SDR model calibration


Regarding the calibration and validation of the SDR model 3.3.3:

In the modeling that I performed, I used the following data:
  • Raster land use - year 2010
  • Raster R - interpolation of the R for the period of 1980-2010
  • Observed sediment yield data - for 2007 to 2011 (obtained using the average annual flow applied to a regression equation sediment x flow)

To calibrate and validate the model:
  • Should I compare the modeling result (sediment export) to the observed value for the year 2010 or can I use the average of observed data for the years 2007 - 2011 ? 
  • If I use a year of 2010 to calibrate, should my R-factor correspond to this year, or can I use the R-Raster of the years 1980-2010  (anual average)? What is the influence of this factor on calibration / validation?
Thank you in advance


  • swolnyswolny Member, NatCap Staff
    Since the SDR model is an annual average model, the modeling result (sediment export) should be compared with the average of observed data for the years 2007-2011. More years would be better (we usually recommend at least 10), but we work with what we have. Also ideally, your R raster would be averaged over those same years to represent the climate more accurately, but if you do not have each year's R raster separately then obviously that's not possible.

    ~ Stacie
Sign In or Register to comment.