Simple Rainfall Assimilation

Problem:

  • in test assimilations, noticed dry zone in South America
  • dry zone were associated with very dry soil moisture
  • positive feedback, especially in data-poor regions
  • wanted a way to constrain the soil moisture (not nudging)
  • must be global

A simple scheme:

  • combined the observed and model precip. to correct the soil moisture
  • Xie-Arkin's pentad precipitation (satellite and gauge measurements)
  • takes observed precip, model precip and model runoff for the
    previous pentad to calculate a correction for the next pentad

  • if model runoff is zero then
    • use (ObsPrecip - ModelPrecip) to correct top soil layer
  • if ObsPrecip   <   (ModelPrecip - Runoff) then
    • correct soil moisture by ObsPrecip - (ModelPrecip - Runoff)
  • otherwise
    • assume error in the model precip affected the runoff


  • misc. details
    • errors in the 5-day precipitation are "corrected" over the next 5 days
      so that "right" amount of water enters the soil.
    • correction is lagged in time
    • daily observed precipitation would be better (smaller lag)
    • nice feature: no error in the model precip. => no correction term

    • no correction is done with the soil is frozen
    • if the correction drys out top layer, soil moisture is
      removed from the middle layer
    • if the correction saturates the top layer, soil moisture
      is added to the middle layer

    Privacy Policy