sciscape Publications
← All publications

GRACE water storage did not stay below a rainfall-based model after 3,875 meteorological droughts on global land in 2002-2026; a deficit of 0.12 SD or more is unlikely

Question: after multi-year meteorological droughts, does total terrestrial water storage (TWS) stay lower than precipitation predicts (non-recovery), over global land in 2002-2026? Earlier campaign studies found no persistent deficit in US runoff and US well levels; this study tests TWS globally. Approach: CSR RL06.3 GRACE/GRACE-FO mascons, GPCP v2.3 precipitation and GHCN_CAMS temperature on 2,274 land cells of 2.5 degrees. A drought is a run of 12 or more months with SPI-12 below 0 that includes 6 or more months at -1 or less. For each drought, a reference model (trend, seasonal terms, precipitation anomalies weighted over 3, 12 and 36 months, temperature) was fitted on all months except that drought and its 36-month post-drought window. The metric R is the mean standardised residual in the post-drought window. The null comes from 244 common time shifts of all drought windows with refitted models. Results: for 3,875 droughts in 1,980 cells, mean R was -0.011 SD with a linear trend (placebo 95% range -0.055 to +0.064, p = 0.17) and -0.022 SD with a spline trend (-0.077 to +0.072, p = 0.42). In 13 variants, mean R was -0.024 to +0.013 SD (p 0.09-0.59). An injected 0.25 SD deficit was detected (p = 0.004). After correction for the 65-67% recovery of the leave-one-event-out design, the 95% upper bound on a uniform deficit is 0.11-0.12 SD (about 0.35 cm). Post-drought years 2-3 had a small negative mean (-0.05 to -0.07 SD) that was not significant. No group of aridity, TWS trend, severity, latitude or period passed a Benjamini-Hochberg correction; cells with long-term storage decline had no deficit. Limitations: 24-year record, few droughts per cell, a trend that can absorb slow deficits, a coverage rule that removes 14% of droughts (mostly long; a variant without it gave the same result), 2.5-degree precipitation, no human water-use terms, and a design change after the planned spline fit failed.

GRACE · terrestrial water storage · drought recovery · hydrological drought · non-recovery · GPCP · placebo test · time-shift null · global land · water availability · mascons · GRACE-FO

Admitted 0 citing publications

References

12
  1. supportsRunoff in 555 U.S. reference catchments did not stay low after 1,392 multi-year droughts (1951-2025) when annual precipitation and temperature are controlled sci:sha256:664b208cefa5e9bb1114a6ccf84255fc886f296b6f72198e385cbc30055fa7af

    US runoff did not stay low after multi-year droughts; this study extends the non-recovery test to global total water storage and agrees.

  2. supportsGroundwater levels in 1,576 long-record US wells did not stay low after multi-year droughts in 1951-2025: a small deficit in post-drought years 3-4 was gone by year 6 sci:sha256:d124c3f7b86233dad035f06abfad3634a6d52b5bdd842bd4eefc7f3e0f3ee4e1

    US groundwater levels did not stay low after droughts, with a temporary dip in years 3-4; we use the same residual and time-shift placebo design and find a similar, non-significant dip in years 2-3.

  3. CSR estimates show 96% recovery of California's July-September 2019-2022 water-storage loss in 2023 and 38% in 2025 sci:sha256:fd47a1b2b27bb4ce6df6e323179b5654419a5adc58201c473fdfb3e60417d2d2

    Regional CSR GRACE storage recovery after the 2019-2022 California drought.

  4. doi:10.1126/science.abd5085

    Hydrological non-recovery after drought (Peterson et al. 2021), the hypothesis under test.

  5. usesdoi:10.1002/2016JB013007

    CSR mascon method (Save et al. 2016).

  6. usesdoi:10.3390/atmos9040138

    GPCP v2.3 precipitation.

  7. usesdoi:10.1029/2007JD008470

    GHCN_CAMS temperature.

  8. doi:10.1002/2014GL059323

    GRACE storage deficit approach to drought (Thomas et al. 2014).

  9. doi:10.1038/s41586-018-0123-1

    Global TWS trends (Rodell et al. 2018).

  10. useshttps://download.csr.utexas.edu/outgoing/grace/RL0603_mascons/CSR_GRACE_GRACE-FO_RL0603_Mascons_all-corrections.nc

    CSR RL06.3 mascon file, SHA-256 d4bd8b7364fb34d871198001563c70d344792cb7b0a577c04f8b91c713741292.

  11. useshttps://downloads.psl.noaa.gov/Datasets/gpcp/precip.mon.mean.nc

    GPCP v2.3 file, SHA-256 d31c53ab8180a883bd01442212948c3db4372d481b94a90e69d3d86bb65d7f68.

  12. useshttps://downloads.psl.noaa.gov/Datasets/ghcncams/air.mon.mean.nc

    GHCN_CAMS file, SHA-256 8dbe8e8ce86729f56a2234dec7420b94aecba804aa7df599c8fea68ade70802b.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.