The following are explanations of the observed data and model data used for the interactive charts/future projections：

Observation data

Using the gridded daily observation data produced by TCCIP, the variables include mean temperature, maximum temperature, minimum temperature (unit: °C) and rainfall (unit: mm/day), the data spatial range is Taiwan including some offshore islands (119.3°E–122.25°E ; 21.8°N–25.75°N), the data time is 1960-2021, and the grid resolution is 0.05°×0.05°.

Model data

Referring to the global climate models used by IPCC AR6 (The Sixth Assessment Report of the Intergovernmental Panel on Climate Change), select models developed from 21 research centers, with a total of 31 sets of model simulation results, and see Table 1 for detailed introduction of the models. TCCIP project uses statistical downscaling method to produce AR6 statistical downscaling daily data, with variables including daily mean temperature, daily maximum temperature, daily minimum temperature (unit: °C) and daily rainfall (unit: mm/day). The data spatial range and grid resolution are the same as the observation data, and the data time is divided into historical simulation 1960-2014 and future projection 2015-2100. The number of models for temperature variables in each scenario are SSP1-2.6 with 25 models, SSP2-4.5 with 26 models, SSP3-7.0 with 23 models, SSP5-8.5 with 26 models, and the number of models for rainfall variables in each scenario are SSP1-2.6 with 28 models, SSP2-4.5 with 29 models, SSP3-7.0 with 27 models, SSP5-8.5 with 29 models.

Temperature future projection change (unit: °C), rainfall future projection change rate (unit: %), where the change or change rate refers to the degree of change of the model future projection relative to the baseline period (climate average of 1995-2014). Note: change = future projection value - baseline average value, change rate = (future projection value - baseline average value) / baseline average value * 100%

Table 1 Description of 31 models (Note: Not every model has simulated results for all scenarios, so the number of models for each scenario is different, ⚫ rainfall and temperature data are both available, ⚪ only rainfall data is available)