To those who may be interested: There has been much confusion created lately by passing around of un-informed emails regarding the GMAO OSSE work. This email sets the record straight. 1. At the GMAO, our original plan 2 years ago was to use a different RT model to simulate obs from the nature run and to assimilate them in the GSI. Imperfections in the RT model are a source of representativeness error. Using appropriately different RT models can thus add some realism to the rep. error in the OSSE. From earlier JCSDA presentations, we expected different models to be available in the CRTM framework. Although we subsequently learned that was not the case when our work began, we preferred to use different CRTM versions since it was possible some portion of appropriate rep. error would still be introduced. Since these were presumably different versions of similar underlying algorithms, however, we expected them to produce results at least as similar as different RTs used at that time. Reported comparisons of such RT models had shown rather small differences in brightness temperature (Tb). 2. Since we needed a CRTM to run outside of GSI, we asked the people in the OSSE project located in the World Weather Building who were already running the CRTM to provide an appropriate code and interface, since we were unfamiliar with the CRTM management structure. F. Weng, as a JCSDA manager, was informed of our request for assistance, and T. Zhu then passed a CRTM version and interface to us in the Fall of 2007. Since then, we have used this version to simulate obs from the nature run. 3. Examinations of our first set of OSSE validation exps. conducted last Fall suggested everything was working remarkably well. Since then we have looked more closely at the results, and we identified a 2K bias in channel 2 AMSU-A innovation statistics for obs over the oceans. We had been running our OSSE GSI without bias correction for radiance obs since we did not want to bother with spin-up of the bias correction files and we did not expect any obs bias to be introduced (The main source of bias should be due to differences in assimilating and NR models in our experiments). So, this was a puzzle. The bigger question was, did this indicate some other underlying problem or bug. 4. Since this bias was easily identifiable, we decide to determine its origin before tackling other OSSE deficiencies noticed. We have since discovered that it is directly a result of our using 2 particularly different versions of the CRTM in our GSI and OSSE obs simulation software. The GSI with this older, pre-release version of the CRTM was the only one available to us at the time we began our work. It dates back to some time in 2007. An updated GSI will not be available to us until July. This will be nearly identical to the latest GSI available at NCEP. 5. We made a test by passing the same background fields into the 2 CRTM versions. Attached (as item 11) are tables of means and standard deviations of Tb differences for AMSUA channels and AIRS channels. The averaging is over about 2000 profiles for AMSUA and 400 for AIRS, but all over only ocean points. For channel 2 AMSUA the bias is -3K and the standard deviation for channel 15 is 0.56K. For AIRS, the largest bias and standard devs. are 0.52 and 0.12, resp. 6. We performed the same test as in (5) to compare the CRTM given to us by Zhu and that in the latest GSI version from NCEP. Results are nearly identical. So anyone using a later version of GSI than the one we had been using at the GMAO will not have the kind of bias we obtained. (GSI bias correction, if used, would likely have eliminated it in any case, but the random error in some channels associated with large standard deviations of CRTM differences would not have been eliminated.) 7. For our initial version of OSSE simulated obs., we will continue to use the CRTM version Zhu gave us since it creates negligible differences with the CRTM in the latest GSI. In the future, we may replace the CRTM by some other RTM to create representativeness error as described earlier. 8. Because we have removed a big source of representativeness error in AMSUA, we will have to re-tune the parameters for our obs+representativeness error-adding software. We also will be creating a new reference run using real observations against which to validate the OSSE, since the CRTM in the earlier version apparently had some problems. Anyone else using our simulated observations in their own data assimilation experiments will have to perform their own tuning error-tuning exercise at this time, since our former tuning parameters are likely inappropriate. We will not have our own tuned values until September at the earliest. 9. Some plots (created at NCEP) of Tb produced by our OSSE obs sim. software have been passed around in emails without sufficient caveats regarding how to interpret them. It must be noted that: (a) We do not attempt to assimilate land or ice surface affected MW obs since we need better guidance on how to simulate the ERRORS in the CRTM emissivity model and associated input parameters. (This will be addressed only after we have a prototype running for the other obs.) Thus, we simply add 100K to such Tb to cause GSI QC to reject those obs, which it does. (b) We do not attempt to simulate cloud or precip affected radiances except to sufficiently identify them to the GSI QC so that it eliminates them. By this elimination we obtain a count of QC-accepted assimilated obs that is similar to the counts with real obs, with the rejections occurring in cloudy or precip regions of the nature run. At some later time, we will also try to do this better, but we note that the real issue that pertains to the present GSI is not getting the cloudy radiances realistic since GSI will reject any such radiances it detects. Rather, the current issue regards thin clouds that act insufficiently to cause rejection of the obs but create an effective representativeness error with respect to a RTM that is now only treating radiances as though they are in clear skies. (c) For the above reasons, the only appropriate comparison with real obs is to look at those obs that have passed QC. For these, aside from the bias problem we have had, the innovation statistics are similar in real and OSSE contexts, when assimilating simulated obs that have had somewhat appropriate simulated obs + rep. error added. 10. If we have a question regarding our work, we will ask the appropriate people ourselves at the appropriate time so that it is posed properly. Be wary of reports about our work not originating from us until such time as we have ourselves publicly released results. Only use versions of our software or its products either provided or directed by us with our knowledge. 11. TABLE of comparison of the Tb output from the CRTM in the GSI we had been using vs. from the CRTM from Zhu. The two routines have been passed identical atmospheric profiles over oceans taken from some GSI background file. The statistics are for the differences in Tb. Corresponding statistics (not shown) for the difference between the CRTM from Zhu and the one in the latest GSI are 0 to the precision shown, except for some visible AIRS channels. inst ch #profs mean st. dev. AMSUA 1 2056 -2.6186 0.5805 AMSUA 2 2056 -3.0825 0.3574 AMSUA 3 2056 -1.2740 0.3699 AMSUA 4 2056 -0.2130 0.1307 AMSUA 5 2056 -0.0312 0.0269 AMSUA 6 2056 -0.0013 0.0014 AMSUA 7 2056 -0.0001 0.0001 AMSUA 8 2056 0.0000 0.0000 AMSUA 9 2056 0.0000 0.0000 AMSUA 10 2056 0.0000 0.0000 AMSUA 11 2056 -0.0001 0.0000 AMSUA 12 2056 -0.0001 0.0001 AMSUA 13 2056 0.0004 0.0005 AMSUA 14 2056 0.0056 0.0022 AMSUA 15 2056 -1.7156 0.5639 AIRS 1 404 0.0040 0.0055 AIRS 2 404 0.0008 0.0030 AIRS 3 404 0.0124 0.0032 AIRS 4 404 0.0202 0.0244 AIRS 5 404 0.0128 0.0267 AIRS 6 404 -0.0002 0.0008 AIRS 7 404 0.0102 0.0301 AIRS 8 404 0.0153 0.0229 AIRS 9 404 0.0058 0.0022 AIRS 10 404 -0.0007 0.0014 AIRS 11 404 0.0016 0.0019 AIRS 12 404 0.0055 0.0106 AIRS 13 404 -0.0007 0.0015 AIRS 14 404 -0.0012 0.0040 AIRS 15 404 0.0132 0.0224 AIRS 16 404 0.0074 0.0100 AIRS 17 404 -0.0020 0.0043 AIRS 18 404 0.0061 0.0060 AIRS 19 404 -0.0092 0.0260 AIRS 20 404 -0.0013 0.0039 AIRS 21 404 -0.0015 0.0037 AIRS 22 404 0.0267 0.0266 AIRS 23 404 0.0006 0.0125 AIRS 24 404 0.0133 0.0091 AIRS 25 404 0.0206 0.0063 AIRS 26 404 0.0019 0.0106 AIRS 27 404 0.0007 0.0033 AIRS 28 404 0.0040 0.0123 AIRS 29 404 0.0041 0.0044 AIRS 30 404 0.0065 0.0017 AIRS 31 404 -0.0014 0.0022 AIRS 32 404 -0.1546 0.1033 AIRS 33 404 -0.4208 0.1474 AIRS 34 404 0.5087 0.1369 AIRS 35 404 0.4497 0.1310 AIRS 36 404 0.2856 0.1003 AIRS 37 404 0.1554 0.0649 AIRS 38 404 0.0967 0.0520 AIRS 39 404 0.0636 0.0451 AIRS 40 404 0.0436 0.0402 AIRS 41 404 0.0203 0.0124 AIRS 42 404 0.0085 0.0638 AIRS 43 404 0.0063 0.0106 AIRS 44 404 0.0119 0.0029 AIRS 45 404 0.0024 0.0034 AIRS 46 404 0.0058 0.0013 AIRS 47 404 0.0007 0.0006 AIRS 48 404 -0.0104 0.0217 AIRS 49 404 0.0066 0.0015 AIRS 50 404 0.0002 0.0008 AIRS 51 404 0.0074 0.0164 AIRS 52 404 -0.0011 0.0021 AIRS 53 404 -0.0004 0.0013 AIRS 54 404 0.0125 0.0170 AIRS 55 404 0.0002 0.0004 AIRS 56 404 -0.0008 0.0018 AIRS 57 404 0.0050 0.0013 AIRS 58 404 0.0105 0.0185 AIRS 59 404 0.0035 0.0032 AIRS 60 404 0.0081 0.0019 AIRS 61 404 0.0059 0.0036 AIRS 62 404 0.0336 0.0151 AIRS 63 404 0.0000 0.0006 AIRS 64 404 0.0052 0.0050 AIRS 65 404 0.0001 0.0004 AIRS 66 404 0.0000 0.0004 AIRS 67 404 0.0000 0.0004 AIRS 68 404 0.0016 0.0022 AIRS 69 404 0.0007 0.0172 AIRS 70 404 0.0002 0.0004 AIRS 71 404 -0.0070 0.0432 AIRS 72 404 0.0049 0.0057 AIRS 73 404 0.0035 0.0039 AIRS 74 404 -0.0124 0.0433 AIRS 75 404 0.0026 0.0039 AIRS 76 404 0.0013 0.0030 AIRS 77 404 0.0009 0.0019 AIRS 78 404 0.0002 0.0008 AIRS 79 404 -0.0144 0.0470 AIRS 80 404 0.0010 0.0021 AIRS 81 404 0.0008 0.0018 AIRS 82 404 -0.0084 0.0277 AIRS 83 404 0.0010 0.0018 AIRS 84 404 0.0007 0.0004 AIRS 85 404 0.0004 0.0002 AIRS 86 404 0.0006 0.0002 AIRS 87 404 0.0005 0.0002 AIRS 88 404 0.0049 0.0025 AIRS 89 404 0.0004 0.0001 AIRS 90 404 0.0006 0.0002 AIRS 91 404 0.0006 0.0002 AIRS 92 404 0.0002 0.0000 AIRS 93 404 0.0005 0.0006 AIRS 94 404 0.0002 0.0001 AIRS 95 404 0.0001 0.0000 AIRS 96 404 0.0003 0.0001 AIRS 97 404 0.0001 0.0000 AIRS 98 404 0.0001 0.0002 AIRS 99 404 0.0002 0.0001 AIRS 100 404 0.0001 0.0000 AIRS 101 404 0.0001 0.0000 AIRS 102 404 -0.0001 0.0002 AIRS 103 404 -0.0001 0.0002 AIRS 104 404 0.0001 0.0000 AIRS 105 404 0.0000 0.0000 AIRS 106 404 0.0001 0.0000 AIRS 107 404 -0.0001 0.0001 AIRS 108 404 0.0001 0.0000 AIRS 109 404 0.0001 0.0000 AIRS 110 404 0.0001 0.0000 AIRS 111 404 0.0010 0.0002 AIRS 112 404 0.0004 0.0001 AIRS 113 404 0.0000 0.0000 AIRS 114 404 0.0000 0.0000 AIRS 115 404 0.0002 0.0000 AIRS 116 404 0.0000 0.0000 AIRS 117 404 0.0000 0.0000 AIRS 118 404 0.0000 0.0000 AIRS 119 404 0.0000 0.0000 AIRS 120 404 0.0000 0.0000 AIRS 121 404 0.0000 0.0000 AIRS 122 404 0.0000 0.0000 AIRS 123 404 0.0000 0.0000 AIRS 124 404 0.0000 0.0000 AIRS 125 404 0.0000 0.0000 AIRS 126 404 0.0000 0.0000 AIRS 127 404 0.0000 0.0000 AIRS 128 404 0.0000 0.0000 AIRS 129 404 0.0000 0.0000 AIRS 130 404 0.0000 0.0000 AIRS 131 404 0.0000 0.0000 AIRS 132 404 0.0000 0.0000 AIRS 133 404 0.0000 0.0000 AIRS 134 404 0.0000 0.0000 AIRS 135 404 0.0000 0.0000 AIRS 136 404 0.0000 0.0000 AIRS 137 404 0.0000 0.0000 AIRS 138 404 0.0000 0.0000 AIRS 139 404 0.0000 0.0000 AIRS 140 404 0.0000 0.0000 AIRS 141 404 0.0000 0.0000 AIRS 142 404 0.0000 0.0000 AIRS 143 404 0.0000 0.0000 AIRS 144 404 0.0000 0.0000 AIRS 145 404 0.0000 0.0000 AIRS 146 404 0.0002 0.0001 AIRS 147 404 0.0000 0.0000 AIRS 148 404 0.0000 0.0000 AIRS 149 404 0.0000 0.0000 AIRS 150 404 0.0000 0.0000 AIRS 151 404 0.0000 0.0000 AIRS 152 404 0.0000 0.0000 AIRS 153 404 0.0000 0.0000 AIRS 154 404 0.0000 0.0000 AIRS 155 404 0.0000 0.0000 AIRS 156 404 0.0000 0.0000 AIRS 157 404 0.0000 0.0000 AIRS 158 404 0.0000 0.0000 AIRS 159 404 0.0000 0.0000 AIRS 160 404 0.0000 0.0000 AIRS 161 404 0.0000 0.0000 AIRS 162 404 0.0000 0.0000 AIRS 163 404 0.0000 0.0000 AIRS 164 404 0.0000 0.0000 AIRS 165 404 0.0000 0.0000 AIRS 166 404 0.0000 0.0000 AIRS 167 404 0.0000 0.0000 AIRS 168 404 0.0000 0.0000 AIRS 169 404 0.0000 0.0000 AIRS 170 404 0.0000 0.0000 AIRS 171 404 0.0000 0.0000 AIRS 172 404 0.0000 0.0000 AIRS 173 404 0.0000 0.0000 AIRS 174 404 0.0000 0.0000 AIRS 175 404 0.0000 0.0000 AIRS 176 404 0.0000 0.0000 AIRS 177 404 0.0000 0.0000 AIRS 178 404 0.0000 0.0000 AIRS 179 404 0.0000 0.0000 AIRS 180 404 0.0000 0.0000 AIRS 181 404 0.0000 0.0000 AIRS 182 404 0.0000 0.0000 AIRS 183 404 0.0000 0.0000 AIRS 184 404 0.0000 0.0000 AIRS 185 404 0.0000 0.0000 AIRS 186 404 0.0000 0.0000 AIRS 187 404 0.0000 0.0000 AIRS 188 404 0.0000 0.0000 AIRS 189 404 0.0000 0.0000 AIRS 190 404 0.0000 0.0000 AIRS 191 404 0.0000 0.0000 AIRS 192 404 0.0000 0.0000 AIRS 193 404 0.0000 0.0000 AIRS 194 404 0.0000 0.0000 AIRS 195 404 0.0000 0.0000 AIRS 196 404 0.0000 0.0000 AIRS 197 404 0.0000 0.0000 AIRS 198 404 0.0000 0.0000 AIRS 199 404 0.0000 0.0000 AIRS 200 404 0.0000 0.0000 AIRS 201 404 0.0000 0.0000 AIRS 202 404 0.0000 0.0000 AIRS 203 404 0.0000 0.0000 AIRS 204 404 0.0000 0.0000 AIRS 205 404 0.0000 0.0000 AIRS 206 404 0.0000 0.0000 AIRS 207 404 0.0000 0.0000 AIRS 208 404 0.0000 0.0000 AIRS 209 404 0.0000 0.0000 AIRS 210 404 0.0000 0.0000 AIRS 211 404 0.0000 0.0000 AIRS 212 404 0.0000 0.0000 AIRS 213 404 0.0000 0.0000 AIRS 214 404 0.0000 0.0000 AIRS 215 404 0.0000 0.0000 AIRS 216 404 0.0003 0.0001 AIRS 217 404 0.0002 0.0000 AIRS 218 404 0.0000 0.0000 AIRS 219 404 0.0001 0.0000 AIRS 220 404 0.0000 0.0000 AIRS 221 404 0.0000 0.0000 AIRS 222 404 0.0002 0.0000 AIRS 223 404 0.0002 0.0000 AIRS 224 404 0.0000 0.0000 AIRS 225 404 -0.0001 0.0003 AIRS 226 404 0.0000 0.0000 AIRS 227 404 0.0011 0.0026 AIRS 228 404 0.0026 0.0034 AIRS 229 404 0.0001 0.0001 AIRS 230 404 0.0012 0.0005 AIRS 231 404 0.0018 0.0006 AIRS 232 404 0.0069 0.0117 AIRS 233 404 0.0019 0.0355 AIRS 234 404 0.0057 0.0093 AIRS 235 404 0.0049 0.0058 AIRS 236 404 0.0060 0.0080 AIRS 237 404 -0.0007 0.0163 AIRS 238 404 0.0019 0.0022 AIRS 239 404 0.0010 0.0014 AIRS 240 404 0.0033 0.0173 AIRS 241 404 0.0003 0.0006 AIRS 242 404 0.0000 0.0002 AIRS 243 404 0.0002 0.0024 AIRS 244 404 -0.0001 0.0007 AIRS 245 404 0.0000 0.0001 AIRS 246 404 0.0001 0.0001 AIRS 247 404 0.0000 0.0000 AIRS 248 404 0.0001 0.0000 AIRS 249 404 0.0000 0.0000 AIRS 250 404 0.0000 0.0000 AIRS 251 404 0.0000 0.0000 AIRS 252 404 0.0000 0.0000 AIRS 253 404 0.0000 0.0000 AIRS 254 404 0.0000 0.0000 AIRS 255 404 0.0000 0.0000 AIRS 256 404 0.0000 0.0000 AIRS 257 404 0.0000 0.0000 AIRS 258 404 0.0000 0.0000 AIRS 259 404 0.0000 0.0000 AIRS 260 404 0.0000 0.0000 AIRS 261 404 0.0000 0.0000 AIRS 262 404 0.0000 0.0000 AIRS 263 404 0.0000 0.0000 AIRS 264 404 0.0000 0.0000 AIRS 265 404 0.0000 0.0000 AIRS 266 404 0.0000 0.0000 AIRS 267 404 0.0000 0.0000 AIRS 268 404 0.0000 0.0000 AIRS 269 404 0.0000 0.0000 AIRS 270 404 0.0000 0.0000 AIRS 271 404 0.0000 0.0000 AIRS 272 404 0.0000 0.0000 AIRS 273 404 0.0000 0.0000 AIRS 274 404 0.0000 0.0000 AIRS 275 404 0.0000 0.0000 AIRS 276 404 0.0000 0.0000 AIRS 277 404 0.0000 0.0000 AIRS 278 404 0.0000 0.0000 AIRS 279 404 0.0000 0.0000 AIRS 280 404 0.0000 0.0000 AIRS 281 404 0.0000 0.0000