Introduction

Seasonal changes of various surface masses can be measured as time-variable gravity with satellites such as Gravity Recovery and Climate Experiment (GRACE) and its follow-on (FO), with a temporal and spatial resolution of 1 month and ~ 300 km, respectively. Such changes also drive seasonal crustal movements that can be measured using global navigation satellite system (GNSS) receivers. The GNSS network has a higher temporal resolution (typically daily) and has a high spatial density in tectonically active regions. GRACE and GNSS now play complementary roles in studying surface mass changes1,2.

Seasonal crustal motions and their link to hydrological mass changes (mainly soil moisture) have been studied in China3, North America4, and Europe5. Studies in Iceland6 and Japan7,8 showed that seasonal crustal movements are mainly driven by winter snowpacks. Zheng et al.9 reported a case of the Gulf of Carpentaria, northern Australia, where interference of the seasonal changes of ocean and hydrological masses results in a phase shift between the vertical and horizontal seasonal movements. Atmospheric loading also drives significant seasonal crustal movements. However, it has been deemed as a quantity that needs to be removed rather than a target of geophysical studies.

As a recent trend, timescales of the crustal deformation studies by surface loading are shifting from seasonal to intra-seasonal loading events. Such studies include crustal subsidence by stormwater loading associated with the landfall of tropical cyclones10,11, heavy rainfall12, and heavy snowfall13. Such studies will lead to a sophistication of hydrogeodesy, a newly developing field in geodesy that is deeply linked to various studies associated with climate change14.

Seasonal crustal movements have not been studied well in dry regions in Southwest Asia. There, hydrological mass changes are very little, but ocean mass changes are often significant. Wahr et al.15 found, using the GRACE data, seasonal ocean mass changes of ~ 30 cm peak-to-peak in the Red Sea, separating the Arabian Peninsula from the African Continent, driven by winds and the ocean-atmosphere exchange of heat. Later, Alothman et al.16 studied seasonal vertical movements of GNSS stations around the Red Sea and confirmed seasonal vertical crustal movements driven mainly by ocean mass changes. Here we perform a comprehensive study, using the GNSS and satellite gravimetry data, of seasonal crustal movements in and around the Arabian Peninsula caused not only by mass changes of the surrounding oceans, including the Red Sea and the Persian Gulf, but also by atmospheric loading.

Datasets utilized

Gravity recovery and climate experiment (GRACE) data

Figure 1 shows the monthly gravity data for 2002–2024 at three points representing the Persian Gulf (also called the Arabian Gulf), the inland region of the Arabian Peninsula, and the Red Sea. We used the mascon solution of GRACE (and its FO) from Univ. Texas, Center for Space Research (UTCSR). We fit the data using the quadratic function of time and the seasonal (annual and semiannual) changes. The inland point shows a significant gravity decrease possibly caused by groundwater depletion17 but little seasonal changes. On the other hand, points in the Persian Gulf and the Red Sea show significant seasonal changes with peak-to-peak amplitudes of ~ 15 cm and ~ 25 cm, respectively. They both show maxima in winter and minima in summer.

The winter mass increases in the Red Sea have already been studied by Wahr et al.15. Mass changes in the Persian Gulf have an amplitude ~ 60% as large as in the Red Sea and are driven by flow through the Strait of Hormuz18,19. If we only consider surface mass changes as measured by satellite gravimetry, we expect that winter subsidence has the largest amplitudes around the Red Sea. We also expect that they have amplitudes of ~ 60% around the Persian Gulf. On the other hand, in the inland region of the Arabian Peninsula, seasonal vertical movements would be negligible. Here we also note that the atmospheric mass changes have already been removed in the GRACE mascon solution. However, atmospheric loading would also cause seasonal crustal movements measured with GNSS.

Fig. 1
Fig. 1
Full size image

(a) We selected three points representing the Persian Gulf, inland, and the Red Sea. (b) Change in equivalent water thickness from GRACE and GRACE-FO mascon solution released from UTCSR. Vertical grid lines indicate year boundaries. Best-fit seasonal (annual and semiannual) changes together with secular changes expressed with quadratic functions are added. Oceanic points show strong seasonal signatures but little secular trends. On the other hand, the inland point shows little seasonal variations but a significant negative trend.

Global navigation satellite system (GNSS) data

We download daily coordinates of 25 GNSS stations with sufficiently long time spans in and around the Arabian Peninsula from the database run by Univ. Nevada Reno20. They provide solutions by precise point positioning without corrections of any non-tidal loading. We fit the vertical component of the position changes using the quadratic functions of time showing long-term changes and average seasonal (combination of annual and semiannual changes) terms. We considered the data up to 2024 May, but few stations continue observations until 2024. Figure 2a shows the positions of GNSS stations near the Persian Gulf (RASZ, BHR4), inland (SOLA), and near the Red Sea (RASH, ALWJ).

Figure 2b shows the time series of vertical positions of these five stations. The time periods shown there are different from station to station. Hence, we do not show labels in the horizontal axis of Fig. 2b. The data of all the other stations near the Persian Gulf and the Red Sea are shown in Figures S1 and S2, respectively. There, many stations show significant long-term subsidences, but they are out of the scope of this study.

Despite the different time windows, all the stations show clear winter subsidence of ~ 1 cm relative to summer, qualitatively consistent with the expectation. However, we did not anticipate such uniform amplitudes, i.e., the GRACE data (Fig. 1) predict that the stations on the Red Sea coast should show much larger winter subsidence than inland stations. Hence, the uniform amplitudes in Fig. 2 suggest that some other factor controls the winter subsidence of these stations.

Fig. 2
Fig. 2
Full size image

(a) Positions of GNSS stations near the Persian Gulf (RASZ, BHR4), inland (SOLA), and near the Red Sea (RASH, ALWJ). (b) De-trended up position time series of the five GNSS stations. Their time spans are different, i.e., 2009-2017.5 (RASZ), 2016.0-2024.5 (BHR4), 2008.0-2016.5 (SOLA), 2010-2018.5 (RASH), and 207.0-2015.5 (ALWJ). ALWJ has a jump in the middle of the time series of an unknown origin. Vertical grid lines indicate year boundaries. It is intentional that year labels are not given (they are different from station to station). (c) Vertical movements by non-tidal atmospheric loading. Best-fit seasonal (annual and semiannual) changes are added to those time series.

Non-tidal atmospheric loading (NTAL) data

Atmospheric pressure changes also drive crustal movements. An effective way to correct them is to subtract the non-tidal atmospheric loading21 (NTAL), available as a product from the Earth System Modeling Group at Deutsches GeoForschungsZentrum (ESMGFZ). They are calculated using the atmospheric surface pressure changes given at 0.5-degree grids every 3 h using the ECMWF (European Centre for Medium-range Weather Forecasts) operational data. Zheng et al.13 demonstrated that NTAL gives movements consistent with those calculated using atmospheric pressure data from a local network.

Figure 2c shows the vertical crustal movements due to NTAL changes at the five stations. They are in phase with the observed crustal movements. Their amplitudes are almost identical to the observation for the inland station. On the other hand, they are somewhat smaller at stations around the Persian Gulf and are significantly smaller at stations near the Red Sea. Such differences in amplitudes can be seen in the two-dimensional distributions of NTAL discussed later in Sect. 3.2.

Fig. 3
Fig. 3
Full size image

Average seasonal variations are drawn for two years for the observed (black/grey) vertical components and for NTAL (blue), for the GNSS points representing the Persian Gulf (right), inland (bottom), and the Red Sea (left). The observed and NTAL seasonal curves are shown so that their maximum values (occurring in July) coincide with each other to highlight differences in the winter subsidence. Grey curves (positions shown with grey circles in the map) represent stations not influenced by ocean mass changes in the Persian Gulf and the Red Sea. Differences of the black/grey and blue curves are given, together with their root mean squares, in Figure S3.

Discussion

Comparison between GNSS and NTAL

Figure 3 compares average seasonal (annual and semiannual components of the model curves in Figs. 2a, S1 and S2) vertical movements for all the stations studied here. To emphasize differences in winter subsidence, NTAL, and the observed movements are plotted so that their July positions coincide. At inland stations such as SOLA and YIBL, the NTAL and the GNSS amplitudes are very similar, suggesting that their seasonal movements come almost entirely from NTAL. The same is true for coastal stations facing the Indian Ocean like JASK, SQUO, and ASAB.

For stations on the Red Sea coast, NTAL amplitudes explain only ~ 1/2 of the observed changes. In contrast, observed subsidences of the Persian Gulf coast stations are only slightly larger than NTAL. These large and small differences between the observed and NTAL winter subsidences for the Red Sea and the Persian Gulf, respectively, correspond to the contributions from the ocean mass changes. Differences between the observed vertical seasonal movements and those from NTAL are given in Figure S3 together with their root-mean-squares.

Fig. 4
Fig. 4
Full size image

(a) Excess mass in January relative to July from the CSR GRACE mascon solution in terms of equivalent water thickness. (b) Crustal subsidence was calculated from the mass anomalies measured by GRACE using the Green’s function. Small white triangles and grey squares represent GNSS stations where significant and negligible winter subsidence occurs due to ocean mass changes, respectively. They are compared in (c) with the observed winter subsidence after the NTAL correction. Those on the Persian Gulf (PG) and Red Sea (RS) are marked with blue and red colors, respectively. The five stations given in Fig. 2 are indicated with larger circles. >1/2 of the stations fall within 1 mm from those calculated using the GRACE data.

Comparison between GNSS and GRACE datasets

Here we compare the excessive winter subsidence of GNSS stations inferred from Fig. 3 (Obs−NTAL), with those calculated using the excess mass from the GRACE data (mainly from the ocean because atmospheric mass contributions have already been corrected in the GRACE data). Figure 4a shows the average difference of mass in terms of equivalent water thickness (EWT) in January relative to July. The GRACE data (Fig. 1b) suggest an excess ocean mass of ~ 25 cm in the Red Sea and ~ 15 cm in the Persian Gulf. Figure 4b shows the vertical crustal movements as the elastic response of the solid earth calculated using the data in Fig. 4a and the load Green’s function22. The expected winter subsidence relative to summer is 3–5 mm around the Red Sea while it is ~ 2 mm around the Persian Gulf. Figure 4c compares the observed (Obs−NTAL) and calculated (Fig. 4b) winter subsidences for all the GNSS stations used here.

Figure 4c shows that the two quantities mostly coincide with each other. Despite the scatter, the majority of the data fall within 1 mm from the thick line. This would suggest the detection of the seasonal mass change of the Persian Gulf and provide support for the Red Sea mass change15,16. An interesting feature is that the nonuniformity in the NTAL seasonal amplitudes (getting smaller toward the southwest) (Fig. 5a, b) is compensated by significant ocean mass changes in the Red Sea. As a result, a fairly uniform winter subsidence of ~ 1 cm is realized in the studied region (Fig. 2).

Many of the GNSS stations used here are located in or near cities. If artificial groundwater extraction activities have seasonal changes, they may influence seasonal crustal movements. The scatter seen in Fig. 4c would partly come from such a groundwater contribution, together with miscellaneous factors like the deformation of geodetic monuments, vegetation changes, and so on.

Fig. 5
Fig. 5
Full size image

Surface vertical movements in January (a) and July (b) by atmospheric loading around the Arabian Peninsula, obtained as NTAL from GFZ. (c) Distribution of the average displacements by NTAL in January 2024 relative to July 2023. White dashed curves show ± 4 mm and ± 8 mm contours.

Figure 5c shows the worldwide distribution of winter-summer contrasts of displacements by NTAL. Lands around the Persian Gulf is one of the regions with the largest winter subsidence due to atmospheric loading, together with northern India, central Asia, China, and inland of Australia. On the other hand, atmospheric contributions are not so large in Europe, South America, Japan, Indonesia, and so on (non-seasonal short-term atmospheric contributions are large even in other regions). Removal of NTAL is deemed as a standard step in crustal deformation studies. Nevertheless, future research of seasonal crustal movements in those regions should pay extra attention to seasonal atmospheric pressure changes as the dominant factor.

Horizontal movements

As discussed in the previous session, seasonal vertical movements in and around the Arabian Peninsula can be understood in terms of seasonal change of loads, combination of the atmosphere and the ocean mass changes. Like vertical movements, seasonal horizontal movements are also considered to be driven by seasonal atmospheric and ocean mass changes. Figure S4a shows the January displacements relative to July by ocean (and partly hydrological) loads, and Figure S4b shows those by NTAL. The former is characterized with winter displacements of 1–2 mm toward the Red Sea. The latter is more complicated, but is characterized by the winter northward movements of the region south of 20 N by ~ 2 mm.

Such seasonal horizontal movements would not be easy to detect considering that both ocean and NTAL contributions do not exceed 1–2 mm. It is expected that we observe the sum of Figure S4a and S4b, but the observed displacements (Figure S5) are only partly consistent with the expected. Those in the southwestern part of the studied area show winter displacements toward NNE, which matches with the expectation. Around the Persian Gulf, however, their behaviors are rather systematic (e.g., winter displacement toward the gulf) and such large displacements are not expected by the winter excess mass in the Persian Gulf (Figure S4a). These misfits would reflect various unknown factors such as groundwater, deformation of geodetic monuments, vegetation around the antenna, and so on.

Conclusion

Geodesists often simply remove atmospheric loading by using products available on-line without paying much attention. This study shows that atmospheric loading plays by far the largest role in seasonal timescales in dry regions like the Arabian Peninsula. Hence, to isolate subtle crustal movement signals from hydrological and ocean loading in such regions, atmospheric loading would need to be correctly understood.

We conclude the essence of this study as follows.

  1. 1.

    Seasonal crustal movements in and around the Arabian Peninsula are dominated by atmospheric loading.

  2. 2.

    Around the Persian Gulf and the Red Sea, ocean loading enhances the winter subsidence.

  3. 3.

    This study gives the first detection of the Persian Gulf mass changes with GNSS.

  4. 4.

    Combination of the atmospheric and ocean loading realizes apparently uniform amplitudes of winter subsidence there.