jma_logo Welcome to TCC wmo_logo
Home
World Climate
Climate System Monitoring
El Niño Monitoring
NWP Model Prediction
Global Warming
Climate in Japan
Training Module
Press release
Links
HOME > World Climate > Impacts of Tropical SST Variability on the Global Climate > Data and Analysis Method

Data and Analysis Method

1. Data

(a) Temperatures and precipitation

The analysis was performed using monthly mean temperature information and precipitation totals from CLIMAT data on states around the world as well as the Global Historical Climatology Network (GHCN) data provided by the National Oceanic and Atmospheric Administration (NOAA)'s National Centers for Environmental Information (Peterson and Vose, 1997). Both sets of data are based on observations conducted at surface weather stations worldwide. CLIMAT data are available only as far back as June 1982. Where both were available, CLIMAT data were used.

(b) Sea surface temperature

The COBE-SST analysis dataset (Ishii et al., 2005) produced by the Japan Meteorological Agency (JMA) was used as the sea surface temperature (SST) data source. Details of this dataset are provide in The Characteristics of the Global Sea Surface Temperature Data (COBE-SST)/Monthly Report on Climate System Separated Volume No.12.

(c) Other supporting data – reanalysis and satellite observation

To support and complement the results from surface observation data analysis, 2-m temperature and total precipitation rate data from the Japanese 55-year reanalysis (JRA-55; Kobayashi et al., 2015) were used. Ooutgoing longwave radiation (OLR) data derived from polar-orbiting satellite observations conducted by NOAA were also used as estimates of convective activity.



2. Analysis Method

The period of the analysis was from 1958 through 2012 (55 years). The analysis method is described below.

(a) Indices of tropical SST variability

Indices relating to tropical SST variability in three areas were defined using the area-averaged SST anomaly normalized by its standard deviation, where the anomalies are the deviations from a sliding 30-year mean for NINO.3 and linear extrapolations with respect to a sliding 30-year period for NINO.WEST and IOBW. These indices are available on the "Download El Niño Monitoring Indices" page.

Area Latitude Longitude

Eastern equatorial Pacific (NINO.3)

5°N–5°S

150°W–90°W

Western tropical Pacific (NINO.WEST)

15°N–EQ

130°E–150°E

Tropical Indian Ocean (IOBW)

20°N–20°S

40°E–100°E


From these indices, warmer (cooler) SST events were identified. El Niño (La Niña) events, in which the SST in NINO.3 is above (below) normal, were identified when the five-month running mean value of the NINO.3 SST index was 0.5°C or above (-0.5°C or below) for at least six consecutive months. Indices relating to warmer (cooler) SST events in NINO.WEST or IOBW were identified in the same way, but the thresholds were 0.15°C for warmer events and -0.15°C for cooler events.

(b) 5° x 5°- grid data on temperature and precipitation

Three-month running mean temperature and precipitation totals for each station were derived from CLIMAT or GHCN data. Three-month averages and totals were produced only if all relevant data for three consecutive months were available. Then, for each station, the temperature anomaly normalized by its standard deviation and the ratio of precipitation to its mean values for the analysis period were calculated if only data availability was 50% or more for the analysis period. Additionally, for levels of data availability between 50% and 80% where data are biased on either side of El Niño/La Niña events (or warmer/cooler SST events in NINO.WEST or IOBW), the station data are not retained. Subsequently, temperature anomalies and precipitation ratios were averaged for each three-month period into 5° x 5° grids with data for all stations located in a grid. Here, the long-term linear trend of temperature was eliminated to exclude impacts from global warming.

(c) Thresholds of classification

For each girds and three-month period, the temperature (precipitation) data were clasified as "Low (Dry)," "Normal" and "High (Wet)" using thresholds based on the assumption of equal probabilities for each class (known as as "climatological probability"). For most grids, thresholds were successfully set to give a 33% climatological probability for each class, but it should be noted that climatological probability far exceeds this value in some cases. By way of example, the climatological appearance probability of "Dry" is far above 33% in desert areas where it aseldom rains, and the threshold for "Dry" and "Normal" in such area were based on the smallest non-zero smallest value in the data.

(d) Statistical testing

Statistically testing was performed to determine whether the appearance probabilities of each class exceeded the climatological probability for El Niño and La Niña events (or warmer/cooler SST events in NINO.WEST or IOBW). Specifically, the population proportion was assessed via binomial testing with 90 and 95% statistical significance levels. The results are shown on the "Detailed Charts" page.

(e) Schematic charts

Based on the result of the above statistical testing, schematic charts were produced. When three or more adjacent grids had equal tendencies with a confidence level of 90% or more, the area was designated a specific color. In determining these colored areas (in particular for those where observation stations are sparsely distributed, such as over oceans), composite analysis results for 2-m temperature, SST and OLR were also taken into account.


References


Tokyo Climate Center, Climate Prediction Division, 1-3-4 Otemachi, Chiyoda-ku, Tokyo, Japan.
Copyright(C) 2002. Japan Meteorological Agency. All rights reserved. - Legal Notice -