Page 79 - 全球气候变化及其影响Global Climate Change and Its Impacts-185×260

Chapter II Evidence for Global Climate Change

mote sensing, and reanalysis datasets. Meteorological station observations, with their long-
term time series records, can accurately reflect temperature changes at specific locations.
However, constrained by uneven spatial distribution of stations, there exists severe data
scarcity in remote areas such as the Antarctic continent and the heart of the Sahara Desert
in Africa. Satellite data demonstrate significant advantages in global coverage, enabling ac-
quisition of large-scale surface temperature information. Yet in practical applications, they
face challenges including limited sensor accuracy and complex atmospheric correction. For
instance, different satellite sensor models exhibit varying accuracy levels, potentially gener-
ating errors ranging from ±0.5°C to ±1°C when measuring surface temperatures. Reanalysis
data, synthesized from multi-source information through numerical model simulations, carry
potential biases due to inherent model uncertainties and excessive reliance on input data.
For example, when simulating polar region temperatures, models often produce significant
deviations due to inadequate representation of unique atmospheric circulation patterns and
underlying surface conditions in these areas.
To verify the reliability of these data, researchers employed a cross-validation method.
They conducted meticulous comparisons between meteorological station observation data
and satellite-retrieved surface temperature data. For instance, in the topographically complex
Himalayan region, comparative analysis was performed using data from multiple weather
stations and corresponding satellite-derived temperature data. The results revealed significant
discrepancies between the two datasets in certain areas. Through thorough investigation, it
was discovered that satellite data in these regions were affected by terrain-induced atmo-
spheric correction errors, leading to overestimated retrieved temperatures with maximum
deviation reaching 2°C. By calibrating with ground-based measurements, researchers es-
tablished correction models using field data to successfully rectify satellite data biases. For
reanalysis data, researchers conducted comparative validation with independent high-quality
meteorological station observations. This revealed deviations between reanalysis data and ac-
tual observations during extreme climate events like the 1998 El Niño phenomenon. The root
cause was identified as numerical models’ insufficiently accurate physical descriptions of
extreme weather processes, failing to precisely capture subtle changes in atmosphere-ocean
interactions during extreme events.
When assessing the representativeness of climate change evidence, researchers con-
ducted spatial distribution evaluations of meteorological station data. The study revealed
that in some African regions like the Central African Republic, meteorological stations were
sparsely distributed, with only 1-2 stationsstations per 1000 km², which were insufficient to
accurately reflect the overall temperature variation characteristics of the region. To address
this issue, the research team established temporary observation stations and utilized satellite
remote sensing data for spatial interpolation. This approach improved the spatial resolution
of temperature monitoring in the region from the originalhundred-kilometer scalelevel from
the original hundred-kilometer level to a ten-kilometer level, improving the spatial represen-

• 71 •