETAS model is confirmed, we then obtain the inversion solution, the maximum a posteriori (MAP) estimate. In this way, the compensated nonstationary ETAS model can estimate background rate fluctuations, verify the existence of externally induced components such as fluid pressure changes and slow slip, and detect their quantitative changes.
3 Results
3.1 Aftershock detection rates immediately following the M7.6 and M6.5 earthquakes
We use all detected aftershocks in the catalog detected and compiled by the Japan Meteorological Agency (JMA). Specifically, we examined all detected aftershocks up to January 12, 2024, including the M7.6 earthquake on January 1 (Fig. 2b), and also detected aftershocks from May 1, 2023 to 12 days after the M6.5 earthquake (Fig. 2a).
The inversion results for the magnitude–frequency model in (1) are shown in Fig. 3. First, from Fig. 3a, the bvalue of the aftershock activity for the M7.6 earthquake is constant. On the other hand, the change in σ values is significant, which may reflect the size of the aftershock area relative to the seismic station layout in the northern Noto Peninsula. Namely, there is a limited number of nearby stations in the interior, while the M7.6 aftershock zone extends to the distant coasts (Fig. 2b), where smaller events are better detected for the earthquakes occurring closer to the seismic station than for the more distant earthquakes. This is unlikely to be the case for the small aftershock area of M6.5 (Fig. 2a), so the σ value is constant throughout.
The difference in detection rate level between day and night is also adapted in the estimation of the 50% detection rate value (red curve) and also in the 97.5% detection rate (ν + 2σ value), represented by the green curve in Fig. 3b, d.
On the other hand, Fig. 3c shows that the σ value of the M6.5 aftershock activity is constant at 0.274 throughout the entire period, while the bvalue varies from 0.9 to 1.3. It decreases immediately after the M6.5 main shock, increases rapidly after the largest aftershock, and then converges to an almost constant bvalue around 0.9.
The blue curve in Fig. 3b, d shows the detection probabilities of M ≥ 2.0 for the M7.6 aftershock and M ≥ 1.0 for the M6.5 aftershock, respectively. These sensitively show decrease in the detection rates of secondary aftershocks immediately after a moderate or large aftershock in addition to some quasiperiodic fluctuations.
3.2 Analyzing aftershocks using compensated ETAS models
We calculate the earthquake occurrence rates with the compensated ETAS model in (4), which takes into account the effects of the detection probabilities shown in Fig. 3.
First, the aftershock activity immediately after the main shock of the M7.6 Noto Peninsula earthquake in January 2024 is analyzed. Equation (4), which multiplies the probability of detecting of an aftershock of M ≥ 2 shown in the green curve in Fig. 4a (the same as the blue curve in Fig. 3b) by the intensity function of the original stationary ETAS model in (3), defines the intensity function of the compensated ETAS model in (4) for the detected aftershocks, and its loglikelihood in (5) is maximized with respect to the original ETAS model parameters θ to obtain a maximum likelihood estimate (MLE) (Ogata 2007). The ETAS model for this MLE parameter value gives the true intensity function of the original ETAS model in (3), as indicated by the red curve of the intensity function in Fig. 4a. The occurrence of M ≥ 2 aftershocks by this ETAS model largely compensates for the missing data in the first few days and continues to compensate in detail for subsequent days. The μ value for this interval is constant (94.12 events per day).
We then applied the nonstationary ETAS model in (6) for the same aftershock sequence, with optimal smoothing by the ABIC method by adjusting penalty weights to the roughness of the time variation of μ and K values. Referring to Table 1a for each weight (hyperparameter), the larger the weights of μ and K values, the smaller the ABIC values. As the weights of the smoothing constraints increase, the maximum a posteriori (MAP) solution becomes eventually equivalent to the stationary ETAS model since both the μ and K curves become almost constant, and its MAP solution is the same as the MLE solution in Eq. (3).
Similarly, for the aftershocks of the M 6.5 earthquake, the intensity function (4) of the detected aftershocks is defined by multiplying the intensity function (6) of the original nonstationary ETAS model by the change in detection rate for aftershocks of M ≥ 1.0, and the minimum ABIC obtained by the Bayesian computation based on the loglikelihood function (7) with smoothing penalty is given in Table 1b. Based on this, the difference in ABIC between the model with the best weights \(\,(\hat{w}_{\mu } ,\hat{w}_{K} ) = (10^{4} ,10^{5} )\) and the model with the heaviest weights \(\,(\hat{w}_{\mu } ,\hat{w}_{K} ) = (10^{6} ,10^{13} )\) is −12.3; this difference is significant. In other words, the former model was exp{−ΔABIC/2} ≈ 468 times more probable than the latter, indicating that the compensated nonstationary model is far superior to the compensated stationary ETAS model. On the other hand, when the constraint of m value is further weakened (i.e.,\(\,\hat{w}_{\mu } \le 10^{3}\)) in Table 1, the performance becomes extremely poor.
In Fig. 4b, the optimal MAP solution of the original nonstationary ETAS is shown as a red curve, which appears to compensate for the apparent occurrence rate since the missing measurements of the first few days are substantially filled in and then the detection probability is close to 1.0 thereafter. The aftershock induction coefficient \(K(t)\) shows little variation. In particular, the background intensity rate μ(t) increases sharply by a factor of 10 in the first few days after the main shock and then decays.
4 Discussion
Hardebeck et al. (2024) review physically hypothetical mechanisms in detailed aftershock generation scheme, including research problems of dynamic aftershock triggering related to pore fluid pressure changes, to improve physicsbased aftershock forecasting.
Regarding the swarm activity in Noto Peninsula, there are a number of papers (Nakajima 2022; Kato 2024; Yoshida et al. 2023a, b; Nishimura et al. 2023) suggesting that fluid was supplied from the deep source, and this fluid slowly slipped through the seismicfree zone and penetrated the surrounding vulnerable areas.
In addition, Kato (2024) indicates that the early aftershock front of the M6.5 mainshock initially migrated at a speed of ∼20 km/hr and suggests that this rapid migration of the immediate aftershocks could be driven by upwelling of crustal fluids along the intensely fractured and permeable fault zone by the dynamic rupture of the mainshock.
Figure 4b shows the time evolution of the background intensity, which might be presumably due to strong fluctuations in fluid pore pressure caused by the M6.5 mainshock throughout the aftershock region (Kumazawa and Ogata 2023b). Similar migrations were observed for the space–time aftershock pattern of the M6 class earthquakes referred by Ogata (2007, 2011) and Kumazawa and Ogata (2024).
On the other hand, the spatiotemporal aftershock distribution of M7.6 is characterized by the timeinvariant, constant background intensity (Fig. 4a; Kumazawa and Ogata 2024). This indicates that there is no extension of the aftershock front as were seen in the first half examples of Ogata (2010) or Fig. 12 in Kumazawa and Ogata (2024). Such a pattern is typical of a sequence in which static stress changes due to the fault motion of the main shock dominate other factors, and is characterized by a very small constant background intensity, as seen in Fig. 4a,
From Figs. 3 and 4, we can see that the bvalue and μ value are constants for the M7.6 aftershocks. In addition, the change in bvalue and μ value for the M6.5 aftershocks appears similar in shape. However, we cannot definitively explain the reason why bvalue evolves significantly with time for M6.5 aftershocks while b remains constant for M7.6 aftershocks, but this may indicate the ratio of the pore pressure increase relative to the shear stress increase caused by the rupture. For the consistent examples, we can only refer to the two classic papers: first, Utsu (1962) showed that the magnitude distribution of aftershocks (bvalues) did not change with time for three aftershock series, including the 1957 Mw9.1 Aleutian earthquake; and second, in Healy et al. (1968), the Denver earthquake (1962–68), a wellknown waterinjectioninduced earthquake, showed that well pressure (which dominates the earthquake rate) and bvalue (0.6–0.9) are positively correlated.
5 Conclusions
The temporal variation of the bvalues considering the missing aftershocks of the M7.6 Noto Peninsula earthquake is constant, but the aftershocks of the M6.5 earthquake show a large variation. Furthermore, the compensated ETAS models of the respective aftershocks lead to different feature inversion results between the M7.6 and M6.5 earthquakes. For the former sequence, the compensated stationary ETAS model fits better, while for the latter, the compensated nonstationary ETAS model fits the latter well. The small constant in the μ value for the background intensity rate after the M7.6 event may indicate that the effect of pore pressure was significantly smaller than the shear stress increments caused by the main rupture. In contrast, the increase in the background intensity rate after the M6.5 event reflects the possibility that the effect of pore pressure became comparable to the shear stress increments.
Since early aftershock, forecasting plays an important role in emergency risk mitigation and immediate response efforts, stable procedures for realtime forecasting immediately after a mainshock need to be implemented.
Availability of data and materials
The Japan Meteorological Agency (JMA) unified catalog is available at the NIED website https://www.hinet.bosai.go.jp/about_data/?LANG=en for Registered Users. The authors last accessed February 2024.
Abbreviations
 AIC:

Akaike information criterion
 ABIC:

Akaike Bayesian information criterion
 ETAS model:

Epidemictype aftershock sequence model
 GR:

Gutenberg–Richter
 MAP :

Maximum a posteriori
 MLE:

Maximum likelihood estimate
 JMA:

Japan Meteorological Agency
References
Akaike H (1974) A new look at the statistical model identification. IEEE Trans Automat Contr 19(6):716–723. https://doi.org/10.1109/TAC.1974.1100705
Akaike H (1980) Likelihood and the Bayes procedure. Bayesian Statistics, J. M. Bernardo, M. H. DeGroot, D. V. Lindley and A. F. M. Smith, eds. Spain; University Press: 143–166, (discussion: 185–203) https://www.uv.es/bernardo/Valencia1.pdf; also in 'Selected Papers of Hirotugu Akaike' edited by Parzen E Tanabe K Kitagawa G, SpringerVerlag. ISBN10 : 0387983554; ISBN13:978 isbn13
Akaike H (1985) Prediction and entropy. in A Celebration of Statistics. A. C. Atkinson and E. Fienberg, eds., SpringerVerlag, New York, (1985) 1–24. https://doi.org/10.1007/97814613856081; also in 'Selected Papers of Hirotugu Akaike' edited by Parzen E Tanabe K Kitagawa G , SpringerVerlag. ISBN10 : 0387983554; ISBN13:978–0387983554
Enescu B, Hainzl S, BenZion Y (2009) Correlations of seismicity patterns in Southern California with surface heat flow data. Bull Seismol Soc Am 99(6):3114–3123. https://doi.org/10.1785/0120080038
Hainzl S, Kumazawa T, Ogata Y (2024) Aftershock forecast based on incomplete earthquake catalogs: ETASI model applications to the 2023 SE Turkiye earthquake sequence. Geophys J Int. https://doi.org/10.1093/gji/ggae006
Hardebeck JL, Llenos AL, Michael AL, Page MT, Schneider M, van der Elst NJ (2024) Aftershock forecasting. Ann Rev Earth Planet Sci. https://doi.org/10.1146/annurevearth040522102129
Healy JH, Rubey WW, Griggs DT, Raleigh CB (1968) The Denver Earthquakes: disposal of waste fluid by injection into a deep well has triggered earthquakes near Denver. Colorad Science 161(3848):1301–1310. https://doi.org/10.1126/science.161.3848.1301
Kato A (2024) Implications of faultvalve behavior from immediate aftershocks following the 2023 Mj6.5 earthquake beneath the Noto Peninsula, Central Japan. Geophys Res Lett. https://doi.org/10.1029/2023GL106444
Kumazawa T, Ogata Y (2022a) Changing features revealed from the nonstationary ETAS model in regional earthquake swarm activity of the Noto Peninsula, Report of the Coordinating Committee for Earthquake Prediction 107 (7–4). https://cais.gsi.go.jp/YOCHIREN/report/kaihou107/07_04.pdf
Kumazawa T, Ogata Y (2022b) Regional features revealed from the nonstationary ETAS model for earthquake swarm activity in Noto Peninsula. Report of the Coordinating Committee for Earthquake Prediction 108 (7–3). https://cais.gsi.go.jp/YOCHIREN/report/kaihou108/07_03.pdf
Kumazawa T, Ogata Y (2023a) Regional features revealed from the nonstationary ETAS model for earthquake swarm activity in Noto Peninsula  Continuation Report 2. Report of the Coordinating Committee for Earthquake Prediction, 109 (7–2). https://cais.gsi.go.jp/YOCHIREN/report/kaihou109/07_02.pdf
Kumazawa T, Ogata Y (2023b) Spacetime features of the Noto Peninsula swarm activity and the May 5, 2023 M6.5 earthquake. Coordinating Committee for Earthquake Prediction 110 (7–2). https://cais.gsi.go.jp/YOCHIREN/report/kaihou110/10_06.pdf
Kumazawa T, Ogata Y (2024) Spatial and temporal variations of the 3year earthquake swarm activities leading up to the M7.6 Noto Peninsula earthquake and interpretations of their activities. Research Memorandum 1228, The Institute of Statiscal Mathematics. https://www.ism.ac.jp/editsec/resmemoe.html
Kumazawa T, Ogata Y (2013) Quantitative description of induced seismic activity before and after the 2011 TohokuOki Earthquake by nonstationary ETAS models. J Geophys 118(12):6165–6182. https://doi.org/10.1002/2013JB010259
Morikawa K, Nagao H, Ito S, Terada Y, Sakai S, Hirata N (2021) Forecasting temporal variation of aftershocks immediately after a main shock using Gaussian process regression. Geophys J Int 226(2):1018–1035
Nakajima J (2022) Crustal structure beneath earthquake swarm in the Noto peninsula, Japan. Earth Planets Space. https://doi.org/10.1186/s4062302201719x
Nishimura T, Hiramatsu Y, Ohta Y (2023) Episodic transient deformation revealed by the analysis of multiple GNSS networks in the Noto Peninsula, central Japan. Sci Rep. https://doi.org/10.1038/s4159802335459z
Ogata Y (1983) Estimation of the parameters in the modified Omori formula for aftershock frequencies by the maximum likelihood procedure. J Phys Earth 31:115–124. https://doi.org/10.4294/jpe1952.31.115
Ogata Y (2001) Increased probability of large earthquakes near aftershock regions with relative quiescence. J Geophys Res 106:8729–8744. https://doi.org/10.1029/2000JB900400
Ogata Y, Katsura K (2006) Immediate and updated forecasting of aftershock hazard. Geophys Res Lett 33(10):L10305. https://doi.org/10.1029/2006GL025888
Ogata Y (2007) Anomalies of seismicity and crustal movement in and around the Noto Peninsula before the 2007 earthquake of M6.9 (in Japanese). Report of the Coordinating Committee for Earthquake Prediction 78 (7–11). https://cais.gsi.go.jp/YOCHIREN/report/kaihou78/07_11.pdf
Ogata Y (2010) Spacetime heterogeneity in aftershock activity. Geophys J Int. https://doi.org/10.1111/j.1365246X.2010.04542.x
Ogata Y (2011) Preseismic anomalies in seismicity and crustal deformation: case studies of the 2007 Noto Hanto earthquake of M6.9 and the 2007 Chuetsuoki earthquake of M6.8 after the 2004 Chuetsu earthquake of M6.8. Geophys J Int 186(1): 331–348. https://doi.org/10.1111/j.1365246X.2011.05033.x. Erratum, Geophys J Int 188(3): 1454. https://doi.org/10.1111/j.1365246X.2011.05306.x. http://onlinelibrary.wiley.com/doi/https://doi.org/10.1111/j.1365246X.2011.05306.x/abstract.
Ogata Y (2017) Statistics of earthquake activity: models and methods for earthquake predictability studies. Annu Rev Earth Planet Sci 45:497–527. https://doi.org/10.1146/annurevearth063016015918
Ogata Y, Katsura K (1993) Analysis of temporal and spatial heterogeneity of magnitude frequency distribution inferred from earthquake catalogues. Geophys J Int 113:727–738. https://doi.org/10.1111/j.1365246X.1993.tb04663.x
Ogata Y (1988) Statistical models for earthquake occurrences and residual analysis for point processes. J Am Stat Assoc 83(401):9–27. https://doi.org/10.1080/01621459.1988.10478560
Ogata Y, Katsura K, Falcone G, Nanjo KZ, Zhuang J (2013) Comprehensive and topical evaluations of earthquake forecasts in terms of number, time, space, and magnitude. Bull Seismol Soc Am 103(3):1692–1708. https://doi.org/10.1785/0120120063
Omi T, Ogata Y, Hirata Y, Aihara K (2014) Estimating the ETAS model from an early aftershock larger events. aftershock forecasts with ETAS sequence. Geophys Res Lett 41:850–857. https://doi.org/10.1002/2013GL058958
Sawazaki K, Enescu B (2014) Imaging the highfrequency energy radiation process of a main shock and its early aftershock sequence: the case of the 2008 IwateMiyagi Nairiku earthquake, Japan. J Geophys Res 119:4729–4746. https://doi.org/10.1002/2013JB010539
Utsu T (1962) On the nature of three Alaskan aftershock sequences of 1957 and 1958. Bull Seism Soc Am 52(2):279–297. https://doi.org/10.1785/BSSA0520020279
Yoshida K, Uno M, Matsuzawa T, Yukutake Y, Mukuhira Y, Sato H, Yoshida T (2023a) Upward earthquake swarm migration in the northeastern Noto Peninsula, Japan, initiated from a deep ringshaped cluster: possibility of fluid leakage from a hidden magma system. J Geophys Res. https://doi.org/10.1029/2022jb026047
Yoshida K, Uchida N, Matsumoto Y, Orimo M, Okada T, Hirahara S, Kimura S, Hino R (2023b) Updip fluid flow in the crust of the northeastern Noto Peninsula, Japan, triggered the 2023 MW 6.2 Suzu earthquake during swarm activity. Geophys Res Lett 50:e2023106023. https://doi.org/10.1029/2023GL106023
Zhuang J, Ogata Y, Wang T (2017) Data completeness of the Kumamoto earthquake sequence in the JMA catalog and its influence on the estimation of the ETAS parameters. Earth Planets Space. https://doi.org/10.1186/s4062301706146
Acknowledgements
The JMA PDE epicenter catalog as of January 30, 2024 was used for this analysis. YO and TK are supported by the “Seismological Research Innovation Promotion Program (STARE)” (Grant No. JPJ010217) of the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT). This research was stimulated by a joint work with Sebastian Hainzl. Comments by Aitaro Kato and anonymous reviewers are useful to clarify the manuscript.
Funding
YO and TK were supported by the “Seismological Research Innovation Promotion Program (STARE)” (Grant No. JPJ010217) of the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT).
Author information
Authors and Affiliations
Contributions
YO led and designed the entire research, and carried out the data analysis, interpreted the results, and drafted the manuscript. TK compiled the data for the analysis, YO drew all the figures. YO and TK discussed the results and commented on the manuscript. Both the authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ogata, Y., Kumazawa, T. Inverse analysis of seismic activity rate changes for severely incomplete sequences: comparison of aftershock activity patterns immediately following the 2023 M6.5 and 2024 M7.6 Noto Peninsula earthquakes. Earth Planets Space 76, 112 (2024). https://doi.org/10.1186/s4062302402057w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s4062302402057w