Note of warning about operational soil moisture. 1) The historical backbone monthly'* soil moisture calculations are based on monthly Climate Divisions (CD) from NCDC. At the end of each month in real time, CPC itself produces a prelim CD data set. The CD from Asheville in semi-final form are not available until a few months later, and the truly final estimates after 5 or 6 months. Every month we rerun the soil model starting from about one or two years back, so as to ingest the last CD data in their respective form of finality and to keep the current soil moisture data sets as consistent as possible with the historical 1931-onward data set. 2) The real time daily updated soil moisture starts from the initial condition at the end of the last month (such as it is), and integrates forward over N days, where the N-day mean precip and temperature are the forcing fields. 3) Each month, the real time daily values change (not only because of the rain that fell yesterday), but also because of reintegration over the last 1 or 2 years. This process of repeated changes, hopefully small and converging, continues until all CD data has been cleared by NCDC as final'**. For instance the Jan 31 2000 estimate will be updated until July 2000, after which it should change no more. 4) At times we make errors. Needless to say that if you want to use our soil moisture estimates at a time after real time you should always check back whether an improved estimate is now available. * The model has a time step much less than a month, more like a day, and can generate output more often. So the name monthly' has to be qualified as it refers only to the input driving fields. ** On very rare occasions there have been retroactive changes back to the beginning in 1931, based on a new insight about instruments, observing times, the weighted average for CD based on a changing number of stations and other tedious work at NCDC. We gratefully acknowledge NCDC for their work to keep the data clean.