Update 6 Oct 2014 ********************** After 3 years of testing many different ways to solve uncertainties with the MSU/AMSU data we are getting close to releasing v6.0. This version will have completely new limb-corrections, diurnal corrections, and hot-target corrections. As noted before, we did not apply any diurnal corrections to the AMSUs (starting in 1998) due to the fact AQUA and MetOP were non-drifting satellites. However, NOAA-15 was a key link in the chain and as an A.M. orbiter it backed into "warmer" parts of the diurnal cycle, and thus contributed to a slight spurious warming (noticeable mainly in LT). This is one factor that is improved. Another key factor is that we now use a multichannel representation of MSU 2 from the AMSUs so that rather than assuming AMSU 5 equals MSU 2 (which is not strictly true) we now have a representation of MSU 2 from AMSU 5,7 and 9). We hope to be out with this as a beta version by the end of 2014. Update 26 Jun 2013 *********************** We now have data from METOP (European AMSU) and the latest NOAA polar oriter NOAA-19. We have added these data into the mix without changing the other merging procedures. Because some of the anomalies in the last four years change by up to 0.04 in LT and MT (more in LS) we've decided to up-version the filenames to v5.6. However, due to some language in a small contract we have with NOAA, we will continue to produce v5.5 without METOP and NOAA-19. So when you see both versions being updated, that is the reason. Though the differences are small (trend in LT is +0.138 in both), we will be displaying v5.6 on our releases and blogposts. The basic change in the grids is that land trends are slightly higher and ocean trends are slightly lower in v5.6 that would be attributed to the over-reliance on NOAA-18 in v5.5 which is drifting in that manner. Version 6.0 is still being developed as we had literally gone back to the drawing board on every processing step. Update 5 Oct 2012 ************************* As noted in the 1 Dec 2011 entry below, the increasing noise in AQUA continued to introduce a bias to the time series. In Sep 2012 the bias was clearly quickly growing as the difference between AQUA and the average of NOAA-15 and NOAA-18 was 0.2 K. This event was obviously beyond our control. As a result we have changed the satellites which provide data for the time series since 2009 for which mainly AQUA provided in v5.4. In new version 5.5 we shall use the average of NOAA-15 and -18 for the time series. Both of these spacecraft have experienced the effects of diurnal drift: NOAA-15 warming slightly and NOAA-18 cooling slightly. By utilizing the average, the errors largely cancel. The difference between the two spacecraft is less than 0.1 K, so this is much better that the difference between them and AQUA. This affects the two products that depend on AMSU channel 5, LT and MT. AMSU channel 9 used in LS is unaffected by the noise. The net effect of this change is minuscule on the trend - from +0.137 to +0.138 K/decade for LT and +0.047 to +0.043 for MT. So updating the list on Update 8 Nov 2011 we have Spacecraft with AMSU used in UAH products as of Sep 2012 v5.5 START TERMINATE NOAA-15 1998/215 Present NOAA-16 2001/032 2003/031 (sensor drift problems) AQUA 2002/221 2009/365 (significant noise) NOAA-18 2005/152 Present Update 1 Dec 2011 ************************* Roy and Danny are looking at a new, more robust method of calculating (empirically) the diurnal drift effect on the AMSU instruments as noted last month. In very preliminary results, it appears that the AMSU on AQUA may have a slight spurious warming in AMSU5 over the last 3 years. Since this AMSU is our backbone satellite since 2002, we have intercalibrated the other satellites to it, which means when all is said and done, we may be lowering the anomalies since 2008. However, there are still many uncertainties in the process, but I wanted to give an update on the preliminary findings. Update 8 Nov 2011 ************************* Roy Spencer and Danny Braswell are working on a diurnal correction for the AMSU channels. To date we have relied on AQUA AMSU due to the fact it was a NASA science spacecraft with on-board propulsion and thus a stable (non-drifting) orbit. AQUA will not last forever, and there are signs of increasing noise, so it will not be suitable as an anchor satellite much longer. The diurnal corrections will allow more data from NOAA-15, -16 and -18 to be utilized. At this point we terminate their use when a drift in temperature is detected: (YEAR/DAY) Spacecraft with AMSU used in UAH products as of Oct 2011 v5.4 START TERMINATE NOAA-15 1998/215 2007/365 NOAA-16 2001/032 2004/233 AQUA 2002/221 Present NOAA-18 2005/152 2010/182 We are also looking to bring in the AMSU data from METOP and NOAA-19. When the testing is completed we will be issuing version 6.0 of the temperature products. We do not know when such testing will be completed. Update 8 Dec 2010 ************************* Preliminary runs show that the new mean annual cycle will be about 0.1 C warmer each month for the global averages, meaning all monthly anomalies will appear to decrease by about 0.1 when the new 30-year base period is used (see below). Here are the preliminary differences (old annual cycle v5.3 minus new annual cycle (v5.4).) The final anomalies will be out next month when Dec is calculated. TLT annual cycle v5.3 minus v5.4 MON GL NH SH TRPCS 1 -0.106 -0.185 -0.026 -0.047 2 -0.093 -0.167 -0.019 -0.033 3 -0.099 -0.186 -0.013 -0.006 4 -0.102 -0.193 -0.010 0.000 5 -0.079 -0.134 -0.026 -0.003 6 -0.052 -0.067 -0.036 0.008 7 -0.070 -0.077 -0.063 -0.050 8 -0.069 -0.096 -0.043 -0.043 9 -0.125 -0.146 -0.104 -0.049 10 -0.119 -0.113 -0.127 -0.050 11 -0.109 -0.139 -0.079 -0.048 12 -0.075 -0.129 -0.021 -0.068 So, for example, here is the old set of 2009 anomalies (v5.3) and the new set (v5.4) v5.3 2009 1 0.251 0.472 0.030 -0.068 2009 2 0.247 0.565 -0.071 -0.045 2009 3 0.191 0.324 0.058 -0.159 2009 4 0.162 0.315 0.008 0.012 2009 5 0.139 0.161 0.118 -0.059 2009 6 0.041 -0.021 0.103 0.105 2009 7 0.429 0.190 0.668 0.506 2009 8 0.242 0.236 0.248 0.406 2009 9 0.505 0.597 0.413 0.594 2009 10 0.362 0.332 0.393 0.383 2009 11 0.498 0.453 0.543 0.479 2009 12 0.284 0.358 0.211 0.506 v5.4 2009 1 0.145 0.287 0.004 -0.115 2009 2 0.154 0.398 -0.090 -0.078 2009 3 0.092 0.138 0.045 -0.165 2009 4 0.060 0.122 -0.002 0.012 2009 5 0.060 0.027 0.092 -0.062 2009 6 -0.011 -0.088 0.067 0.113 2009 7 0.359 0.113 0.605 0.456 2009 8 0.173 0.140 0.205 0.363 2009 9 0.380 0.451 0.309 0.545 2009 10 0.243 0.219 0.266 0.333 2009 11 0.389 0.314 0.464 0.431 2009 12 0.209 0.229 0.190 0.438 Remember, Dec 2010 has yet to be included, so there will be some slight changes when the final version 5.4 is generated. Update 13 Sep 2010 ************************* As an alert, we will be generating anomalies when the December data have been processed to be based on the 30-year mean annual cycle of 1981-2010 to match the 30-year normal time frame of many meteorology anomalies. This will replace the older reference annual cycle of 20-years (1979-1998). Update 13 Apr 2010 2200 CDT ************************ The addition of NOAA-18 has now been corrected. The problem was a mismatch in the overlap dates of NOAA-16 and NOAA-18 (there was no overlap of "good" data). For the gridded files, which are basically the patterns of the anomalies, NOAA-18 is merged directly with NOAA-15, and this has been coded accordingly. Corrected v5.3 files will be on the data site on 14 April. Update 13 Apr 2010 ********************************* The addition of NOAA-18 on the gridded monthly anomalies has created a sudden divergence between land and ocean temperatures beginning in 2005 (when NOAA-18 began) in v 5.3. I will update v5.3 through March 2010 without NOAA-18 and place it on the website. There is likely some error in the merging of NOAA-18 that creates this rather spurious redistribution. The version 5.3 files without NOAA-18 are appended with 5.3a, i.e. tltmonamg.2000_5.3a and for the sections as uahncdc.lt53a. Update 7 Apr 2010 *********************************** With March 2010 we have now included AMSU data from NOAA-18 beginning in June 2005 to the present. This is an operational satellite in the afternoon (about 2 p.m.) orbit, so it will experience drifting into cooler diurnal temps as time goes on. At this point, the equatorial crossing time is only about 30 minutes from the nominal time, so has very little impact. AQUA, with its stationary crossing time is still the backbone satellite for this period. You will notice some slight changes in anomalies as NOAA-18 is added into the mix. In particular, the gridded maps that had depended only on NOAA-15 patterns will now be a bit smoother as another satellite is added. Update 5 Mar 2010 *********************************** Since the discussion in Jul 09 below points out that the largest differences between UAH and other dataset regarding specific monthly trends occurs in Feb for v5.2, we are instituting the adjustment discussed last July and calling this v5.3. In essence, the mean annual cycle of the AMSUs (NOAA-15, NOAA-16 and AQUA) will not include the accumulated annual cycle anomalies determined for the MSUs (through NOAA-14). Evidently this accumulation of annual cycle anomalies created a spurious rather than corrected annual cycle by the time the AMSUs were on orbit. For Dec78 to Feb10, the trends of both v5.2 and v5.3 are +0.132 C/decade, so the overall trend is not affected by this readjustment of the mean annual cycle for the AMSUs. If anyone wishes to see the document prepared last Jul, please email me at christy@nsstc.uah.edu. Update 18 Jul 2009 ************************************ Corrected trend values (1700 CST) It was brought to my attention by Anthony Watts that there has been some discussion about the noticeable annual cycle in the LT and MT trends when done by months. In other words, the trend for Februaries is on the order of 0.12 C/decade warmer than the trend for Mays. Other data sets don't have such a large range in trends when calculated by months, RSS for example has a range of 0.05 C/decade. (Note, this issue doesn't affect the overall trend.) The feature arises when the AMSU data are adjusted and merged into the MSU data stream beginning with NOAA-15 in Aug 1998, then carries forward with NOAA-16 and AQUA (both of which are AMSUs too). The process involves at one point the removal of a mean annual cycle in the anomaly differences from one satellite to another. It turns out that all satellites have a residual annual cycle due to each instrument's peculiarities. In the end, all annual cycles are matched to NOAA-6 and NOAA-7. Detecting the impact of this peculiarity is difficult. For example, it is not seen when gridded data are directly compared against radiosondes (see Christy and Norris 2006 and 2009.) However, an annual cycle in the difference time series is clear in RSS data when compared with balloons (see Fig. 2 of both papers.) I've tested a number of alternate processing methods (basically versions of not removing the annual cycle in the difference time series from the first AMSU onward) and the range from the highest to lowest is reduced to just under 0.09 C/decade. This in effect establishes a new annual cycle for the AMSUs based on the first AMSU. I think the magnitude of the annual cycle in the monthly trends is a legitimate problem to address. The range in the current v5.2 LT looks too large (about 0.12 C/decade). However, one should expect differences from month to month, especially when ENSOs and a volcano have different impacts by months so so the range shouldn't be zero. I'll keep looking into this and if a reasonable result is produced, I'll rename the dataset v5.3. The important point in all of this is that the overall global trend of the entire time series ranges insignificantly from +0.123 to +0.125 C/decade even under the different merging methods used to date. This is because the removal of the annual cycle of differences from satellite to satellite does not add any bias to the time series, so the overall trend doesn't change. Update 6 Jul 2009 ******************************* It is likely that the grids for Dec 1978 to May 1979 for LS and MT files (not LT files) were over written by wrong values during the past six months. Please reaccess these grids.(i.e. tmtmonamg.YYYY_5.1, tlsmonamg.YYYY_5.1). The uahncdc.mt and uahncdc.ls files would also be affected. The tlsglhmam_5.1, tmtglhmam_5.1 files were not affected. Again, the LT files were not affected. Update 17 April 2009 *********************************** A new format for the daily files has been created for easier reading and to add the tropical anomaly (20S-20N). This will be used beginning with the release of the April data in May 2009. tltday_5.2 New format is (2I5,I6,A5,4I5) for IYR,IDY,IDY78,ACHA,MGL,MSH,MNH,MTR Where IDY is the day of the year and IDY78 is the number of the day since 1 Jan 1978. MTR is the tropical anomaly. This will also be true for the first line of the daily zonal anomalies, tltdayamz_5.2. Update 3 Jan 2008 We now have data from AQUA added to the time series beginning with day 221 of 2002. AQUA is a spacecraft with on-board propulsion and thus has stable station-keeping. Thus, AQUA's AMSU will not be subject to diurnal temperature drifts. Upon comparison with NOAA-15's AMSU, we find only minor differences for their 5+ year overlap, with NOAA-15 being slightly warmer near the end of the time series for LT and MT. The error values for NOAA-15 are much smaller than what we indicated below. At this time we are merging AQUA into the time series while keeping NOAA-15, with its slight, spurious warming, in the mix through the present. There will be some slight month-to-month differences versus the version from last month, but these are very small so we will keep the version numbers as they are since nothing has changed except the merging of AQUA into the time series. We note that 1-11 Dec 2007 are missing in AQUA and are hoping these will be recovered. We do have all daily values from NOAA-15 for December. Update 19 Dec 2007 ***************************** As noted below, the diurnal drift of NOAA-15 is becoming more obvious. We are still working on a correction scheme to remove this spurious warming effect. Through comparison with other AMSUs, the warming drift by Oct and Nov is on the order of 0.2 C (i.e. the values of v5.2 are too warm for TLT by that amount.) This has been a long ordeal because we want to create a correction that will stand the test of time. Update 10 Sep 2007 ***************************** Note that NOAA-15 is drifting backward into a warmer part of the diurnal cycle which will induce a spurious warming in the values for the last couple of years in LT and MT (see 12 July 2007). Information from Carl Mears indicates the spurious warming in our LT and MT global anomalies will be about +0.08 and +0.04 respectively by mid 2007 (for the anomalies). Because we have not solved all of the peculiarities with NOAA-15 (especially channel 6) our intent to generate a multi-channel AMSU replacement for old MSU LT and MT is still on hold. Be advised that the LT and MT products are a little too warm in the past couple of years, but less than 0.1K. Update 23 Aug 2007 ***************************** A new operating system is being adjusted to right now. Our trends are not showing up on the output. Hopefully this will be remedied by next month Update 12 July 2007 ***************************** We are still relying on NOAA-15 for the current MT and LT products. We have not instituted a diurnal correction on these, so they are likely running a little on the warm side as NOAA-15 is "backing" into a warmer period of the diurnal cycle. Regarding the long term trend, this will have a very small impact. When the new MSU diurnal corrections as well as the AMSU diurnal corrections are applied, we suspect the resulting trend will be almost identical to the current trend, though there will be some changes in the interannual variations. Update 15 Dec 2006 ****************************** Due to a dumb mistake, the values for MT were in error when loaded up for the period ending Nov 2006. Rather than eliminating NOAA-16 data (the bad satellite) I had eliminated NOAA-15 (the good satellite) after Sept 2005. So, the values for MT have all been rerun and replaced. There are slight changes throughout the time series since the mean annual cycle was affected. I've also replaced all of LT to make sure they were consistent. Update 5 Dec 2006 ******************************* Data products are still 5.2 and 5.1. For LT 5.2 and MT 5.1 we have eliminated the data from NOAA-16 after September 2005 when NOAA-16 began to diverge in a manner that suggested NOAA-16 was having problems. Thus, the data since Oct 2005 is based on NOAA-15. The net effect on this change was to increase post-Oct 2005 temperatures slightly, and thus the global trend is increased by about 0.01 C/decade. Update 10 Nov 2006 ******************************* Notice that data products are back to version 5.2 for LT and 5.1 for MT and LS. We had hoped to solve the inconsistencies between NOAA-15 and NOAA-16 by this time, but we are still working on the problem. The temperature data for LT and MT are diverging, and we had originally thought that the main error lay with NOAA-15. However, after looking closely, there is evidence that both satellites have calibration drifts. We will assume, therefore, that the best guess is simply the average of the two. This is what is represented in LT 5.2, MT 5.1 and LS 5.1. These datasets have had error statistics already published, so we shall stick with these datasets for a few more months until we get to the bottom of the calibration drifts in the AMSUs. However, the error statistics only cover ther period 1978 - 2004. The last two years cover the period where the two AMSUs are drifting apart, so caution is urged on the most recent data. Update 12 Oct 2006 ****************************** This is a note just to emphasize that the current products will be changing and conclusions should not be made on versions appended with "p". The older version 5.2 has been analyzed in the literature with published levels of confidence. There are no confidence levels for any of the "p" versions, which means confidence should be assumed to be poor until published magnitudes are available. Update 6 Oct 2006 ****************************** Another month and the same story, we are still finalizing the production of version 6.0. The values which will be placed on the website contain the earlier v5.2 data through Dec 2004 with the preliminary values from v6.0 from Jan 2005 to the present (Sep 2006). The largest changes occur in the last two years as NOAA-15 was drifting into warmer temperatures and its target temperature effect became evident. Again, this is a preliminary dataset designed only to give an idea of what is going on at the moment. We will update the daily files when the full v6.0 is ready. The monthly gridded files and the uahncdc files also reflect the use of the newer data in 2005 and 2006 only. Update 9 Sept 2006 ***************************** We are nearing completion of the new diurnal corrections and the conversion of AMSU data to mimic MSU data (rather than substituting AMSU5 for MSU2, as in v5.2). We will likely use a statistical combination of AMSU4-9 to generate a more realistic MSU2 from which LT and MT are derived. This will make the time series more consistent. The diurnal drift and hot target effects of NOAA-15 and NOAA-16 render the recent months of v5.2 too warm since we haven't adjusted for those effects. As a quick solution, we are subsituting a preliminary version of LT (v6.0p) for Jan - Aug 2006 for which these adjustments have been applied. We caution that there are still likely to be some changes to v6.0 when it is released in a month or two, but this seemed the best path to take given the growing errors especially in NOAA-15 LT. The remaining months prior to Jan 2006 will be v5.2 as before, so we will still label those months as v5.2. When we have completed v6.0, we will relabel all of the datasets accordingly. Update 9 May 2006 **************************** We are continuting to work on the diurnal correction, so the current AMSU data have not been corrected for diurnal drift. We note that experience indicates that as NOAA-15 is backing up, it is moving into warmer temperatures, while NOAA-16 is advancing and moving into cooler temperatures. The net impact of these drifts could be very slight as they tend to compensate one another. [Note the sign of the drift error is opposite that reported below in 20 Aug 2004 due to the discovery of the diurnal correction error being applied at that time in the tropics.] Update 6 Apr 2006 **************************** Roy is working on a diurnal adjustment for the AMSU instruments as they have now drifted over an hour from their initial crossing time. NOAA-15 has backed up from 7:30 to 5:48 and NOAA-16 has drifted forward from 1:54 to 3:10. Be on the lookout for a new version that will have these additional adjustments. There is also a divergence between NOAA-15 and NOAA-16 that has developed in the last 12 months. We don't know if N15 is spuriously warming or N16 is spuriously cooling. As soon as this is resolved, we hope to include this correction in the next version as well. Update 4 Jan 2006 **************************** For those who access the files with section anomalies (uahncdc.lt) note that an error was found in the calculation of the USA48 list. Rather than set to missing, the surrounding grids were set to zero, so many zero grid values were included in the USA48 average, muting the real anomalies. The correct values are now available. None of the other sections were affected. Update 7 Aug 2005 **************************** An artifact of the diurnal correction applied to LT has been discovered by Carl Mears and Frank Wentz (Remote Sensing Systems). This artifact contributed an error term in certain types of diurnal cycles, most noteably in the tropics. We have applied a new diurnal correction based on 3 AMSU instruments and call the dataset v5.2. This artifact does not appear in MT or LS. The new global trend from Dec 1978 to July 2005 is +0.123 C/decade, or +0.035 C/decade warmer than v5.1. This particular error is within the published margin of error for LT of +/- 0.05 C/decade (Christy et al. 2003). We thank Carl and Frank for digging into our procedure and discovering this error. All radiosonde comparisons have been rerun and the agreement is still exceptionally good. There was virtually no impact of this error outside of the tropics. Update 20 Aug 2004 ***************************** NOAA-15 and NOAA-16, the first two spacecraft carrying the AMSU instrument, are due for diurnal corrections. The global mean intersatellite drift error is 0.016 C/year for MT for the period of their overlap (Feb 2001 - present). In general, the newer spacecraft drift less than the earlier ones, so corrections should not be substantial and the net effect may be a wash. NOAA-15 is drifting earlier, which introduces a spurious cooling trend in LT, while NOAA-16 is drifting later in the diurnal clock which introduces a spurious warming in LT. The opposite occurs for both MT and LS. Sometime in the next few months we shall apply the new diurnal corrections, likely changing the file names to 5.2. We are also experimenting with a more direct method of calculating the brightness temperatures which avoids the NESDIS nonlinear adjustments altogether. Preliminary results suggest trends could be very slightly warmer, but less than 0.02 C/decade different. Update 5 Feb 2004 ****************************** We have now updated the ephemeris corrections from their estimated values to actual values for 2003. There will be a very slight increase in 2003 temperatures to account for orbital decay. The net effect on the time series of TLT is to increase the trend by +0.002 C/decade. Update 7 Mar 2003 ***************************** We have made some changes to the data processing that were quite minor. Even so, we decided to change the version number to 5.1 from 5.0. These changes will not affect scientific results for those of you in the process of publishing work from version 5.0. For all three products we have strengthened the requirement a bit for acceptable data to entire into the routine that calculates the intersatellite biases. This resulted in a very slightly more negative trend in LT by 0.004 c/decade and for MT by about 0.003 C/decade. In addition as noted in the 10 Jan 03 entry, we have updated the Target Temperature coefficients since 2002 added some MSU data from NOAA-14. The update only affected LS (T4). One of the coefficients barely exceeded the 40% explained variance threshhold this time (NOAA-11), so it was employed in the processing. This helped reduce the daily error variance and the difference in trends between NOAA10 v. 11 and NOAA11 v. 12. The net effect on the trend was about 0.02 C/decade (more positive) We thank John Bates of NCDC for finding earlier MSU data from TIROS-N. As a result, version 5.1 daily data now begin at 16 Nov 1978 and the monthly grids now begin with Dec 1978. Update 10 Jan 2003 ***************************** Roy Spencer has updated the ephemeris corrections which had been estimated for the past 9 months. Changes are minor with a few monthly global averages a few hundredths C warmer in 2002 than as shown last month for LT. With the completion of 2002, we will be recalculating the target coefficients and if the changes impact the trends much, we may up the version number to 5.1 from 5.0. This should be complete with the calculation of the January 2003 values. Coming in 2003 - diurnal drift corrections for the AMSUs on NOAA-15 and NOAA-16. There should be little impact. We will also be merging data from NOAA-17 this year. Also, the paper describing Version 5.0 of the microwave data will be appearing in J. Atmos. Oceanic Tech. this year. Christy, J.R., R.W. Spencer, W.B. Norris, W.D. Braswell and D.E. Parker, 2003: Error estimates of version 5.0 of MSU/AMSU bulk atmospheric temperatures. 2003. J. Atmos. Oc. Tech. Update 14 Aug 2002 ********************* I've noticed a noise problem with the gridded datasets in which the AMSU data, because of its much finer resolution is not being properly smoothed to match the old MSU spatial resolution. This creates spurious east-west oscillations of anomalies primarily from 40S to 40N where the stripping is most evident (see Christy et al. 1998 for complete discussion of this sampling problem.) The scale of this problem is wave number 14 in the east-west dimension due to the fact there are 14 orbits per day per satellite. I'll be testing some better longitudinal downscale averaging techniques in the next few months to try and eliminate this feature in the anomalies. At present, I will likely smooth the anomalies of NOAA-16's AMSU to wave number 12 to greatly reduce the wave number 14 noise. The goal here is to reproduce an MSU-like spatial resolution from the AMSU high resolution data. This does not effect the global or even zonal average anomalies. Update 9 May 2002 ********************* We are now using the non-linear correction for the diuranal drift of the satellites. The version will now have appended "_5.0" to indicate this. Note that the name of the file begins "tlt" rather than "t2lt" because channel 2 is not used from the new AMSU instruments, so no need to use the "2". "tlt" means "temperature lower troposphere". Update 8 April 2002 ********************** Roy Spencer and I are in the process of upgrading the MSU/AMSU data processing to include a new non-linear approximation of the diurnal cycle correction (currently the approximation is linear). In preliminary results, the effect is very small, well within the estimated 95% C.I. of +/- 0.06 C/decade. In the products released today, some minor changes have been included (though not the new non-linear diurnal adjustment). The 2LT trend is +0.053 C/decade through Mar 2002. The difference in today's release vs. last month's is a slight warming of monthly data after 1998. Essentially, this release corrects an error in the linear diurnal adjustment and produces better agreement between the MSU on NOAA-14 and the AMSU on NOAA-15. The single largest global anomaly impact is a relative increase of +0.041 (April 2001) while most are within 0.02 of the previous values. The net change in the overall trend was toward a more positive value by +0.012 C/decade. Again, this is still an interim change, and we anticipate a final version ("E" or "5.0") next month. The current release will be designated t2ltmonamg.YYYYdx. Update 8 Jan 2002 ******************* Revised values for the correction of orbital decay for NOAA-15 (Aug 1998) and NOAA-16 (Feb 2001) has been applied to the AMSU data. These corrections are based on a more analytical approach for the cross-scan biases that arise as the spacecraft falls in altitude. The net effect is minor, being a slight cooling of the early values and a warming of the more recent values observed by the AMSUs. The overall 23-year trend is essentially unchanged. Update 2 Nov 2001******************* The comparison between NOAA-14 and NOAA-15 had been superb up until about 21 Aug 2001. At that time NOAA-14 showed significant cooling relative to both NOAA-15 and NOAA-16. This is likely due to the diurnal drift of the spacecraft as it is now over 6.5 years old. The diurnal corrections are typically useful until the spacecraft drifts significantly (NOAA-14 is a p.m. bird, starting with a 2 p.m. orbit that is now in late afternoon.) I've determined that NOAA-14 for 2LT is likely in a situation where the diurnal corrections as well as instrument body temperature effects are not well characterized for this new solar shadowing regime. So, for now, I am eliminating the post 21 Aug 2001 NOAA-14 data in 2LT, relying on the excellent agreement between NOAA-15 and NOAA-16. As a result, the values for global anomalies for Aug and Sep (and now Oct) will be slightly warmer than previously reported (about 0.06 C). The trend has become more positive by only 0.002 C/decade as a result of this termination. We are investigating a fix for NOAA-14 so that the data may be used again. Please be advised that as the data from NOAA-16 are merged into the datastream there will be slight adjustments to the recent anomalies as we wait until a resolvable annual cycle is available. NOAA-16 has provided data since 1 Feb 2001. Update 24 Aug 2001 ********************* I've discovered a Y2K error in the program which reads the diurnal corrections. The corrections for NOAA-14 were not applied after 1999. These will be applied when the August data are processed. Preliminary checks indicate the impact is less than 0.01 C/decade. Because the diurnal corrections were not completely applied, I recomputed the PRT coefficients to adjust for heating of the instrument. This impact is between .001 and .003 C/decade on the full trend, so it is tiny. The net effect of these corrections over the current dataset is to give a slight warming to MSU2 and a slight cooling to MSU2LT. As of July 2001, the trends in MSU2 and MSU2LT are virtually identical. This seems odd because the overall stratospheric trend is quite negative. However, the stratospheric component of MSU2 has "seen" a warming due to Mt. Pinatubo and a gradual warming since June 1996 in the stratosphere. Thus, relative to MSU2LT, MSU2 has seen more warming since 1996 and now expresses the same trend since 1979 as MSU2LT. This is not an artifact of NOAA-14, but is also present in the AMSUs on NOAA-15 and NOAA-16. Update 9 May 2001 ********************** NOAA-14, with the last MSU, will soon be decommissioned. We will soon be phasing in data from NOAA-16. Through April, the dataset has been built from the MSU on NOAA-14 and the AMSU on NOAA-15. During the next several months, as we phase-in NOAA-16, there will likely be a little instability in the anomalies as it takes many months to accurately determine the bias between NOAA-16 and older satellites. Thus there may be changes from one month to the next for anomalies calculated for previous months. Update 19 Jan 2001 ********************* A new algorithm for the NOAA-15 AMSU has been applied to match the NOAA-14 MSU2LT product. The new multi-angle method reduces the intersatellite noise from the previous method by 10 percent. The bias correction now is only 0.126 vs. 1.084. A few of the monthly values are changed by 0.01 C or so. The overall 79-01 trend under the new algorithm is identical to the old (0.044 C/decade). Update 6 Oct 2000 ********************** Orbital decay values for NOAA-14 have been updated for 2000. This effect is very minor - a couple of hundredths of a degree on a monthly, global average. The global trend warmed up by +0.002 C/decade as a result. Update 7 Mar 2000 ********************** We have now included the NOAA-15 AMSU data into the 2LT time series. After several experiments with multi-channel representations of the old MSU 2LT, we concluded that none quite reproduced the old temperatures. So, we are using a multi-angle retrieval from AMSU channel 5 which reproduces the multi-angle 2LT from MSU very closely. So, data in the 2LT time series since September 1998 now include AMSU. Update 4 Feb 2000 *********************** We have made a slight change to the t2lt files. The orbit decay correction had been determined as a function of the daily average orbit altitude. We have recalculated the orbit decay effect on a scan-by-scan basis (i.e. over 3000 times per day) to account for the alititude change within a particular orbit. The time series is virtually identical to the earlier version, with the trend becoming more positive by 0.013 C/decade. We shall have one more alteration to the t2lt time series in the next few weeks - that of merging NOAA-15 (AMSU) into the time series. T4 and T2 already include NOAA-15. (See below for earlier discussion). We are closer to a multi-channel solution now and hope to include that soon for 2lt. Update 27 Dec 1999 ************************ The daily zonal files have now been made Y2K compatible with the value for year as four-digits. I am still working on the grid point files, some of which have four digits, and some which don't. This is a bit more tricky than it seems. NOAA-15 became operational in September 1998 carrying the new Advanced Microwave Sounding Unit. This instrument has 16 channels, 12 more than the MSU of TIROS-N, NOAA-6, 7, 8, 9, 10, 11 ,12 and 14. The channel frequencies are not exactly matched to the MSU, but we are working on merging the new data into the MSU stream. For T2 (MSU channel 2) we are using AMSU channel 5. AMSU5 peaks slightly lower in altitude than MSU2 and has a mean bias of 2.55 K (AMSU5 warmer than MSU2). AMSU5 global mean annual cycle is about 0.15 K greater than MSU2. In terms of anomalies, this should not be a problem, and the intercomparison of MSU to AMSU anomalies looks as good as any other MSU to MSU comparison. Therefore, the T2 anomalies now have AMSU5 data merged. We shall continue to call this product T2 in honor of the MSU heritage. For T4 (MSU channel 4) we are using AMSU channel 9. AMSU9 has a slightly different weighting function than MSU4, but the global mean bias is only 0.02 K warmer than MSU4 but still has a global mean annual cycle which is 0.1 K in amplitude greater than MSU4. The MSU4 to AMSU9 anomaly comparison is excellent. Therefore, the T4 anomalies now have AMSU9 data merged since September 1998. We shall continue to call this product T4 in honor of the MSU heritage. We are still working on an AMSU retrieval which matches MSU 2LT. At present, we have generated a retrieval whose weighting function is similar to 2LT using AMSU channels 4, 5, 6, and 7. There is a global mean bias of about 1.3 K, but the distribution of the bias is non-uniform, being much greater near the poles (about 5 K) and less in the tropics (about 0.6 K). At present, we have not merged the AMSU data into the T2LT time series until a better match is determined. Update 19 Nov 1999 *********************** Data are now available for version D in both zonal and gridpoint. The NOAA-12 non-linear coefficients have been applied to both zonal and gridpoint files. The description of version D is contained in Christy, J.R., R.W. Spencer and W.D. Braswell, 2000: MSU tropospheric temperatures: Dataset construction and radiosonde comparisons. J. Atmos. Oceanic Tech. (in press). The gridded files will now be in this directory as the corrections have now all been applied and the version extension is now simply "d". Update 13 July 1999*********************** Recent research has shown that the time series of t2lt is affected by orbit decay. This effect causes artificial cooling on the time series when examined in isolation. (Wentz and Shabel 1998) There are offsetting warming effects related to the time-drifting of the spacecraft due to east-west drift. The new designation for MSU 2R (lower troposphere) will be t2lt representing a channel 2 retrieval of the lower troposphere (hence t2lt). The third character in t2lt is an "l" as in "lower", not a one. Note that the base period for the mean annual cycle for t2lt is now 1979-1998, or 20 years instead of the previous 1982-1991 ten years. John Christy. christy@nsstc.uah.edu