Page 71 - 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

of the Earth’s surface and atmosphere. For instance, thermal infrared sensors can monitor
global surface temperature distribution, providing crucial data for studying global tempera-
ture changes; microwave sensors can detect precipitation, improving spatial coverage of
precipitation observations to some extent. The temporal resolution of satellite data is also
continuously improving, with some satellites achieving multiple daily observations. Howev-
er, satellite data is affected by sensor performance and atmospheric interference, leading to
issues such as radiometric calibration errors and atmospheric correction errors, making pre-
cisemeasurements ofcertain meteorological elements remain challenging, such as accurate
detection of cloud internal structures and specific trace gas components.
Reanalysis data is a dataset generated by integrating various observational data through
numerical models and assimilation techniques. It combines weather station observations,
satellite data, and other observational materials, featuring complete spatiotemporal coverage
and a unified data format, facilitating long-term climate analysis. Reanalysis data employs
model simulations and data assimilation processes to provide comprehensive estimations of
climate system states. However, due to uncertainties in numerical models, limitations of as-
similation algorithms, and quality variations in input observational data, reanalysis data may
exhibit systematic biases. Significant discrepancies between simulation results and actual
conditions may occur in complex terrain areas or under extreme weather conditions.
To achieve effective fusion of multi-source climate data, multiple technical approach-
es must be employed. Data interpolation is one of the common methods. For processing
weather station observational data with uneven spatial distribution,inverse distanceweighted
interpolation and Kriging interpolation can be applied.inverse distanceThe inverse distance
weighting interpolation method follows the principle that data closer to observation points
receive greater weights, performing weighted averaging of known station data to estimate
climate values at unobserved locations; Kriging interpolation considers the spatial autocor-
relation of data throughconstructing semi-variograms to more accurately estimate values at
unknown points, filling gaps in meteorological station spatial distribution. When integrating
satellite data with ground observation data, regression analysis can be used to establish math-
ematical relationships between them. For instance, by conducting regression analysis be-
tween ground-observed air temperature data and satellite-retrieved atmospheric temperature
data, models can be developed to convert satellite data into forms comparable with ground
observations, thereby achieving data fusion.
Data assimilation techniques play a central role in multi-source climate data fusion.
They organically integrate observational data with numerical models, continuously adjusting
the model’s initial conditions and parameters to make simulation results align more close-
ly with real observations. In marine climate data fusion, assimilation methods such as the
ensemble Kalman filter are commonly employed to assimilate ocean buoy observations of
temperature, salinity, current data, and satellite altimeter measurements of sea level data with
ocean numerical models. This effectively enhances simulation accuracy of ocean circulation,

• 63 •