This preview shows page 1. Sign up to view the full content.
Unformatted text preview: cs, autocorrelation can be established as:
Trandom can be determined by mean and σ if it is a normal distribution.
α× t is a linear trend. A time series model provides a concise representation.
T(t) = T cycle (t) + T auto (t) + T random (t) + α × t
Time series models: * represent variability conceptually; * create the missing data; * study probability of certain scenarios; produce probabilisticforecasts; * generate scenarios of climate change and study its impacts. (3) Frequency‐domain methods
Frequency‐domain approaches transfer the data into series of periodic functions (e.g. sine and cosine), each with a characteristic amplitude and frequency. (3) Frequency‐domain methods
Frequency‐domain approaches transfer the data into series of periodic functions (e.g. sine and cosine), each with a characteristic amplitude and frequency.
The frequency of the functions represents the timescale of the cycle, the amplitude indicates how many variance is at that timescale. 4) Spatial analysis
The section of Mapping Considerations is not required.
(1) Spatial autocorrelation
One of the fundamental properties of climatological variables is their spatial coherence (i.e. how rapidly a variable changes with distance from a location). One way to examine the coherence is to estimate the correlation of a variable at one location with the one from another location. The spatial autocorrelation function (SAF) expresses the spatial correlation as a function of distance. How quickly a SAF decays indicates:
(a) the spatial scales of variability,
(b) whether spatial climate variability is resolved by the station network,
(c) how reliable spatial interpolation will be.
(d) the optimal timescale for performing spatial analysis.
The above four items can also be consider SAF’s application. Spatial data should have all non‐stationarities removed before spatial auticorrelation function are estimate. (3) Spatial interpolation All isoline maps produced from observational data require interpolation. In general, methods of spatial interpolation estimate variable values at grid points by using a combination of values at sample points. Different weighti...
View Full Document
- Winter '14