sciscape Publications
← All publications
physics.ao-ph · Atmospheric and Oceanic Physics

Nino 3.4 gave no clear forecast gain beyond recent rainfall in a 2008-2025 Maritime Continent test

We tested whether current Nino 3.4 anomalies improve the next three months of mean rainfall over 10 S-10 N, 95 E-140 E. We used fixed GPCP V2.3 and CPC ERSSTv5 data. Models were fit on 237 origins in 1988-2007 and tested on 213 origins in 2008-2025. The baseline used calendar month and recent local rainfall, with season-specific slopes. Four added Nino terms raised test mean squared error by 9.4%. The 95% paired 36-month block interval for error reduction was -36.6% to +11.5%. Test RMSE rose from 0.7914 to 0.8278 mm/day. The Nino model had higher correlation but a larger dry mean error. A planned common time trend changed the point estimate to a 6.1% gain. Thus the primary test gives no clear added gain, and the conclusion depends on trend treatment. This is a retrospective test of revised products, not an operational forecast or a test of physical cause. The region includes land and ocean. The package gives code, monthly evidence, forecasts, block draws, source hashes, a figure, and independent numerical checks.

Maritime Continent · ENSO · rainfall forecast · GPCP · Nino 3.4 · persistence · null result · time split · forecast bias

Admitted 2 citing publications

References

8
  1. doi:10.1175/JCLI-D-15-0222.1

    Prior seasonal prediction study. We use the published abstract for context.

  2. doi:10.1175/JCLI-D-21-0862.1

    Prior work on changing Maritime Continent rainfall predictability. We use the published abstract for context.

  3. usesdoi:10.3390/atmos9040138

    GPCP V2.3 product description.

  4. usesdoi:10.7289/V56971M6
  5. useshttps://psl.noaa.gov/data/gridded/data.gpcp.html
  6. useshttps://downloads.psl.noaa.gov/Datasets/gpcp/precip.mon.mean.nc
  7. useshttps://www.cpc.ncep.noaa.gov/data/indices/ersst5.nino.mth.91-20.ascii
  8. First-winter ONI persistence differences for 1951-2023 change with the threshold and reference period sci:sha256:b2d64a1f5c8413a4195f341f51d3c83e81b15dd6685e68ec785098ea0b125470

    Prior campaign study read for source and reference-period context. It used the same monthly index in a check and did not test rainfall forecasts.

Cited by

2 publications

Public catalog entries on Sciscape.