I am using the SDR and SWY models in watersheds located in mountainous areas of the southeastern region of Brazil. Field measurements of turbidity (monthly) and streamflow (bimonthly) were carried out at two points within the watersheds during one year. Therefore, we have 12 turbidity data points and 6 streamflow data points for each location.
Although the recommendation is to have at least 10 years of empirical data for validation and calibration, given the difficulty in obtaining long-term time series data, is it still possible to perform any type of analysis or comparison considering this data context?
Thank you for writing in to the forum with your question.
Depending on your study region, it can be difficult to obtain ample empirical data to calibrate, validate, and compare these model results with the real-world confidently. Certainly some observations are better than none. But, it’s really temporal mean values that should be used for calibration (annual or monthly). When you only have measurements at a given point in time, your data could be skewed by storm events, miss capturing variation, and likely do not represent the reality throughout the seasons.
If you draw any conclusions using these comparisons, I recommend that you be very diligent in explicitly detailing all the assumptions and limitations of your data and their interpretation.
Alternatively, it may be possible to use observed data from neighboring and/or similar watersheds as proxies, but you should be very careful in doing so and critically assess and deeply understand the differences between the catchments.
These are my personal impressions, but I’m curious what others think.