program nnic ! This program makes a consensus intensity forecast by combing model forecasts ! and variabales from the SHIP lsdiag.dat file using a 3-layer ! Neural Network ! v1.3.0 ! Modified Mar 2023 by MD ! 1. Version number to be consistent with nnic_module.f90 ! 2. Documentation in 3. below was added. ! 3. The nnic_module.f90 code was upgraded to allow for swapping HWFI ! input with HFAI when HFAI is available. This was needed to accomdate ! the migration from HWRF to HAFS planned for the summer of ! 2023. The swap is controlled by the ihswap parameter in ! nnic_module.f90. No changes to nnic.f90 or nnic_module.f90 ! will be needed when the switch happens. The program assumes that the ! HFAI forecasts will be added to the nn_input.dat file by the ! script that runs nnic. ! ! Modified Apr 2020 by MD and SS to use nnic_module.f90 to fill in missing ! HWFI forecasts using TCLP or the consensus tendency. ! NN baseline consensus forecast written to ATCF output file. ! Input: lsdiag.dat - SHIPS diagnostic file for a single forecast ! nn_input.dat - File containing the model intensity ! forecasts. Currently LGEM, DSHP, HWFI and AVNI are used. ! TCLP is also included as an option to fill in missing HWFI forecasts ! nnic_coefs.dat - NN coefficients and normalization facators ! Output: nnic.dat - nnic intensity forecast (ATCF format) ! nnic.txt - nnic intensity forecast (very old ATCF format) use nnic_module implicit none ! Open the input and output files call ioset write(lulog,200) 200 format('Start nnic forecast ') ! Read the lsdiag.dat input file call lread ! Specifiy the models to be included in the consensus call mselect ! Read in the model forecasts call mread ! Read the NN coefficient file call cread(atcfidl) ! Calculate the equally weighted consensus and deviation forecast call concal ! Calculate the neural net based unequally weighted forecast call nniccal ! Write the nnic forecast to the old ATCF format file call write_nnicoa ! Write the nnic forecast to the new ATCF format file call write_nnicna stop end program nnic