|
Cleaned and Enhanced Data
Although the national meteorological agencies in many countries have
compiled extensive records of historical weather data over several
decades or more, such data often is not suitable for weather trading
purposes in its raw form. To provide weather market participants with a
more reliable basis for pricing and risk management, RMS and EarthSat
produce two types of historical data for individual weather stations:
Cleaned Data
Cleaned data is a version of the official historical record in which
missing or erroneous individual daily values have been corrected.
Occasional missing values are not uncommon in most historical records,
and RMS and EarthSat also analyze the records to detect any erroneous
data, such as minimum values that are greater than maximums. All missing
or erroneous values are replaced with estimated values derived from
comparisons with neighboring station recordings and analyses of local
micro-climate biases. The final cleaned data provides a continuous and
complete historical time series of daily values.
Enhanced Data
Enhanced data is a version of daily historical values that has been
adjusted to be consistent with how temperatures are being recorded by
the current instrumentation at each individual weather station. Periodic
changes in weather station location, instrumentation, or environment
over time have introduced measurement discontinuities – permanent
increases or decreases in temperature observations – in the historical
records for many stations. The existence of such discontinuities in
historical data can make the data unreliable for valuing weather
derivatives that will be settled based on observations taken with
current instrumentation. An example of a measurement discontinuity is
shown below for the Charlotte Douglas International Airport weather
station in North Carolina. This graph of the difference in monthly
average temperature between Charlotte and a neighboring station shows a
significant cooling in temperatures at Charlotte in July, 1998,
coinciding with the commissioning of a new ASOS instrument package.
Over the past 3 years, RMS and EarthSat have developed and continued to
refine a complex methodology for identifying and quantifying
discontinuities in historical data. This data enhancement methodology
involves an extensive series of statistical tests that compare
historical temperature recordings at a particular weather station to
recordings at a series of highly-correlated neighboring stations. In
conjunction with original research into the histories of each weather
station, these tests reveal dates at which station changes or events in
the nearby environment have caused measurement discontinuities. Manual
analyses and checks of the data by meteorologists serve as a final step
to confirm the existence and magnitude of discontinuities in historical
data. The enhanced data is developed by adjusting all historical values
in the cleaned data prior to the dates of any confirmed discontinuities
to bring them into consistency with current and recent data.
RMS believes that enhanced data is critical for accurately interpreting
history in the modeling of risk for individual weather contracts. As a
simple example, a shift of 0.5 degrees in the daily measurement of
temperatures would cause an increase or decrease of 75 heating degree
days relative to historical experience for a standard 5 month November
to March winter contract. With typical contract structures involving
payouts of $5-10,000 for each additional degree day above or below a
specified strike level, the difference between using untreated
historical data and enhanced data for pricing and risk analysis can be
considerable.
RMS and EarthSat also conduct ongoing montoring of
weather stations to alert clients to recent and pending changes that may
cause new measurement discontinuities.
|
