Climate Exchange

[

] 154

isaster

isk

eduction

of these historical records, respectively. On Tuesdays

and Fridays, JMA predicts whether temperatures will

reach these criteria for each region and issues EWIEW

accordingly. If an expected criterion is not met, the

agency issues information to notify of a change in its

outlook and reports on the progress of temperature

conditions since the latest EWIEW.

JMA roughly estimated the skill of EWIEW based on

3,600 cases for all regions in Japan from March 2008

(when EWIEW was first issued) to January 2011. For

the categories of very high and low temperatures, the

hit ratios (that is, the number of occurrences of the

predicted phenomenon against the number of EWIEW

issuances) were 62 per cent and 57 per cent respectively.

While there is certainly room for improvement in these

rates, the information is well received and appreciated

by various user sectors.

In addition to the four elements of EWIEW listed

above, JMA also provides detailed probabilistic predic-

tion products for individual regions through the same

website independently of EWIEW issuance. These prod-

ucts show the scientific basis of EWIEW and provide

users with information on probabilities within the range

of their own threshold values. The products are:

• Daily tables of predicted T7d anomaly probabilities

in five categories (very high, high, normal, low, and

very low)

• Time sequences of these probabilities

• Cumulative probability functions (CPFs) and

probability density functions (PDFs) of predicted

T7d anomalies.

These data are provided not only for the 11

EWIEW regions but also for 152 stations where

JMA performed observations from 1981 to 2010.

The information is calculated using both numerical

prediction data from EPS and regression coefficients

estimated from hindcast experiments (re-forecasts)

conducted using the same EPS for the 30-year period

1981-2010. The skill levels of these products were

evaluated using hindcast experiment data and were

found to be satisfactory in terms of probabilistic

prediction metrics such as Brier skill scores. CPF

and PDF graphs are considered to have high potential

for possible future customized use of the scientific

basics used to produce EWIEW. For example, JMA

regularly provides the energy (electric power supply)

sector with information on the most likely predicted

T7d value together with more tailored information

covering the period one week ahead, which can be

estimated from its graph-form representation. Using

this newly available data, operators in the sector can

now estimate electric power demand for the coming

two weeks more efficiently and reasonably.

Use of EWIEW

EWIEW is expected to be used in a variety of fields

in addition to its current main user sectors of agricul-

ture and health. In one notable case, the information

toring and outlooks, and climate change projections. Until recently,

however, no climate information was supplied to specifically urge

the implementation of appropriate measures against high temper-

atures and other extreme events. Based on the increased skill of

seasonal forecasts and the above-mentioned social demand, JMA

therefore commenced the operational provision of Early Warning

Information on Extreme Weather (EWIEW) in March 2008 with

the aim of contributing to meteorological risk management in

climate-sensitive sectors. EWIEW is a probabilistic forecast high-

lighting the possibility of extreme events (such as very high or low

temperatures) from one to two weeks ahead, and is issued to 11

regions in Japan.

EWIEW is provided in a number of ways. JMA makes the

information widely available on its website

in map, table and

text format indicating current nationwide issuance, and its local

offices issue EWIEW in text form within their areas of respon-

sibility. In addition to information for use by the general public

on the website, EWIEW is also sent directly to national and

local governmental organizations, media providers and private

weather-related companies so that they can broadcast the warn-

ings themselves and/or prepare and issue information tailored to

users’ needs. For example, local governments may send messages

to farmers urging them to take measures against potential damage

from high temperatures.

Scientific basis

JMA operates the Ensemble Prediction System (EPS) for one-month

forecasts, using the Atmospheric Global Circulation Model, which

provides one-month numerical prediction data on temperature and

other conditions. EWIEW is based on predicted data for regional-

scale seven-day mean temperatures (T7d) in individual regions for

each day during periods of 5-14 days ahead. When a high probability

(30 per cent or more) of a very high or very low regional-scale T7d

anomaly is predicted for one or more regions of Japan, JMA issues

EWIEW including:

• Estimates of predicted temperatures meeting the criteria for

EWIEW issuance

• The relevant duration, region and a brief message on the possible

social impacts of such high/low temperatures

• The predicted probability of the expected condition occurring

• Details of expected general weather conditions considered likely

to induce extreme climate situations.

Criteria for announcing the probability of very high or low

temperatures in EWIEW are calculated for each region using

JMA’s historical records of regionally averaged observed T7d

values covering the 30-year period from 1981 to 2010. The crite-

ria for very high and very low are the 90th and 10th percentiles

Probability of predicted seven-day mean temperature anomaly for

five categories for 4-10 August 2010, issued on 30 July for the

Tohoku region (originally in Japanese)

Source: JMA