Skip to content
- Annual homogenisation
- Homogenisation using annual data (averages or sums). May also include the magnitude of the seasonal cycle.
- ANOVA
- A joint correction method computing all corrections of a network simultaneously assuming all have the same regional climate signal and breaks are a step function. The method minimizes the noise using a least square method and as such the equations are the same as those of the statistical test for differences in means called ANalysis Of VAriance.
- Benchmarking
- Performance testing of homogenisation methods using realistic open data, which typically involves multiple methods, contributions or operators; it is a community effort. As such benchmarking is more than just a validation of a method.
- Corrections
- Adjustments made to raw observations aiming to make them more homogeneous. Together all corrections can also be called the estimated break signal in a multiple breakpoint setting. A corrected value is naturally not exactly right, in science we quantify how accurately we know a value.
- Daily homogenisation
- Homogenisation using daily data (averages or sums). In case of daily homogenisation the default is to consider all days serially in one long time series. Often the term homogenisation of daily data is used to imply that the distribution of the daily values is homogenised, not just the mean.
- Joint methods
- Homogenisation methods that jointly detection or correction all inhomogeneities in multiple stations in the same step.
- Monthly homogenisation
- Homogenisation using monthly data (averages or sums). This can be done with 12 monthly time series in parallel (default) or with all months considered serially in one long time series. Often in combination with annual homogenisation.
- Multiple breakpoint methods
- Homogenisation methods that detection or correction multiple inhomogeneities in one station/time series in the same step.
- Seasonal homogenisation
- Homogenisation using seasonal data (averages or sums). This can be done with 12 monthly time series in parallel (default) or with all months considered serially in one long time series. Often in combination with annual homogenisation.
- SNHT
- Standard Normal Homogeneity Test.