Using Water Years in APEX-CUTE

[Lauren Cohen]

Hello,

I’m trying to use APEX to model the nutrient fluxes in a small watershed in Northwestern Washington State for my thesis research. I’m currently in the process of setting up my calibration files for APEX-CUTE. In Washington, we have very distinct wet (October – April/May) and dry (April/May – September) seasons. Consequently, hydrologic work here tends to use “water years” that run from October 1^st through the following September 31^st. I have seven water years of measured data in my watershed (October 2015 through September 2022). I’m planning to use the 1^st, 3^rd, 5^th, and 6^th water years of data as my calibration dataset and the 2^nd, 4^th, and 7^th water years as my validation data because of some changes in the stream over time that I want the model to consider. The APEX-CUTE manual says that I need an annual file for my outlet watershed in addition to monthly and daily. But the manual also says that it is “recommended that observation data starts January 1^st”. Starting January first does not make ecological sense to me, so I would like to use water years.

All of this poses a problem for creating the annual files. If I sum my variables across the years, those values are going to include data that are part of the validation dataset, and they’re also going to include data from water years that are potentially very different (i.e. December 2020 is far more similar to January 2021 than to January 2020). How should I handle these issues when I’m creating my files?

My biggest concern is the annual file, but I would appreciate hearing about if anyone has used water years in APEX-CUTE or if anyone has recommendations about using water years.

Thank you so much,

Lauren

[Jaehak]

If your data are monthly values, you only prepare monthly observation data file(s). Your obs file may start from January to December. If no data are available, you put '-999'. Note that -999 is no value that is different from zero. Here's my recommendation for setting up the CUTE. 

• Build monthly reach obs file with observed values for those odd years and put -999 in all months of even years.
• Set the CUTE to conduct only calibration, monthly time step, and reach N output. 
• After calibration, calculate performance statistics for the even years outside the CUTE.