************************************************************************ LOG OF ETA MODEL TEST RUNS ************************************************************************ 1992 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 13/12Z: ETA archive changed to 17-lyr, generalized (pressure) Betts, and drag table over elevated water points. ETAY archive has 16-lyr, old Betts, no drag table. 28/00Z: ETAX archive changed to 7 days, ETAY archive to 35 days. 28/12Z: Fedor's experimental Betts profiles put into ETAY. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 17/12Z: Integrated Richardson number in PROFQ2 put into ETAY; at same time, viscous sublayer removed and no drag table over lakes. 21/00Z: The ETAX pcp is now being put into the ET8P 35-day precipita- tion archive instead of the ETAY pcp. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 02/12Z: The ETA, ETAY, and ETA30 runs now include: (1) Fedor's new convection profiles with unified sea/land; (2) corrected AQ1, BQ1, and BQ2 in RADFS. The ETA and ETAY runs also now include Great Lake temperatures derived from the 14km/50km satellite data (Bill Gemmill's routine). ETAY switched to newest Betts with quadratic temperature reference profiles and subcloud outflow. 04/12Z: A typo was removed from the ETAY code dealing with the quadratic reference profiles. 11/00Z: ETA, ETAX, and ETAY resumed after Cray disk crash caused all 80 km runs to be missed at 12Z the 10th. 13/00Z: Bug removed in 30 km operational version of RSTINI (incorrect SLP on western boundary) thereby eliminating the large gravity wave that was being generated in each forecast. None of the rerun versions with tops at 100 mb or 80 km operational versions of RSTINI had the bug. 17/00Z: 30 km operational run was restarted improperly after prev. maintenance and was not completed. 18/12Z: ETAY turned off terminating test from 12Z March 4. 22/12Z: ETAY is now 80 km with 46 layers (concentrated in lowest elevations) and 200 iterations in vertical diffusion loops of all variables except Q2. Surface exchange coefficients derived through integrated Richardson number with no viscous sublayer. 25/00Z: A rotating 31-day (square) tape archive was begun for the 30 km runs. Included are 3-d fields of omega, T, u, v, and q as well as surface pressure, surface geopotential, accum- ulated and convective precipitation, sea level pressure, skin temperature, soil moisture, and lowest layer u and v. 26/00Z: ETAY continues as above except depths of the soil slab and of ice are changed from 0.100m to 0.016m to account for the thinner layers over the surface. 27/00Z: Sensible and latent heat flux were added to the 31-day 30 km tape archive. 30/12Z: The ETAY from 12Z 22 March ended with this run. 31/00Z: The 30 km run was lost again when the job was restarted after being suspended for Cray preventive maintenance. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 03/12Z: ETAY turned on to test single point mountains. 11/00Z: The current ETAY test ended prematurely with this forecast due to Cray procedures destroying /tmp/eta. 13/12Z: ETA runs resumed; last successful run was 12Z April 11. ETA30 runs resumed; last successful run was 00Z April 11. ETAY test begins using explicit cloud water in 80km/17lyr version. 22/00Z: Latent heat added to 30 km tape archive. 23/00Z: The 30km forecast failed again due to a faulty restart by the Cray after the job had been suspended for preventive maintenance. 28/12Z: ETAX changed from 16 layers to 17 layers with 50 mb top. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 04/12Z: ETAY switched from cloud water (13 Apr) to 2nd test of single point mountains (this time with steps along coasts also single point). 18/12Z: (MEB) ETAY switched from single point mountains to 2nd test of Lech Lobocki's PROFQ2 code (integrated Richardson number, no viscous sublayer, no drag table over lakes). ETA and ETAY were modified to include Joe Gerrity's implicit vertical diffusion in PROFS and PROFQ2. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 01/12Z: ETAY switched from test beginning 12Z 18 May to what was the standard ETA run before the inclusion of implicit vertical diffusion in order to see the impact of the change in diffusion. 04/00Z: The 30km run killed because it was restarted from zero around midday. 08/12Z: ETAY switched from test beginning 12Z 1 June to test of modification of turbulent kinetic energy computation conceived by Tom Black and Joe Gerrity. 19/00Z: The 30km run was turned off. 22/12Z: The ETAY test from 8 June turned off. 23/00Z: The MESO version (40km/38lyrs) begins. This is the optimized code with J. Gerrity's vertical implicit diffusion and TKE production/dissipation (without Helfand), and updated GFDL/MRF radiation. 24/12Z: The updated and optimized version placed into ETAY. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 02/00Z: Horizontal smoothing of initial surface potential temperature was removed in MESO. 09/00Z: Bug removed in RADFS in the updated fast code which was resulting in improper description of the stratiform cloud effects; thus the change will occur in the MESO and ETAY runs. 13/00Z: ETAY test from June 24 was turned off, ending with this run. 13/12Z: Mysterious missing ETA and ETAY runs. The IJP job that initiates the ETA run never was executed after the NGM run. 14/00Z: ETAY was switched to Fedor's average L scheme in PROFQ2. The "slow" code was used. 16/00Z: Missing COMMON block in CUCNVC was added to the fast code which provides RADTN with the convective precip rate. This allows for the inclusion of convective clouds in radiation. This change effects the MESO run. 21/00Z: INIT was modified in the fast code to initialize the saturation specific humidity (QS) over water. This removed an error in which the surface moisture exchange at timestep 1 had felt completely "dry" water. Because this is the fast code, only MESO is changed. 22/12Z: The ETA run is replaced with the "standard" fast code which until now has only been in the MESO run. The old slow code will be kept in /wd2/wd22/wd22tb/opnl in file fcstdrag_implicit. The ETAY test from 14 July is terminated and is replaced by the testing of the current fast code with the MRF's large scale rain algorithm. 28/00Z: RADTN was modified in the fast code to spread the stratiform clouds into adjacent layers where the relative humidity is high enough to allow stratiform cloud. Also, for a 30 mb layer above the ocean surface, the critical RH for stratiform cloud was increased from 0.8 to 0.95 (CLDRH2). Stratiform cloud computation will not take place over points where convective clouds will occur. These changes were made in order to reduce the problem with light, large-scale rain over ocean points that existed in the fast code. This will modify the MESO, ETA, and ETAY runs. 31/00Z: The value of FSL in CUCNVC in MESO is changed from 0.90 to 0.85. This will make the moist profiles moister and thus should generally decrease the heavy convective rain. 31/12Z: The ETAY test from 22 July was terminated and replaced by testing the current fast code with MRF's large scale rain algorithm along with Joe Gerrity's scheme for upward transport of condensed water above the mixed layer, beginning with the 7/31 12Z run. A bug in RADTN in the fast code was removed that caused the stratiform cloud spreading algorithm (see 7/28) to fail to thicken some stratiform clouds. This modification will affect the ETA, ETAY, and MESO runs. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 7/00Z: The MESO archive jobs on the front end were changed to remove a problem with packing the precipitation forecast fields. All MESO forecasts will now use the updated packer/unpacker found in 'NWS.WD22.TLB.ETA.SOURCE40(WXAIEME)'. 10/12Z: The lbulk scheme is replaced with the integrated Richardson number in PROFQ2. Also, the mixing lengths at each inter- face are now computed taking into account the different layer thicknesses above and below the interfaces. These changes are placed in the standard code affecting all runs. 13/12Z: The ETAY test of 31 July is terminated and replaced with a test of the MRF's use of an inversion check in the lowest 100mb over sea points; if the inversion falls within this layer and is dry above, then clouds are allowed within the layer with an RH cutoff of 90%, otherwise no clouds are allowed below 100mb over sea. 31/12Z: The 12Z forecasts of MESO are now turned on; they will run only to 24 hours (the 00Z MESO still runs to 36 hours). ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 02/12Z: The ETAY test of 13 August is terminated and replaced with a test of the standard code with 38 layers and 80km resolution. 09/12Z: The ETAY test of 2 September is suspended and replaced with a test of the 80km/17lyr model with convection modified as follows: (1) deep convection adjusts to cloud bottom and not LMH-1; (2) the slope of the first guess reference temperature profile in deep convection changes at the ambient freezing level, not where the reference drops below freezing (check in 420 loop is 'IF(T(K,IVI+1).LT.T0) GO TO 430'); (3) cloud top is not automatically lowered one level (check in 350 loop is 'IF(TREF(K,L).GT.T(K,L)-DTTOP) LTOP(K)=L') and DTTOP is changed from 1.0 to 0.0; 11/12Z: The 12Z MESO run was given the same CUCNVC changes as are in the ETAY. The 00Z MESO run still uses the standard code. 24/12Z: The changes in CUCNVC tested in ETAY from 12Z 9 Sep are put into the standard code. In conjunction, FSS is changed to 0.8 in all versions. PROFQ2 is changed in the standard code to include the viscous sublayer. These changes thus apply to all ETAL and MESO. The ETAY test of 9 September is terminated and replaced with the 80km/38lyr version of the standard code. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 02/00Z: The ETAL AND ETAY were lost because they ran so late that the 12-h old AVN files for the boundary data were gone. 05/12Z: New PBLOUT creates data for charts of surface sensible and latent heat fluxes, vertical gradients of heat and moisture at the ground, and friction velocity. ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 18/12Z: The ETAY test from 24 September was terminated and replaced with the new 80km/38lyr version which includes: (a) the expanded horizontal domain reaching slightly farther north, east, and south but significantly farther west; (b) a slight modification of the vertical structure so that the secondary maximum in resolution is near 250mb; (c) COMMON blocks are now in INCLUDE statements; (d) the second TTBL which is used only at pressures over 700mb and which has higher resolution within the table. 30/12Z: The ETAY version was turned off after moving it to the standard late slot. ----------------------------------------------------------------------- ----------------------------------------------------------------------- 1993 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 06/12Z: A test of doing deep convective adjustment down to the second lowest layer above ground was placed in ETAY. 25/12Z: The ETAY test commencing 6 Jan was terminated. 26/12Z: The ETAY began testing second-order horizontal diffusion (SECOND=.TRUE. in HDIFF). It was discovered that the ETAL and ETAY disk archiving has captured only the first 8581 points (the number of points on the old 80 km grid) since the new expanded grid was introduced on 30 Dec 1992. The problem has now been rectified. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 12/12Z: The ETAY test commencing 26 Jan was terminated and replaced with a test of Joe Gerrity's latest changes in SLINT, PROFQ2, and PROFS. In SLINT, the lowest layer winds around each H point are used to compute the wind there. PROFQ2 and PROFS include the newest changes to the TKE prod/dissp equation (a la Galperin) as well as performing the vertical diffusion of Q2 before the prod/dissip calculation is carried out. Also, in calculating the windshear at H points for use in th Richardson number, only those winds in the given eta layer that are above ground are used. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 01/12Z: A bug was removed in the code which averages only the winds which are above ground in PROFQ2 for computing the Richardson number (an index for VTM was KHN rather than K+KHN). This is part of the 12 Feb ETAY run as well as the MESO run. 03/12Z: The ETAY from 12 Feb is terminated. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 01/12Z: Penman-Monteith placed in ETAY. 06/00Z: ETAY now has the same surface files as ETAL and MESO. 06/12Z: The ETAY 36-day disk archive begins. 09/12Z: The fourth-order diffusion was changed to second-order as it was in ETAL starting with this run. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 12/12Z: The initial soil moisture was changed so that use of the NGM's moisture availability is replaced by the GDAS first guess soil moisture. The Penman-Monteith code remains unchanged. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 03/12Z: The ETAY test of 12 May is finished and replaced with a test of the 80km/38lyr model with convection modified as follows: (1) original cloud top is the first buoyant layer above cloud base where the layer directly above it is negatively buoyant (rather than the highest buoyant layer) (2) depth of cloud base search (PBM) changed to 150 mb (was 130 mb) (3) depth of shallowest cloud (PNO) is changed to 25 mb (was 90 mb) (4) transition from shallow to deep clouds (PSH) is changed to 300 mb (was 290 mb) (5) maximum allowable shallow cloud top is changed to 300 mb (PSH) above cloud base (from no shallow cloud top above 500 mb level) (6) the shallow cloud top is chosen by finding the layers where D(RH)/DP is a maximum from cloud base to 300 mb above cloud base and the upper (drier) layer of the two is selected as cloud top (was layer where RH dropped off by 10%) (7) shallow mixing line slope is modified by using the average of theta and q at cloud base and theta and q one layer (or 10 mb, whichever is higher) above cloud top to define the top of the mixing line (was theta and q at cloud top) (8) reference humidity profile for shallow convection follows original Betts formulation of a constant DSP (40 mb) first guess and a subsequent correction resulting in no precipitation (was Janjic formulation) (9) the reference temperature profile for deep convection is anchored at the moist adiabat running through cloud base (rather than the ambient temperature at the cloud base level) (10) in all tests of pressure depth above the surface (such as cloud base search), the surface pressure is PD(K)+PT (rather than the pressure in the middle of the lowest layer above ground) 07/12Z: The ETAY test beginning 03 June 12Z continued to use the GDAS soil moisture of the previous parallel run. This was modified to use the NGM soil moisture starting with this run. 25/12Z: The ETAY beginning 3 June was turned off. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 06/12Z: Penman was placed into ETAY with the stomatal resistance set to zero. 10/12Z: The Penman version of SURFCE still contained the accumulation of total precipitation which was moved to RAIN and CUCNVC in the standard code, thus the rainfall was being counted twice. This problem was removed. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 15/12Z: The Penman parallel was terminated and the sigma version was placed in ETAY. 17/12Z: A problem with mountains in the 3rd row with sigma was discovered after this run. The ETAY was turned off as of the 18/00Z cycle until a fix could be determined. 20/12Z: BOCOH was modified to permit the sigma coordinate to work properly where the steps exist in the third row of the domain; the sigma run began again. 28/00Z: The sigma test from 20 September ended with this cycle. 29/12Z: The Penman test was reintroduced with one difference from the August 93 test: the soil moisture is now taken from the climatology file rather than basing it on the NGM's moisture availability. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 25/00Z: The Penman parallel from 29 September ended with this cycle. 28/12Z: A test of modifications to the shallow convection code is introduced with this cycle which allows it to work more often. The depth of the shallowest cloud (PNO) is reduced to 10 mb (from 90 mb), and the shallow cloud top is selected by finding the maximum D(RH)/Dp (rather than finding a RH drop of 10% over one layer). ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 19/12Z: The shallow convection test from 28 October ended with this cycle. 22/12Z: Another test of modifications to the shallow convection code is introduced into the ETAY. The modifications from the 28 October test are included along with setting EPSTH=0. (from EPSTH=6.0). Also, a test of the "crippled" horizontal advection was placed into the ETAL beginning with this cycle. ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 07/00Z: The test of the crippled horizontal advection ended with this cycle of the ETAL. 07/12Z: Yet another test of modifications to the shallow convection code is introduced into the ETAY. The maximum D(RH)/Dp (see 28 Oct test) is used to find the shallow cloud top ONLY for "swap" points. The modification (EPSTH=0.0) of the 22 Nov test is also valid. Also, the layer above the shallow cloud top is used to determine the slope of the mixing line. 14/00Z: The gravity wave drag test was placed into the ETAL. ----------------------------------------------------------------------- ----------------------------------------------------------------------- 1994 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 05/00Z: The convection test in ETAY from 7 Dec 1993 ended with this cycle. 18/12Z: A test in the ETAY of a combination of the Penman surface evaporation scheme (29 Sept 93) with the modified shallow convection (7 Dec 93) began with this cycle. 27/00Z: The gravity wave drag test in ETAL ended at the previous cycle and was replaced with the 'N+1'-layer test. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 02/12Z: An error in the computation of the climatological soil moisture in the ETAY test of 18 Jan was fixed. The error consisted of not setting the soil wetness to the maximum value over snow cover and water. 25/00Z: The ETAY (shallow cnvc/Penman) ended with this cycle. 26/00Z: The new silhouette/average mountains were placed in ETAY (including the analysis, of course). ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 01/12Z: The executable from the previous ETAY parallel (Penman/Shal) was used for the new mountain parallel before this cycle. The code was recompiled beginning this cycle. 13/12Z: The test of layer n+1 from 27 Jan in ETAL ended with this cycle. 14/00Z: Joe Gerrity's test of modified evaporation of grid-scale rain was placed into ETAL. 15/00Z: The ETAL test was changed to include an inhibition on surface evaporation for RH > 70% in addition to the modification in RAIN. 30/12Z: The ETAL test was changed to include the modifications to shallow convection that are being tested in the EDAS. These changes match those made in the Jan 18 parallel along with the addition of a test for negative q in the reference profile. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 01/12Z: The ETAL now includes modifications of evaporation of grid- scale rain, shallow convection, and short wave radiative flux at the surface. The previous ETAY test has been terminated and replaced with modifications to shallow convection and short wave radiative flux. The changes to RAIN are from the previous ETAL run, changes to CUCNVC are from the Jan 18 ETAY, and the changes to radiation improve the diurnal cycle. 15/00Z: The ETAL test of Apr 1 has been replaced by a test of the Penman surface evaporation scheme using climatological soil moisture. This also includes modifications to shallow conv and surface radiation found in the Apr 1 ETAY test, making that the control for this run. 16/00Z: The climatological soil moisture used in the ETAL test has been corrected to adjust from a max bucket depth of .15 m (climo file) to .1125 m used in the model. 20/12Z: The ETAY was replaced by a test of several changes to the static analysis. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 03/00Z: The ETAL test of Apr 16 has been replaced by a test of the standard surface evaporation scheme using climatological soil moisture. 14/00Z: A bug in the ETAY test of Apr 20 involving the boundary conditions in the first 6 hours was fixed as of this cycle. 17/12Z: The ETAY test of Apr 20 now includes the climatological soil moisture used in the ETAL run. 24/12Z: The ETAL test of 3 May terminated with the previous cycle and is replaced by Zavisa's 'swamp' test with climatological soil moisture. 26/00Z: The ETAY test of Apr 20 now does not include the climatological soil moisture, but analysis changes only. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 01/12Z: A small change requested by Zavisa was made in the swamp test version of SURFCE in the ETAL run (the 185 and 280 loops). 20/12Z: The correction of RADIN was made to ETAL (see code log). 22/12Z: The correction of RADIN was made to ETAY (see code log). 28/12Z: Both the ETAL and ETAY were recompiled to include new profiler output (see code log). ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 23/00Z: The ETAY was replaced by a pre-implementation version of the analysis changes. This is a clean test of the current operational ETAX model with the CAFTI approved analysis changes. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 06/12Z: A new 40km/38lyr control and parallel was established. The control uses the new 'standard' forecast and post-processing codes without cloud water or 'plus 1'. The parallel uses a version of the 'standard' code to which the cloud water and 'plus 1' have been added. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 18/12Z: The ETAY was replaced by a version of the operational ETA system which used SSM/I total column water data in the OI analysis. 21/00Z: The control 40km run began using the new standard code (see code log) WITH cloud water and WITHOUT 'plus 1'. The 40km parallel is still clouds AND 'plus 1'. ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 11/00Z: The ETAL was replaced with the 80 km EDAS and a 48 h forecast from the on-time EDAS analysis. This test uses the "standard" version of the code (without cloud water) and the native grid version of FERR. 22/12Z: The test of SSM/I total column water data in the ETAY was turned off after this cycle was completed. 30/00Z: ETAY was turned on to test (a) a 5cm soil layer and (b) the new evolving albedo. The control is the operational Early Eta. ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 02/00Z: Corrections/improvements were put into the 40km/38lyr parallel regarding the 2m diagnostic quantities and the background value of exchange coefficients (for the supercritical case). 19/12Z: The parallel test of the 40km 'plus 1' was discontinued and replaced with a 40km test of the new EDAS which will be implemented in the 29km run. 20/12Z: The parallel test of the 40km 'plus 1' was turned back on after problems were noted in the timely availability of off-time PREPDA files in the EDAS parallel run. 31/00Z: The prepda file used in the T-0 h 80 km EDAS analysis in the ETAL parallel was changed from the regional file to the early file. ----------------------------------------------------------------------- ----------------------------------------------------------------------- 1995 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 06/00Z: The boundary tendency code (BCTEND) was revised to provide correct boundary conditions to the 80 KM EDAS running in the ETAL parallel. Previously, at each model start time during the EDAS the boundary file was read from the beginning; thus at the start of the 48 h forecast the AVN boundaries were 12 h old. 30/12Z: The ETAY was replaced by a version of the operational 80 km Eta with NESDIS moisture bogus data withheld from the analysis. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 01/12Z: The Eta model code used in the 80 km EDAS running in the ETAL parallel was recompiled with the correct constants for 1995 in ZENITH to compute the solar zenith angle. 28/12Z: The 40km/38lyr control remains the fundamental cloud water version; the 40km parallel was changed from 'plus 1' to the cloud version with Zhao's fixes of the saturation vapor pressure over ice (in PRECPD, GSCOND, RADTN, and CHKOUT) and of the critical value of Q for precipitation in PRECPD. For T>-10C then MI0=1.E-4; for T<-25C, MI0=5.E-4; between these two temperatures, MI0 varies linearly. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 07/12Z: The 40km parallel from 28 Feb ended with this cycle. 13/12Z: The impact test of no moisture bogus data in the ETAY parallel was replaced by an 80 km EDAS / forecast to run in parallel to the system in the ETAL parallel but with SSM/I total column water data included in the analysis. 23/00Z: The 40km/38lyr parallel was reinstated with the new "column" turbulence codes. 28/12Z: The value of KTMQ2 in the 40km parallel was increased from 2 to 8 in order to see if more vertical mixing of moisture would result. 29/00Z: Three occurrences of LP1 in MIXLEN were corrected to LMH(K)+1. 30/12Z: EPSQ2 and EPSL were changed from 1.E-4 to 1.E-2 in the 40km parallel. The affected routines are INIT, HDIFF, TURBL, MIXLEN, PRODQ2, and VDIFQ. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 03/12Z: Implicit solution to the vertical diffusion was placed into the 40km parallel and EPSL was increased from 1.E-2 to 1.E-1. 04/00Z: The top of the 40km runs was raised to 25 millibars. 04/12Z: The SSM/I total column water test in the ETAY parallel was replaced with an 80 km EDAS/forecast including GOES-I precipitable water data. The 40km runs for this cycle were corrupted by an error in the analysis. 05/12Z: The GOES-I test in the ETAY was turned off at the request of NESDIS; it will resumed in about 1 month after NESDIS upgrades their retrieval software. 20/00Z: The 40km parallel now contains Zavisa's new turbulence scheme as well as the new soil package. 21/00Z: The 40km parallel code has been updated to contain the RADFS albedo bug fix. (see code log) 21/12Z: Subroutine TURBL in the parallel code has been updated to fix a bug in setting THZ0 and QZ0. Also, updates in SFLX were made to change the sign of the sub-surface heat flux and bring the code up to date with Fei Chen's version. 25/12Z: Subroutines SURFCE and SFLX were brought up to date yet again. The reduction of over-moist GDAS soil moisture was put in SURFCE, and some additional diagnostics were added to SFLX. The entire code running in the parallel is in /dm/wd20mb/para40. 26/00Z: The 40km parallel was switched to using guess from the AVN rather than from the FNL. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 10/12Z: A bug in TURBL was fixed, VDIFQ was moved up to the end of the first loop. The 40 parallel was recompiled using code in /dm/wd22tb/mike, removing Zp from SURFCE and SFLX. 12/00Z: The 2-d version of the code (/dm/wd22tb/eagan/soil/full2d) began with this cycle. This includes the parallelized version of TURBL. 12/12Z: Bugs were removed from CRST21.f, but /wd2 disk crash affected 12Z and 13/00Z runs. 12/12Z: The ETAY parallel using GOES-I total column water was turned back on starting with this cycle. 15/12Z: Bugs removed from CHKOUT.f, similar to previous bugs in CRST21. Posted fields in previous runs are ok except for fields involving u, v, q, q2, cwm, ttnd, train, tcucn. 18/12Z: Subroutines (2-d) SURFCE and TURBL were further parallelized and the 40km parallel was recompiled for this cycle. 19/12Z: 40 km parallel is now 1-d standard code with new soil model. Non-standard code comes from /dm/wd22tb/mike and consists of ACMSFC, CHKOUT, INIT, RADTN, SFLX, SOIL, and SURFCE. 22/12Z: The 40km parallel was changed back to the newest version of the soil/turbulence 2-D code. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 21/12Z: An error was discovered in the GOES-I total column water processing by FERR in the ETAY parallel. The index of array MAXBUF was incorrectly incremented for the GOES-I total column water, causing FERR to throw out 90% of the data. The problem was fixed starting with this run. 23/12Z: The new 2-d soil/turbulence code in the 40km parallel was changed to include: (1) Zhao's fixes in PRECPD regarding excessive evaporation of precipitation and improper conden- sation of vapor directly to precip and (2) Mahrt's fix to the surface exchange coefficients. 26/12Z: The FERR code used in the GOES-I total column water test in the ETAY parallel was modified to exactly partition the TCW data onto the 38 layers of the 80 km model. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 06/12Z: It was discovered that the 40km parallel script was utilizing an old version of the parallel (marriage) executable. This was corrected so now the 40km parallel includes Zhao's fix of the vapor-to-precip bug and his control on evaporation of precip as well as Mahrt's fix to the surface exchange coefficients. A new 40km/38lyr mini-parallel was created to test Zavisa's replacement of the Mahrt fix with Paulsen functions. 31/00Z: The test of GOES-I precipitable water data in the ETAY parallel was turned off after this cycle. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 03/12Z: The mini-parallel was modified to test Zavisa's changes to get variables at 2 and 10m (and remove SHLTRX and AMEMLV6). Other diagnostic and bufr routines were also stripped out of the code to run in a separate post-processor for the sounding output. The routines that were removed for this were WETBULB, CALHEL, CALWXT, BFRIZE, and BFRHDR. All new source was copied from /dm/wd20mb/current/sndpost/2m/. The new sounding post-processor was copied from /wd2/wd20/wd20mb/post/postsnd.f using modified bufr tables. This code was been jif'ed for the operational Meso as well. Also, the 40km post was recompiled to remove the processing of accum precip type. This is in anticipation of its removal from the restart files, since it will not be computed in the model anymore. This was done by copying source found in /dm/wd20mb/current/sndpost/post/. No permanent changes have been made to the sources for the MESO or standard codes. 12/00Z: In an effort to discover the cause of degraded oceanic precipitation forecasts in the 48 km EDAS, processing of SSM/I precipitable water was turned off starting with this cycle. 14/12Z: A test of GOES-I sounding data in the 80 km ETAY parallel was begun with this cycle. 16/00Z: After 3 weeks of degraded 48 km forecasts of precipitation over the ocean (since 00Z 7/24), the processing of moisture bogus data in the 48 km EDAS was turned off. 17/12Z: Dennis Keyser corrected the processing error of moisture bogus data in the prepbufr files use by the 48 km EDAS (RH profiles were 10 times too small) and the processing of moisture bogus data in the 48 km EDAS was resumed with this cycle. 18/12Z: Processing of SSM/I precipitable water data in the 48 km EDAS was turned back on. 22/12Z: A bug was fixed in ETA3 in the computation of QOL2ST (AUBR needs to be multiplied by CUBR). 22/12Z: Test of GOES-I sounding data in ETAY was turned off after this cycle due to problems with data in prepda file and surface pressure update in OI code. 28/12Z: OI analysis in 48 km EDAS changed to reduce observation error standard deviation of surface data. 28/12Z: Test of GOES-I sounding data in ETAY turned back on after fixes to FERR (use GOES data on every input level) and to PREPDA (add suspect quality mark to GOES surface pressure data to keep them out of the surface pressure update in the OI analysis 30/00Z: GOES-I sounding test in ETAY turned off after this cycle; fixes described in 28/12Z entry had no impact on extrapolation problems along edge of the GOES data in the Gulf of Mexico. 31/12Z: Attempt to test NOAA-14 soundings in ETAY revealed that no TOVS soundings were being used in both the ETAL, ETAY, 48 KM EDAS, and the operational 80 km eta (switch NOSATS=TRUE in FERR). A test with the TOVS data in the ETAY parallel was begun with this cycle. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 21/00Z: The test of TOVS soundings from NOAA 12 and 14 in the ETAY parallel was turned off after this cycle. 26/12Z: SFCDIF was modified in ETA3 to incorporate the Zilitinkevitch fix for Z0T using C=0.10. The 40km post was recompiled to include changes to GRIB parameter and level numbers that are to be included in the next documentation. (and in the Meso) 27/00Z: Processing of TOVS soundings in the Eta OI analysis was turned on in the ETAL parallel. 27/00Z: The ETAY parallel was tuned back on to test the impact of turning off the processing of wind profiler data in the OI analysis. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 05/12Z: What was in ETA3 is now in ETA1, the 40km control run. This is basically the jif'd Meso code, obtained from ~copyprod. The 40km parallel (ETA2) has a modified version of MIXLEN with ALPH=0.3. ETA3 has been put to rest. 13/12Z: ETA2 now has a test of the Mahrt surface exchange coeff. All other code is the same as ETA1. 23/12Z: The parallel 48 km run was changed to point at the operational executables and fixed fields rather than those in /ptmp1/wd22gd/ eta4838. ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 01/12Z: The source of output files populating the nic GCIP directories was switched from the parallel 48 km run to the operational 48 km run starting with this cycle. 02/00Z: The test w/no wind profiler data in the ETAY parallel was turned off after this cycle. 02/12Z: The input prepda files for the 80 km EDAS running in the ETAL parallel were changed from the RDAS/RGL prepda files produced on the NAS to the EDAS/Early Eta BUFR prepda files. 06/00Z: The parallel 48 km EDAS/Eta model running under wd22dd was turned off after this cycle. 07/12Z: The pressure of the model top in the 80 km ETAL/ETAY parallels was raised to 25 mb. 21/12Z: The 40km parallel/ETA2 was changed to mimic the 29km Meso except for using TREL=3000 in cuparm (cuparm_r4038 in ~standard). The 40km control/ETA1 was recompiled using the 29km opnl Meso code and cuparm just to make sure. ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 18/12Z: The 80km ETAL parallel was changed to use the 2-d code with the 2-layer soil model and the column turbulence (so-called "marriage"), identical to what is running in the meso eta. A 12h EDAS is still used to initialize the 48 h forecast. ----------------------------------------------------------------------- ----------------------------------------------------------------------- 1996 ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 01/12Z: A continuously cycled EDAS started from this cycle in the 80 km ETAY parallel. This run uses the same code running in the ETAL parallel (2-d w/soil model and column turbulence). ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 23/00Z: The 80 km ETAL and ETAY parallels were turned off on cray3 after this cycle and will be moved to cray4. 26/00Z: The 48 km parallel EDAS/48 h forecast was turned on for a test of Fedor's new form drag. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 01/12Z: The 80 km ETAY parallel (cycled EDAS) was restarted on cray3. The 80 km ETAL parallel (non-cycled EDAS) will run on cray4. 03/00Z: The form drag experiment in the 48 km parallel was slightly modified to use the operational version of MIXLEN. 16/00Z: The form drag experiment in the 48 km parallel was replaced by a test of the new post-processor for Geoff Manikin. 24/12Z: The 48 km parallel was turned back on to test GOES 8-9 total precipitable water data. 30/09Z: COAC was changed from >2.0 to 0.16 in the 10km nests. 31/09Z: COAC was changed from 0.16 to 0.30 in the 10km nests. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 01/09Z: COAC was changed from 0.30 to 1.0 in the 10km nests. 07/09Z: DATA variables without SAVES were found in RADTN. CC and PPT were moved to COMMON RD1TIM. 08/09Z: The Lynch digital filter was incorporated into the 10km nests. NRLX1 changed from 70 to 100 and NRLX2 changed from 35 to 75. 09/09Z: The western nest was turned off and the eastern nest is now running every day. 11/09Z: The Lynch filter was put into the 10km nest. 15/09Z: NTIM in the Lynch filter was reduced from 150 to 125. 17/12Z: The GOES 8-9 precipitable water test in the 48 km parallel was turned off after this cycle. 18/12Z: A test of the 2-d flux correction for linear advection was started in the 48 km parallel. 21/15Z: The initial relative humidity field in the 10km nest (not modified by the Lynch filter) is being smoothed prior to the recovering of q using T from the Lynch. The RH is smoothed with a 25-pt Bleck filter modified for the E-grid. NTIM reduced from 125 to 110 in DIGFILT. 23/15Z: The initial T field in the 10km nest is being passed through the 25-pt filter before the Lynch filter is called. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 02/12Z: The test of the 2-d flux correction for linear advection in the 48 km parallel was replaced with two items: 1) Scaling of input GDAS soil moisture in SURFCE was being erroneously applied every time the model was started, even if RESTRT=TRUE. The code was fixed to apply this scaling only if RESTRT=FALSE. 2) Modified monthly green vegitation fraction. 02/15Z: The Lynch filter was introduced after the interpolation of the fields from the Meso Eta to the nest's 10km grid to reduce the considerable noise that was present in the first guess for the subsequent 3D-var. 08/15Z: The artificial truncation of soil moisture at timestep 1 (from the global) was eliminated in the 10km nest. 10/12Z: The scaling of the input GDAS soil moisture in the 48 km parallel was moved from SURFCE to INIT, since RESTRT is reset to FALSE before SURFCE is called; therefore, in the test starting 12Z 7/2, the scaling was still erroneously applied every time the model was started. 12/00Z: The OSU base soil evaporation scheme in SFLX was replaced with a Noilhan-type scheme in the 48 km parallel. This was added to the changes added on 12Z 7/2 and 12Z 7/10. 19/00Z: A new cloud fraction scheme was added to the 48 km parallel; this involves changes to CHKOUT, CLO89, PRECPD, RADTN, RADFS, and SWR92 in the forecast model and ETAFLD2 in the post. 30/00Z: 48 km parallel was turned off after this cycle 30/15Z: A new version of the 3dvar analysis code running in the 10 km olympic run was implemented, the major change being the use of surface data by the analysis. In addition, the script and codes used to create the prepbufr file for the 3dvar analysis was changed. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 01/03Z: Switches in the new data processing script for the 10 km olympic run were not set correctly. The script defaulted to operational versions of these codes, not the special versions which Dennis Keyser set up for the olympic run. This resulted in some observation types (such as aircraft temperatures) not being processed by the 3dvar analysis. The switches were correctly set starting with this cycle. 02/00Z: The "bundle" of changes planned for the Early Eta was placed into the 48 km parallel. This includes the following: 1) The fast model and post code, with the correction/addition of fields in the post. 2) The Noilhan-type bare soil evaporation scheme in SFLX 3) Zhao's new cloud fraction scheme involving changes to CHKOUT, CLO89, PRECPD, RADTN, RADFS, and SWR92 in the forecast model and ETAFLD2 in the post 4) Fedor's form drag scheme. 5) Zavisa's new scheme for advection of q and cloud. 6) A 3% reduction in the solar constant to compensate for the lack of aerosols in the shortwave radiation parameterization. 7) Modification of solar constant as a function of the day of the year to account for the ellipticity of the Earth's orbit around the sun. 8) Modified RADTN to include ozone and to squeeze the entire column of ozone into the model column even though the model top is > 0 mb. 9) Increasing the minimum exchange coefficient from 1.E-2 m2/s to 1.E-1 m2/s. 02/12Z: A new 80 km parallel (designated ETAV) was started on cray4. It currently runs a 12-h EDAS (from a GDAS guess)/48 h forecast using the regional 3dvar analysis. 05/03Z: This is the last run of the 10km Olympic nest on the Eagan C90 (Rain). 06/12Z: The input parameters for the 80 km 3dvar analysis (e.g., correlation lengths) were changed by Dave Parrish for the ETAV parallel to be consistent with 80km/38 level resolution. 08/12Z: The 3dvar analysis used in the ETAV parallel was changed to perform higher order (linear to 2nd order) interpolation of potential and streamfunction for improved fit of the analysis to wind data. The input parameters to the code were also modified to have the data better fit the analysis. This was also changed in the 10 km olympic run. 13/12Z: The version of grdeta in the 48 km parallel was changed to use the new 0.15 deg lat/lon file of vegetation fraction. 15/03Z: The value of sinf was increased from 1.3 to 2.0 in the 10 km 3dvar analysis. The maximum eigenvalue is multiplied by this number. It has the effect of slowing down the convergence towards the observations in the 3dvar code. 26/03Z: The 10 km olympic run was turned off after this cycle. 27/03Z: Subroutine INIT for the 48 km parallel was modified to make the scaling of the initial GDAS soil moisture a function of the month. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 18/00Z: The test of continuous cycling in the 80 km ETAY parallel was turned off after this cycle. 19/00Z: A new station list for the hourly output was placed into the 48km parallel EDAS. 21/00Z: The 48km parallel post was updated to include grid 216, and 2 more stations were added to the hourly station list. 23/12Z: The 48km parallel post was updated to fix a problem with the time range indicator for several fields. 25/00Z: The 48 km parallel with the "bundle" of changes/enhancements (see 00Z 8/2/96 entry) was turned off after this cycle. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 02/12Z: A 48 km parallel with the degraded (50% less) U.S. rawinsonde network was started with this cycle in the ETAY slot. 09/12Z: The 48km parallel was turned back on with a test of the adaptive rawinsonde network over the CONUS. 12/00Z: The test of the adaptive raob network in the 48 km parallel was changed to have all raobs turned on at 12Z and none at 00Z. ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 05/12Z: The 80km ETAL and ETAV parallels on cray4 were temporarily turned off to make a set of changes (see 08/12Z entry). 07/12Z: The 48km ETAY parallel (1/2 RAOB test) and the PARA48 parallel (12Z RAOBS only test) were turned off after this cycle. 08/00Z: The 48km "bundle" was placed into the PARA48 parallel. 08/12Z: The 48km PARA48 parallel post was modified to fix a problem with the time range indicator. 08/12Z: The two 80 km parallels on cray4 (ETAL - 80km EDAS control; ETAV - 80km 3DVAR) were changed to 1) run the "bundle" of changes planned for the Early/Meso Eta and 2) converted to continuous cycling mode. 09/00Z: The 48km PARA48 fcst executable was modified to fix a problem with the sign of the diagnostic arrays for evaporation. 13/00Z: The 48km PARA48 is now using the grib sst in grdeta. 13/12Z: The 48km PARA48 post was modified to fix a problem with the accumulation time for the evaporation. 19/12Z: The 48km PARA48 post was modified to back out changes to PDS regarding boundary layer fields in GRIB. 20/12Z: The 48km PARA48 post was modified to back out changes to PDS regarding the boundary layer LI in GRIB. (see cmeb1 in GRIBIT, MISCLN, and CLDRAD to put DRG changes back in) ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 03/00Z: The 80 km 3DVAR parallel on cray4 was modified to use satellite temperature data over the ocean in the analysis. This was done by modifying errtable file read in by prevents. 04/12Z: Modified errtable file in prevents to allow for use of surface land temperatures in the 3DVAR parallel. 06/00Z: Restarted cycling in both the ETAL and ETAV parallels. 06/12Z: The 48 km upgrade ("bundle") in the PARA48 parallel was replaced with a test of pre-processed observations files made from BUFR input from the Cray instead of ON29 input from the NAS. 10/12Z: The PARA48 parallel is now a true "no-NAS" test, using GRIB sst files and snow analysis files (both USAF and SAB) created on the c90, in addition to the BUFR input to prepdata. 12/00Z: The regional 3dvar analysis used in the 80KM ETAV parallel was changed to process SSM/I precipitable water data. 20/12Z: Switched to Wan-Shu's new errtable for the regional 3DVAR parallel. 24/00Z: The 00z 48 km parallel was turned off to facilitate the 09z meso/ 10 km western U.S. run. 28/12Z: The 12z 48 km parallel was turned off to free up computer time for debugging the 09Z meso/10 km runs. 30/00Z: The 80 km parallels on cray4 (ETAL and 3DVAR) were turned off after this cycle to free up cray4 for debugging of the nested version of the eta model. The ETAL parallel (EDAS with OI) was running with NOSATS accidently set to TRUE in feta26.parm since the parallel was moved to cray4, therefore ETAL ran without TOVS thickness retrievals since May 1996. The 80 km ETAY parallel on cray3 did not have this error. ----------------------------------------------------------------------- ----------------------------------------------------------------------- 1997 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 06/09Z: The 09Z meso was turned back on using the "bundle" versions of the model/post with nesting to the 10 km western domain. 06/09Z: First successful run of the revived 10 km nest in the west using the 09Z meso for BC. A 3dvar analysis using the 09Z meso OI analysis for a first guess is used for initial conditions. 17/09Z: The length of the Lynch in the nest-in-the-West was increased from 110 to 180 timesteps in each direction. 21/09Z: NEXRAD radial velocity data (NIDS) turned back on for 10KM 3dvar analysis in 09Z 10 km nest in the west. 23/12Z: The 48 km parallel was turned back on (12Z only) to test the changes to be implemented (the "bundle") in early February. 24/09Z: The land-sea mask for the 10 km western run was modifed by a change in CNSTS.f (in code INTRST) to make the Great Salt Lake points water (SM=1) and ice free (SI=0). Since SNOHIRES made these points snow-covered, CNSTS was resetting the sea ice mask to one and the land-sea mask to zero. 26/09Z: The length of the Lynch in the western nest was reduced back to 110 from 180 timesteps since instability was arising, apparently due to the absence of diffusion. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 03/12Z: The 48k parallel is now a true "no-NAS" test, with: 1) The "bundle" package of changes 2) BUFR observation dumps 3) The new version of Bill Collins' cqc code 4) The "bundle" grdeta code with Ken Mitchell's modifications to use the cray snow analysis files in /dcom/us007003 04/09Z: The 09Z meso grdeta was modified to use the cray snow analysis files. 05/09Z: INTRST for the 29km to 10km interpolation was modified to use the cray snow analysis files needed by Ken Mitchell's revised version of SNOHIRES 17/12Z: The 48km parallel (12Z only) run started at 12Z 2/3/97 was turned off after this cycle. 27/03Z: The nest-in-the-West was moved from being initiated by the 09Z Meso parallel to the 03Z operational Meso. 27/12Z: The 80km ETAL parallel (cycled EDAS with OI analysis) was restarted on cray3 and the 80km ETAV parallel (cycled EDAS with 3DVAR anl) was restarted on cray4. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 05/00Z: Started new cycle for ETAL and ETAV parallels after Dave Parrish fixed a bug in the 3dvar analysis in the processing of rawinsonde moisture data which created an unrealistic initial RH field. 06/03Z: Began using new version of 3dvar analysis in the 10km western run (see 3/5/00Z entry). Error in previous version had no impact on past 10 km runs since there are no raobs at 03Z or 09Z. 11/12Z: Cray3 outages caused break in ETAL cycled EDAS on 00Z 3/9. New cycle for both ETAL and ETAV parallels started from this cycle. 13/03Z: The 10km nest-in-the-West began using 1-hrly BC's and the version of CUCNVC that enhances deep convection over elevated terrain. 17/12Z: The make_prepbufr script used by the 80 KM 3DVAR parallel was modified based on the current operational script, which has no connection to the NAS. 18/03Z: Several changes were made to the 10 km western run: - A bug was removed from RADTN in the 10km nest-in-the-West: HBM2 and HTM need to use the IRAD array for the proper offset of I when in the "double row" I=1,IMT, but they were not. - A bug was fixed in intrst in which lat/lons of zero were fed to subroutine PUTVEG, which interpolates the 0.15 deg vegetation fraction data to the e-grid. The bug caused a value of 0.55 to be assigned to every point in the 10 km grid. - The make_prepbufr script was modified to mirror the operational script, which has no connections to the NAS. 26/00Z: Machine problems caused break in ETAV cycled EDAS on 12Z 3/24. New cycle begun from this time for both ETAL/ETAV parallels. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 04/03Z: Turned on surface winds over land in the 10 km 3DVAR analysis for the nest in the west. 05/12Z: Cray3 outages caused break in ETAL cycled EDAS. 08/00Z: Started new cycle in ETAL and ETAV parallel EDAS runs. Also modified these parallels to bring in updated snow analysis only once/day at the beginning of the 00Z EDAS cycle (12Z valid time). 23/00Z: Discovered that ETAL parallel (80km cycled OI) had depth of deep soil layers set to 2 m, instead of 1 m as advised by Mitchell/Chen for cycled EDAS. (3DVAR EDAS was set correctly to 1 m). Depth was changed to 1 m in ETAL starting with this cycle. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 13/03Z: Put the new dynamics driver and routines (Fedor's) into the Nest-in-the-West. 20/12Z: Discovered a bug in the ETAV (3dvar) script. The 3dvar analysis had no date check for the observations vs. guess, so the code could use a 3-h old prepda file if one was not created for any reason. This caused the terrible forecast in the ETAV run for 12Z 4/3/97. The problem was fixed by deleting the prepda file after each 3dvar execution. 22/03Z: The code used prior to 13 May (without the new driver) was put back into the Nest-in-the-West due to strange precip patterns in forecasts using the new code. 24/03Z: The old driver code was put back into the 32km/45lyr Early Meso. 27/00Z: The code using the new dynamics driver was reconstructed and put back into the Early Meso parallel. 27/12Z: The ETAY parallel was reactivated as an 80KM cycled EDAS with OI analysis and assimilation of Baldwin's hourly analysis of radar-estimated precipitation using Y. Lin's codes. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 03/00Z: Fixed error in computation of the adjoint of the winds in the regional 3DVAR analysis caused by inappropriate vectorization in subroutine SCRIPTG. Changed in both the 00Z 80 km parallel on cray4 and the 03Z 10 km nest-in-the-west. 04/00Z: Modified Early Meso parallel to use off-time AVN forecast for boundary conditions from t-6 h to t+48 h. 05/00Z: Correction made in RADTN of the Early Meso to fix diagnostic output of cloud fraction by not excluding stratiform cloud when convective is present and by not skipping the diagnostic when the surface pressure was too low (elevation too high). 05/12Z: Fixed error in 3dvar analysis in which the guess temperature was not converted to virtual temperature before the 3dvar computed the first guess geoptential heights. Changed in both the 80 km parallel on cray4 and the 03Z 10 km nest-in-the-west. 10/00Z: The number of relaxations in Fedor's SLP reduction was increased from 100 (at 0 hrs) and 75 (all other hours) to 200 and 150 in the Early Meso parallel. 10/12Z: The following changes were made in the 80 km parallels: 1) ETAL was changed from cycled OI EDAS to cycled 3DVAR EDAS with 4-layer soil model (did not run this cycle due to unknown error in forecast code) 2) ETAY was changed from cycled OI EDAS with precipitation assimilation to cycled 3DVAR EDAS with GOES PW data (did not run this cycle due to script error) 3) ETAV remains the control cycled 3DVAR parallel but the cycle was restarted. 10/12Z: The Early Meso parallel was converted to run 3DVAR analysis and with a cycled EDAS. The 00Z run still does a 12-h EDAS with a 48-h forecast, while the 12Z run will do a 12-h EDAS only. This cycle did not run due to script error. 11/00Z: ETAY parallel (3DVAR with GOES PW data) turned off until a fix to Parrish's GOES-PW interface code can be found. 12/00Z: First successful cycled run with the new ETAL parallel (3DVAR with 4-layer soil model) 12/03Z: Zhao's fix to the diagnostic cloud in RADTN removed the 'holes' previously seen in cloud fraction in the Nest-in-the-West. 17/00Z: First successful cycled run with the Early Meso parallel (with 3DVAR). 18/12Z: First successful run with cycled 3DVAR/GOES-PW in ETAY parallel. 19/00Z: Restarted cycle on all 3 80km parallel 3DVAR EDAS runs. 19/00Z: Increased NRLX and NRLX2 in CHKOUT in the Early Meso from 200 and 150 to 500 for each. 27/03Z: The Nest-in-the-West was switched from the original domain to the new small one covering 50% less area. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 01/12Z: Two changes made in the Early Meso: (1) Fedor's newest steps (less raising of points to fill holes) with accompanying z0eff were introduced; and four points over the Great Salt Lake were turned to water by hand. (2) New EBU, INIT, and RADTN were put in to fix failures of the 0-hr posts. 02/00Z: An error in the implementation of 2) above corrupted this run of the early meso and caused the EDAS cycle to be broken. Bug was fixed in time for the 12Z 7/2/97 cycle. 03/00Z: NOAA-12 retrievals turned off in the 3DVAR parallels. 03/12Z: Fixed glitch in 4-layer soil EDAS (ETAL) parallel in which surface runoff routine was not generalized for more than 2 soil layers. 09/00Z: Turned on wind profiler data in 3DVAR prepbufr, using ob error statistics obtained from Bill Collins. Also increased ob errors for pibal observations to be ~50% higher than rawinsonde winds instead of the same errors as raobs. 09/12Z: For this cycle only VAD winds were turned on in the 3DVAR prepda files using observation error statistics obtained from Bill Collins. 10/00Z: NRLX set back to 250 (0 hrs) and 100 (all later hours) in CHKOUT in the Early Meso. Also, the fix to ETA2P was put into the Early Meso's post for getting cloud onto p surfaces. 10/03Z: Fedor's new dynamics driver put into the Nest-in-the-West with NRLX=500. 10/12Z: Dave Parrish corrected GOES-PW version of the 3dvar analysis code to allow it to process all GOES data over land. In older version any PW ob which was not surrounded by 4 eta points at the same elevation was excluded from the analysis. 11/03Z: In the Nest-in-the-West: (1) NRLX was reduced to 250 and 100; (2) the corrected ETA2P was put into the post (for correct computation of cloud water on pressure). 14/12Z: Changes to output from Early Meso: 1) EGRD3D file will only have surface files 2) GRBGRD (90 km polar stereographic) turned off 3) 32 km North America lambert conformal grid (#217), to be called AWIP32, was turned on. The post-processor was modified to process grid 217. 19/00Z: At Fedor's request the GRBGRD (90 km polar stereographic) file was turned back on in the Early Meso parallel. 22/12Z: Added processing of tropical cyclone synthetic data in the prepbufr script for the eta 3DVAR parallel runs (except for the nest in the west). 23/12Z: UTIM is now hardwired to 1.0 in the Early Meso so that the artificial aid to moisture spinup is removed. 25/00Z: Dave Parrish made the following changes to the regional 3dvar analysis running in the 80 km ETAV parallel, the 32km Early Meso, and the 10 km nest in the west: 1) Background error : Increased the background error smoothly by a factor which varies from 1 at eta=0.5 to 2 at eta=1, appiled to all variables. Also decreased the horizontal and vertical correlation lengths for all variables smoothly by a factor which varoes from 1 at eta=0.5 to 0.85 at eta=1. 2) Observation error: Decrease observation error for temperature and moisture as a function of distance from model terrain. The function is: factor = 1./(1+(a0-1)*exp((-d/d0)**2) where for temperature, a0 = 3, d0=.1 and for moisture, a0 = 1.6, and d0 = .09 d is the distance in eta units from the observation to the model terrain When at the model terrain, factor = 1/a0. The temperature error is reduced by a factor of 3 at the model surface, and moisture by a factor of 1.6. This change was not put into the ETAL parallel (4-layer soil) or the ETAY parallel (GOES-PW data). 29/00Z: MIXLEN was modified in the Early Meso to put back Fedor's part of the computation for the mixing length that was removed from the original marriage code. Other fixes were related the an improper search for the top of the PBL which began occurring when EPSQ2 was increased from 0.02 to 0.2. Finally the 165 loop was fixed so that EL is not referenced out-of-bounds. 29/03Z: The Nest-in-the-West was recompiled with the preceding fixes to MIXLEN. 31/00Z: The following changes were made to the regional 3dvar code running in all 80 km parallels, the 32 km Early Meso, and the 10 km nest in the west: 1) The observation error modification (see 00Z 7/25) in the code was removed. 2) The GOES-PW version is now the current "operational" version running in all parallels (only ETAY parallel is using this data). 3) The background error change in the 00Z 7/25 entry is retained. The observation error table read into prevents was changed: 1) Rawinsonde RH errors from SFC-500 mb were reduced, set to 5% at the surface increasing to 20% at 500 mb and above. Previous table had RH error=20% at all vertical levels. 2) SSM/I PW ob error increased from 4 to 6 g/kg, while the maximum GOES PW ob error was also increased from 4 to 6 g/kg. 3) Observation errors for TOVS temperatures (clear and cloudy retrievals) were increased by 2C at all pressure levels. 31/12Z: Changed UNIX script running the GOES-PW test (ETAY parallel) to remove precipitable water files after each 3dvar execution. This prevents old data from being used in case of missing data. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 02/00Z: Discovered bug in processing GOES-PW data for the ETAY parallel 12Z cycle (introduced around 12Z 7/21/97) in which the tm12 GOES-PW data was being used at tm00 and the tm00 data was being used in the tm12 analysis, while tm09, tm06, and tm03 were missing. 04/12Z: The following changes were made to the 80 km parallels: 1) The ETAL parallel (cycled EDAS/3DVAR with 4-layer soil) had VAD wind observations turned on in the 3dvar analysis. The cycle was restarted. 2) The ETAV parallel (cycled EDAS w/3DVAR) and the ETAY parallel (cycled EDAS w/3DVAR, GOES-PW data) were converted to use the 4-layer soil model. The cycle for both parallels were restarted. 07/12Z: The soil scheme was changed from 2 to 4 layers in the 32km Early Meso. 07/12Z: The 80 km parallels were converted to use the same model code as the 32km: 1) Elevated convection; 2) new version of EBU.f 3) Fedor's changes to MIXLEN (see 00Z 7/29/97 entry). 11/12Z: The ETAV parallel (80 km 3DVAR control) was moved to cray3 and the cycle for all three 80km parallels was restarted from the GDAS. 12/12Z: The observation errors for VAD winds were reduced by 1/2 at all levels in the ETAL parallel. 21/03Z: This was the last regular run of the Nest-in-the-West (reduced size). 21/12Z: The 3dvar analysis running in the 32km and 80km parallels was changed by Dave Parrish to include a surface pressure analysis. 26/12Z: The input parameters for the 32km 3dvar analysis were changed by tuning the background error to be greater, thus giving less weight to the guess and more to the observations. This was done by changing bshrink* from 1.6 to 1.0, pshrink* from 0.8 to 0.6 and bscale* from 1.35 to 1.6. 26/12Z: The ETAL parallel (80KM cycled EDAS with 3DVAR and VAD winds) was turned off after this cycle. 27/00Z: VAD winds were turned on all parallels running the 3DVAR analysis (ETAV, ETAY, and the 32 km Early Meso) starting with this cycle. 27/00Z: Changed value of REFKDT from 1.288 to 3.0 in SFLX.f in the 80 km parallels. This has the effect of decreasing surface runoff and increasing storage in the top soil layer. 27/12Z: Three changes were made in the 32km Early Meso: (1) SFLX changed as described in the previous cycle; (2) The operational version of CUCNVC was put back in (thus the SWAP is on and enhanced elevated convection is off); (3) CHKOUT was modified so as to rotate the 10m winds that go into the BUFR output. 28/00Z: Changes (2) and (3) above were made to the 80km parallels (ETAV and ETAY). 28/00Z: An analysis of the geopotential height data was added to the regional 3dvar analysis by Dave Parrish and was turned on in the 80 km parallels 28/12Z: Two changes to the 32km Early Meso: (1) The change to the 3dvar analysis described in the previous entry was put into the Early Meso. (2) The new vertical distribution of the 45 vertical layers was introduced. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 02/00Z: The 80km ETAY parallel (cycled 3DVAR EDAS with GOES PW data) was turned off after this cycle. The last cycle with data (read from Burt Katz's data tank) was at 00Z 8/29/97. 02/12Z: Two changes made to the 32km Early Meso and 80km parallels: (1) NTDDMP changed from 2 to 1 (2) Swaps from deep to shallow convection were turned off over land. 02/12Z: Parameters bscaleh and bscalet were increased from 1.5 to 2.5 for the Early Meso 3dvar analysis. This has the effect of increasing the background error and drawing more closely for the height and temperature observations. 08/12Z: Two changes were made to the 3dvar analysis running in the Early Meso and 80 km parallel: (1) Parameters errgeo0,errhgeo0 in the r3dvar.parm file were increased from 1 to 2. This has the effect of decreasing the geostrophic coupling between height and wind data (2) Use of ob type 183 (surface data with missing station pressure/ elevation) in the height analysis was turned off. These data have assigned to them a surface pressure of the standard atmosphere at the observed temperature, which lead to erroneous surface pressure analysis. The EDAS cycle was restarted for both parallels. 11/12Z: An error was corrected in the 32km Early Meso: CLDEFI was not being initialized at all (a remnant of the switch to the new driver). 11/12Z: An 80 km cycled OI EDAS was started in the ETAL parallel for this cycle. 12/12Z: The 11/12Z fix of CLDEFI in CUCNVC was put into the 80 km parallel. 13/00Z: The following changes were put into the 3dvar analysis running in the Early Meso and ETAV parallels: (1) An error was fixed which was causing significant errors in the analysis of geopotential heights. (2) The code was modified to ingest GOES-PW data from the prepbufr file. The EDAS cycle was restarted for both 3dvar parallels and the OI EDAS parallel running in the ETAL slot. 13/12Z: The GOES-PW data was turned off in the 3dvar analysis due to an unknown code error which caused the analysis to fail. 15/12Z: GOES-PW data were turned back on in the the 3dvar analyses (80 km and 32 km) after Dave Parrish fixed a multi-tasking error in the code. 16/00Z: At Geoff's request the GOES-PW data were turned back off in the 80 and 32km 3dvar analyses. 16/12Z: Increased LSM (number of allowed pressure levels for posting) from 19 to 39 in the 80 km parallel codes. 21/00Z: Turned off the 3dvar parallels (32 and 80km) temporarily since Dave Parrish has found several major errors in the code which caused problems in the analysis of geopotential height data. 22/12Z: The parallels running with the 3dvar analysis (ETAV and the Early Meso) were turned back on after Dave Parrish fixed several errors in the use of geopotential height data in the analysis. Also an error was fixed in CUCNVC in the EMSO and 80km parallels: negative precip was being counted over land where SWAP is turned off. 24/12Z: Changes to the 3dvar analysis in the 80 and 32km parallels: (1) GOES-PW data were turned back on since these data were also switched on in the OI analysis running in the operational EDAS/Early Eta and the 80 km ETAL parallel. (2) Dave Parrish changed the 3dvar analysis code to fix a bug in the treatment of SSM/I sfc wind speed data. With this fix the first guess wind direction was used to compute the u- and v- components. In the previous version the code set u=0 and v equal to the observed wind speed. 25/12Z: Turned on observation type 284 (surface winds with missing station pressure) in the 3dvar analysis with the same observation error variance (3.5 m/s) as surface land data with valid station pressure. 30/12Z: Modified 3dvar code to introduce latitudinal dependance for the rotational part of the wind by reducing background error. This was done to either eliminate spurious tropical waves/storms or reduce intensity of preexisting systems. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 03/00Z: Restarted EDAS cycle for 80 km (ETAL/ETAV) and 32 km parallels. 09/12Z: (1) Dave Parrish modified latitudinal variation of background error for rotational wind, temperature, and geopotential heights in the 3dvar analysis to be similar to the error statistics from the global ensembles. This change adjusts the 9/30/97 modification by making the adjustment to the background error (giving it more weight) more abrupt in the mid-latitude/tropics transition zone between between 20N and 30N than was done previously. (2) In addition, a similar modification was done for moisture and divergent part of the wind (variation is reversed with less weight given to the background in the tropics). (3) The EDAS cycle was restarted for the 80 and 32km parallels. 09/12Z: Turned back on the ETAY parallel (80km cycled EDAS w/3dvar analysis) for a test of Joe Gerrity's new version of CUCNVC (version= CUCNVC.OCT97). 18/00Z: Two modifications to CUCNVC by Joe Gerrity were put into the ETAY parallel (His "CUCNVC.NOV97" version). 18/12Z: The 32km parallel EDAS was converted to "partial cycling" in which PD, T, Q, U, and V at the beginning of the EDAS are obtained from the GDAS, while all other variables (e.g. cloud water, soil moisture/ temperature, TKE) are cycled continuously. 27/12Z: Swap was turned back on over land points (already on over sea points) in the 32km parallel and the 80 km ETAV and ETAL parallels. 27/12Z: All 80 km parallels (ETAV, ETAL, ETAY) were converted to "partial" cycling in the EDAS as was done for the 32km parallel at 12Z 10/18/97. Additionally, Joe Gerrity modified his CUCNVC code (new version = CUCNVC.DEC97) in the ETAY parallel. 28/12Z: Joe Gerrity fixed a bug in his CUCNVC.DEC97 code in the ETAY parallel. The partial EDAS cycle was restarted for this parallel. 29/12Z: The subroutine SSTBAJA was turned on in the 32km parallel because a permanent cold spot in the SST field was found in the Gulf. 30/12Z: SSTBAJA was turned off in the 32km parallel since it was warming the water in the Gulf too much (to ~30C). 30/12Z: The ETAY parallel with Joe Gerrity's modified version of CUCNVC was turned off after this cycle. ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 03/12Z: ETAY parallel was turned back on with a fixed version of Joe Gerrity's CUCNVC subroutine. 04/12Z: The ETAY parallel with Joe Gerrity's new version of CUCNVC was turned off after this cycle. 12/12Z: ETAY parallel turned back on with a test of high density GOES IR cloud drift and water vapor winds. Parallel running at 80 km with 3dvar analysis, partial cycling. ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 05/12Z: The test of GOES high density winds in the ETAY parallel was turned off after this cycle. 06/00Z: A 80-km 3DVAR parallel with cycled EDAS and assimilation of hourly observed precipitation (radar estimate only) was started in the ETAY parallel 08/12Z: EBU was modified in the 32km EMSO so that RADTN was called before the first call to CHKOUT providing radiative temperature tendencies to CHKOUT for the 00-hr output. 11/12Z: The cycled 80km EDAS w/OI analysis in the ETAL parallel was turned off after this cycle. 12/00Z: A 80-km 3DVAR parallel with cycled EDAS and Joe Gerrity's new version of CUCNVC was started in the ETAL parallel 17/12Z: Two changes were made in the 32km EMSO and the 80-km parallels: (1) TURBL and SFCDIF were modified to eliminate Td's greater than the temperature (2) STC and SMC were reordered in the bufr output 17/12Z: The 12Z 12/8/97 change was put into 80-km parallels. ----------------------------------------------------------------------- 1998 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 02/12Z: The ETAY parallel 80-km 3DVAR parallel with cycled EDAS and assimilation of hourly observed precipitation (radar estimate only) was turned off after this cycle. 07/00Z: The post was modified to correctly output cloud water on the first eta level above ground. 07/00Z: The 32-km Early Meso parallel was turned off after this cycle. ----------------------------------------------------------------------- FEBRUARY ----------------------------------------------------------------------- 06/03Z: The 10km El Nino runs began using the reduced Nest-in-the-West domain. 19/12Z: ETAY parallel reactivated (80-km) with test of new versions of CUCNVC from Joe Gerrity. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 13/12Z: Joe Gerrity's test of CUCNVC in the ETAL parallel was replaced by Ying Lin's test of the new IMSSNOW analysis in grdeta. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 02/12Z: ETAL parallel reverts back to a test of Joe Gerrity's CUCNVC with a different cuparm from the ETAY parallel. ETAV restart file used to continue the cycled soil/TKE/cloud water. 07/12Z: First successful run of the eta-10 over the SAMEX domain. Forecast is initialized from the eta-32 analysis and is run out to 12-h from 1200 UTC. 10/00Z: Test of cycled EDAS with no GOES/SSMI precipitable water data started in the ETAY parallel. ETAV restart file used to continue the cycled soil/TKE/cloud water. 15/12Z: ETAY reverts back to a CUCNVC test (with operational cuparm) for for Joe Gerrity. ETAV restart file used to continue the cycled soil/TKE/cloud water. 16/00Z: ETAL parallel changed to use operational CUCNVC but with UNIS=TRUE (sea DSP's used everywhere). ETAV restart file used to continue the cycled soil/TKE/cloud water. 27/12Z: Changes to 80km parallels: 1) GOES-PW data turned off (as done in eta-32) in ETAY, ETAL, ETAV 2) ETAY : Operational etafcst code, fully cycled EDAS, new snow analysis. 3) ETAL : ETAY plus modified CUCNVC with UNIL=.TRUE. (land profiles everywhere) Both ETAL and ETAY restarted from ETAV guess restart file. ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 01/12Z: First successful run of the mini 32-km parallel (ETAX) with the modified convection routine running in the ETAL parallel. 08/12Z: Start of real-time runs of the ETAZ parallel on cray4; containing assimilation of observed hourly precipitation from NEXRAD. Parallel archive for this test goes back to 4/30/97, from runs made by Ying Lin in retrospective mode. 19/00Z: Fix to routine easydescent.f in the r3dvar code being tested in the ETAY and ETAL parallels. Code failed in the computation of the condition number if no moisture data was present. 22/12Z: Fixed some logic errors (undefined variables) in 80 km 3dvar code in all 4 80 km parallels. 26/00Z: ETAZ parallel (ETAV plus assimilation of hourly precipitation from NEXRAD) was turned off after this cycle. 31/12Z: The 10-km SAMEX run was turned off after this cycle. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 03/03Z: The 10-km nest in the west was turned off after this cycle. 08/12Z: Changes to the 80 km parallels: 1) ETAV changed to mimic new operational eta-32 (EDAS w/full cycling, new NESDIS snow analysis) with Dave Parrish's new 3DVAR analysis. 2) ETAL = ETAV plus modified CUCNVC with UNIL=.TRUE. (land profiles everywhere) and modified SFLX to reduce surface evaporation. 3) ETAY = ETAV plus Zhao's modification to the cloud scheme (PRECPD) setting a minimum precipitation threshold per timestep of 0.01 in and with modified CUCNVC running in ETAL parallel. All three parallels were restarted from the ETAV guess restart file valid at 00Z 6/8/98. 11/12Z: CUCNVC modified in ETAY and ETAL parallel by changing EPSDN from 1.05 to 1.25, to be more consistent with the modified time scale for shallow convection. 17/12Z: Changes to 80 km parallels: 1) ETAY (precip threshold in PRECPD) turned off. 2) ETAL modified in SFCDIF by increasing the Zilitinkevich coefficient from 0.1 to 0.2. This will modify the surface exchange coefficients so surface evaporation is decreased and and ground heat flux is increased. EDAS cycle for this parallel started from ETAV restart file. 3) All 80 km parallels revert back to using the operational 3dvar analysis. 18/12Z: The mini-32, running Joe's set of CUCNVC changes, had the modification to the DSP's for clouds with bases above the freezing level put back in after it was inadvertantly dropped on 12 June. 23/00Z: ETAZ parallel on cray4 revived : ETAZ = ETAV plus assimilation of hourly precipitation data from NEXRAD. 25/00Z: Mini-32 parallel test of new cumulus convection turned off after this cycle. 25/12Z: ETAY resumes = ETAV plus changes to surface physics running in ETAL parallel. Cycle for ETAY and ETAL started from ETAV guess restart file. 26/12Z: Changes to the 80 km parallels: 1) Changes to surface physics removed from ETAL parallel, reverting back to a test of the modified convection with full cycling. 2) ETAY = ETAL with partial instead of full cycling. Cycle for ETAY and ETAL started from ETAV guess restart file. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 02/12Z: ETAY changed : Full cycling, operational CUCNVC with UNIS=TRUE (use sea profiles everywhere in the cumulus parameterization). Cycle restarted from ETAV guess restart file. 07/12Z: ETAY changed to ETAV + GOES precipitable water data. Cycle restarted from ETAV guess restart file. 13/12Z: Gross check limits in all 80 km 3dvar parallels for temperature and moisture changed from 10 deg C, 60% RH to 15 deg C, 90% RH. 14/12Z: ETAY changed to exclude GOES-9 PW data from 3dvar analysis. 17/12Z: CUCNVC modified in ETAL parallel with a change of the time scale factor for shallow convection; cycle restarted from ETAV guess. 20/00Z: ETAZ parallel (ETAV plus assimilation of NEXRAD precipitation) turned off after this cycle. 27/12Z: Modifications to CUCNVC added starting with the cycle to the ETAL parallel. 30/12Z: The test of GOES-PW data in the ETAY parallel was replaced with a test of a modified version of 3DVAR designed to improve the fit of the analysis to low-level observations. Parallel restarted from ETAV restart file. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 05/12Z: ETAY test of new 3DVAR analysis restarted from ETAV guess after repeated failures of the new 3DVAR code were (hopefully) fixed by Dave Parrish. 19/00Z: Subroutine SFLX in ETAL parallel modified to remove dependence of surface thermal diffusivity on soil moisture. When thermal diffusivity is calculated a constant SMC of 0.27 will be used. 9/10/98 update : This fix inadvertently set SMC=0.25 everywhere in the domain. 19/00Z: ETAY converted to use the Fortran-90/Y2K system, which includes: - Modified convection run in ETAL plus change to SFLX - New 3DVAR code w/previous modifications running in ETAY parallel, Bill's new conjugate gradient algorthim which will reduce the number of iterations in 3DVAR, and the ability to assimilate radiances (turned off for now). 28/12Z: Numerous bugs in the ETAY parallel were fixed by Tom Black. This cycle is the first to run cleanly with most of the major bugs fixed. ----------------------------------------------------------------------- SEPTEMBER ----------------------------------------------------------------------- 02/12Z: Last run of ETAV, ETAL, ETAY on cray3. These were turned off to facilitate a crisis global model parallel. ETAV and ETAY will be restarted on cray4 or cray5 soon. 14/12Z: ETAV (80 KM control) revived on cray4; ETAY (see 00Z 8/19/98 entry) restarted on cray5. Both EDAS cycles initiated from GDAS first guess. 23/12Z: The following changes were made to the 3dvar analysis running in the ETAY parallel: - Slight reduction (0.6 to 0.5) of the vertical correlation for moisture data; done to draw more closely for raob moisture - 50% increase in background error correlations for all variables (gives less weight to guess and more to observations) - errgeo0 reduced from 1.0 to 0.5; this will constrain height/wind corrections to be more geostrophically coupled - Maximum number of iterations increased from 60 to 90 25/00Z Observation error covariance for all satellite precipitable water data (GOES and SSM/I) increased from 5.0 g/kg to 8.0 g/kg in the ETAY parallel. ----------------------------------------------------------------------- OCTOBER ----------------------------------------------------------------------- 03/12Z: Following changes made to ETAY parallel: - Shallow convection changes (lengthed time-scale and increased EPSDN) from CUCNVC were removed. - Dependence of surface thermal diffusivity on soil moisture restored in SFLX. Cycled EDAS in ETAY was restarted from ETAV restart file. 12/12Z: Decreased scaling of background error by 10% for all variables (except moisture) in the 3DVAR analysis used in the ETAY parallel. 14/12Z: Parameter errgeo0 in ETAY 3dvar code changes from 0.5 to 0.8; this will loosen geostrophic coupling of height/wind increments and help the analysis fit the observations more closely. ----------------------------------------------------------------------- NOVEMBER ----------------------------------------------------------------------- 03/00Z: The following parameters for the 3DVAR analysis in the ETAY parallel were changed to improve performance: - pshrinkq (vertical correlation length scale for moisture) increased from 0.6 to 1.0 - bscaleq (background error variance for moisture) decreased from 0.6 to 0.5 03/12Z: The 3DVAR analysis running in ETAY was implemented into the operational eta-32 starting with this cycle. Thus, the new 3DVAR analysis and all operational F90/Y2K compliant codes were put into the ETAV parallel. The ETAY parallel remains unchanged (ETAY=ETAV + convection changes as described in 10/3/98 note above). 04/12Z: To ensure a clean test the cycled EDAS in the ETAV parallel was restarted from the ETAY cycle. 16/12Z: Unknown problems and a UNIX script error in the ETAV parallel caused 80 km parallel EDAS' to be restarted from the GDAS on 12Z 11/15/98 (ETAY) and 12Z 11/16/98 (ETAV). ----------------------------------------------------------------------- DECEMBER ----------------------------------------------------------------------- 24/00Z: CUCNVC in ETAY changed set minimum depth for convection to be (290 mb * (PSFC/1000)) (the same as ETAV) instead of 200 mb as was done previously. ----------------------------------------------------------------------- 1999 ----------------------------------------------------------------------- JANUARY ----------------------------------------------------------------------- 10/00Z: Convection test running in ETAY parallel on cray5 was turned off after this cycle due to NCO use of cray5 for end-to-end Y2K testing. 12/12Z: First successful run of 80-km cycled 4DVAR parallel in ETAZ slot; both ETAZ and control parallel (ETAV) restarted from the GDAS. Due to script errors ETAZ output files were not saved until 12Z 1/13 cycle. 27/00Z: 80-km 4DVAR parallel running in ETAZ slot was turned off after this cycle. ----------------------------------------------------------------------- MARCH ----------------------------------------------------------------------- 11/12Z: Restarted ETAV (80km control run) on cray3 13/12Z: Started test of new 3dvar code in ETAY parallel running on cray5. 31/12Z: Test of new 3dvar code in ETAY parallel ended with this cycle. ----------------------------------------------------------------------- APRIL ----------------------------------------------------------------------- 01/12Z: Test of new convection (with unified sea profiles in Betts-Miller- Janjic scheme) started with this cycle in the ETAY parallel. Modified thermal conductivty to use fixed soil moisture of 0.26 (subroutine SFLX/TDFCND). ----------------------------------------------------------------------- MAY ----------------------------------------------------------------------- 03/12Z: Both 80 km parallels (ETAV and ETAY) switched to use the new 3dvar analysis without interruption of EDAS cycle. Discovered that a script error caused the 00Z EDAS in the ETAY parallel to be started from a 12-h Eta forecast, not the previous EDAS. 24/12Z: Due to /dm problems on 80 km control parallel did not run from 00Z 5/22 - 12Z 5/23, breaking EDAS cycle. Problems with para on cray3 caused break in ETAY parallel. Therefore, EDAS cycles in both 80 km parallels were restarted from the GDAS. ----------------------------------------------------------------------- JUNE ----------------------------------------------------------------------- 16/12Z: Cycled EDAS in ETAY parallel restarted from ETAV guess due to break in ETAY cycle on cray5. Put newest 3dvar analysis (with fix of underground RH and wind superobs) into ETAV and ETAY parallels. ----------------------------------------------------------------------- JULY ----------------------------------------------------------------------- 02/12Z: Cray3 glitch caused cycled EDAS in ETAV parallel to be restarted from the GDAS. ----------------------------------------------------------------------- AUGUST ----------------------------------------------------------------------- 15/12Z: Cycled EDAS in ETAY broken due to extended cray3 outage.