Hopefully you aren't using the temp sensor included with the PICAXE kit? They don't work well. Except between 19C and 20C. After calibration they are dead-on accurate in that range.
State of Charge is not very daunting for Lead-acid batteries, but it's not quite as simple as a spot reading of voltage. You need to know the history before making that determination. There are different levels of "truth" as well.
For instance, take a full battery at rest and you just hooked up your datalogger, and the measured voltage is 25.0V or so (using the 24 volt system reference). If voltage is your logger's only source of information, then it hasn't a clue. If your logger also knows the charging current has been zero for several hours, then there's more confidence that the state of charge is close to 100%. If recently the datalogger measured a charging current of (roughly) C/80 that raised the voltage up to 28V, then this is a good sign that the batteries are fully charged, because only a float currrent (very small) was needed to maintain float voltage. This requires certain knowledge of the battery bank, its capacity, required charging set points (gell-cells use 28.2 while flooded batteries should see more than 29), current and voltage, AND a record of the progress over time. Hope you're not trying to cram all that info into a 08M micro chip because it doesn't have enough pins.
Even better than that is a constant measure of AH in and AH out of the battery, with periodic status checks to correct the projected SOC no matter how much load and charge are happening. I can keep making this better and better, but at some point it is good enough. The more "correct" you want the meter to be, the more work and cost it requires.
This is a funny coincidence Simen. I've been working on an enhanced datalogger / controller for my solar/wind system for some time now. I'm using a PICAXE 28x2 and I'm using up pins fast. I could clean up the drawing and show you a copy without much trouble, if you are interested.