Florida Climate Institute - Florida State University Seasonal Hindcasts at 50km resolution (FISH50)

FISH50 is a global seasonal hindcast (or reforecast) for the period 1982-2008. The model is initiated in the first week of June and run for 6 months through end of November of the year for the summer FISH50. Similarly the winter FISH50 is initiated in the first week of December and run for 6 months through end of May of the following year.

The model used in FISH50 is the Florida Climate Institute-Florida State University Global Spectral Model (FGSM). A brief outline of the physics of the FGSM is provided in the table below.

Table 1: A brief outline of the physics of the FISH50 AGCM

It is run at a triangular spectral truncation of T248, which corresponds to approximately 50km grid resolution. It has 28 unevenly spaced terrain following  (=p/p_s) levels in the vertical with higher resolution in the boundary layer and in the upper troposphere.

FISH50 is shown to have high fidelity and its seasonal forecast skill is found to be comparable to some of the state-of-the art climate models run as part of the NMME (http://www.cpc.ncep.noaa.gov/products/ctb/nmme/) in many parts of the globe. FISH50 is not a coupled ocean-atmosphere model but a global atmospheric model forced with SST. It used the multi-model SST anomalies forecasted from two of the NMME models (CFSv2 and CCSM3) on a time varying observed SST climatology that has a stationary seasonal cycle, a secular change signal and a decadal component. The latter two components of the time varying climatology is updated at the beginning of each seasonal hindcast.

There are 6 ensemble members for each of the winter and summer seasonal hindcasts. The 6 ensemble members is generated by starting the hindcast from atmospheric initial conditions borrowed from the atmospheric reanalysis that are one day apart. The land initial conditions are kept identical for all ensemble members and they correspond to the reanalysis initial condition that corresponds to 01 June for summer and 01 December for winter FISH50.

For example, the first member of the summer FISH50 starts on 29 May, the second member starts on 30 May, the third member starts on 31 May……and the sixth member starts on 03 June. The details of the summer FISH50 is summarized below.

Table 2: Summer FISH50 Experiment Design

Similarly, the first member of the winter FISH50 starts on 28 November, the second member starts on 29 November ... and the sixth member starts on 03 December. All members however end their integration on the same day, 31 May for winter FISH50 and 30 November for summer FISH50. The details of the winter FISH50 is summarized below:

Table 3: Winter FISH50 Experiment Design

Although FISH50 is global, we are serving daily values of the following variables for the US domain shown in Fig. 1.

Table 4: Description of data served on FISH50

PLEASE NOTE: It would be incorrect to treat the FISH50 data as a continuous run. For example, do not use this data to do any applications for May-June or November-December. The winter and summer FISH50 are part of separate model runs. Likewise each of the ensemble members of the FISH50 is to be treated independent of each other. Watch this website for postings on peer reviewed publications on FISH50.

Fig. 1: The US domain of FISH50 over which data is being served. Shaded contours show the model topography (m).