Why is it?

Discussion in 'Alpine & Southern' started by Vermillion, Jan 6, 2007.

  1. Vermillion

    Vermillion Pool Room Ski Pass: Gold

    Joined:
    Nov 13, 2005
    Messages:
    71,490
    Likes Received:
    25,929
    Location:
    Cranhole, Melbourne, VIC
    That all the models over-estimate moisture content with approaching systems. Ok this does not happen exclusively, but on the whole the vast majority of systems that are progged will over-estimate the moisture/rainfall content. Why is this? Surely if they kept on over-estimating it then that would be added to their database and they would 'learn' from it and the model would predict a rainfall amount that is closer to the real amount in the future. I mean, when was the last time u saw a little system pop by and BANG 30mm+ comes out of it when you were expecting less than 5mm? Not since ive been forecasting. On the other hand how many times in the past months have we seen a system destined to drop 30mm+ and end up with a bee's fart worth of rainfall? Too many times. If data was continously added to the model then shouldnt it realise that its getting it wrong, and from its catalogue of past systems use this data to predict systems more accurately?

    What im really getting at here is why do models consistently over-estimate the amount of rainfall/moisture instead of under-estimating it?
     
  2. BlueHue

    BlueHue One of Us

    Joined:
    Apr 17, 2003
    Messages:
    2,015
    Likes Received:
    575
    Location:
    Sydney
    Just guessing but maybe a limitation in the data input to the models.

    I would assume that a fair amount of data is required from the oceans surrounding Australia, since they have such a big influence on our weather. If some errors occuring in the models require changes in input data ie extra variables, changes in precision/accuracy/resolution of data collected, then getting out there and changing the instrumentation would not be a simple task logistically I imagine! Even on land Australia has a lot of remote areas from which data is probably required for input.

    Going by some of the modelling I've seen done by work colleagues, even if you have the right data, making adjustments is not always a simple matter. Figuring out where the errors are occuring, finding a solution, recalibrating the model, ground truthing and testing to ensure it still functions to an acceptable accuracy can be very time and resource consuming.

    Anyway, thats my thoughts, I am sure there is more to it though.
     
  3. Croweater

    Croweater First Runs

    Joined:
    Jul 24, 2000
    Messages:
    168
    Likes Received:
    0
    Location:
    adelaide
    I HAVE ASKED MYSELF THE SAME QUESTION VERM.
    COULD IT BE THAT THE MODELS ONLY REPORT THE MOISTURE CONTENT OF SYSTEMS AND THE POTENTIAL OF SYSTEMS TO PRODUCE RAIN AND CAN'T PREDICT EXACTLY WHERE IT WILL FALL --- IF AT ALL. CLOUD SEEDING COULD TAKE THE GUESS WORK OUT.