More models

Species specific water requirements

Flow assessment

Generalised and process models

Colwells index

About Colwells index

Colwells index enables for the identification of temporal patterns in hydrological systems. It allows for the calculation of predictability, consistency and contingency across a data series.

Model purpose

The purpose of the model is to calculate the predictability, constancy and contingency of a data series. These values are calculated across the whole series, based on a number of input parameters. These values help identify temporal patterns in the data. All 3 predictors are 0-1 bound, with 1 being maximum and 0 minimum. The predictors describe the following:

  • Constancy – How consistent the state (flow, depth, etc) is for all times across all years (I.E. always very high flow).
  • Contingency – How consistent the state (flow, depth, etc) is at a given time (month, season) across the years (I.E. always very high flow in summer).
  • Predictability – A combination of constancy and contingency.

Development context

This model has been developed from the original literature, ‘Predictability, Constancy, and Contingency of Periodic Phenomena’, Robert K. Colwell 1974.

Spatial application

This model and its default parameters were created for application in South East Queensland.

However, the model parameters could be edited to suit other locations.

Model description

Ecohydrological rules

This model first determines the classes to bin data into, based on the class boundary method specified. The model then creates the time state matrix, by splitting the timeseries into time sections, taking the summary of each time and finding which bin this falls into. This matrix is produced in the result, along with the calculated predictors.

Assessment method

This model has no assessment due to the format of its results.


  • Daily time series data. This can be any type (flow, depth, rainfall, evaporation, etc.) provided it is at the daily or sub-daily time step.
Parameter Sections
  • Data – define the data type to run analysis on. All time series data types are valid options.
  • Classes – define the parameters required for binning of data. This includes the time period to split data into (seasonal or monthly), the summary method (mean, median, min or max), the number of classes (n), the class boundary method, the log base, the Gan from and the Gan step. The available class boundary methods are:
    • Equal – Splits the data into n equal sized bins, based on the data range.
    • Transform – Apply a log10 to the data, then apply equal bins.
    • Log – Generate bins on the logarithmic scale, using the base specified in the parameters.
    • Weighted Log – Generate bins on the logarithmic scale, multiplied by the data seasonal mean. Uses the base specified in the parameters.
    • Gan – Use the binning method proposed by Gan et al. (1991). Uses the Gan from and Gan step parameters provided.


  • Custom results table detailing the calculated predictors along with the state-time matrix.

User interface

Underlying code

This plugin is written in Python and its underlying code is publicly available from the Eco Risk Projector computation repository.


Colwell, R.K. 1974, Predictability, Constancy, and Contingency of Periodic Phenomena.

Gan, K.C., McMahon, T.A., and Finlayson, B.L. 1991. Analysis of periodicity in streamflow and rainfall data by Colwell’s indices. Journal of Hydrology 123(1-2): 105-18.