Hello,I am a person who recently started learning the Invest model,I have a question regarding the precipitation data requirements for the model. According to the user guide, the description for annual precipitation data states that “Ideally, the gauges will have at least 10 years of continuous data, with no large gaps” However, in the area I am trying to quantify, the annual precipitation varies greatly over the 10-year period (with the lowest and highest values differing by a factor of 2.2). The precipitation in 2014, which corresponds to the same year as LULC, is about 0.7 times the 10-year average. Given these precipitation conditions, if I want to quantify the annual water yield for 2014 in the area, should I use the precipitation data for 2014 alone, or would it be better to use the 10-year average data?
Hello @rebj, and welcome!
The Annual Water Yield model was made to represent long-term annual average water production. So we do recommend inputting the average precipitation over a 10-year period, rather than inputting annual values separately. It’s totally normal for precipitation to be different each year, but using long-term averages are one of the limitations of this simple model.
That said, people do use the model to represent each year separately. Even though it is not what the model was designed for, it’s ok to model one year at a time, but not to model sub-annual (like monthly) values.
~ Stacie
Thank you for your reply.
Hello @swolny!
I recently asked a question about the user guide’s recommendation to use precipitation data with a minimum period of at least 10 years. How could this recommendation be justified, considering that various authors and academics suggest that climate variability in terms of precipitation should be analyzed in intervals of approximately 25 to 30 years?
I appreciate your guidance in advance on this matter.
Hi @bmnoz_2001,
My take is that generally, the broader your temporal coverage of precipitation data, the better. It is often difficult or impossible to obtain data for certain locations. Ten years’ worth is certainly preferred to one year’s worth, while 20, 25, or 30 would be even better, if available. But, the reality is that many users cannot obtain these data for their study areas that cover so many years.
It’s important to detail the limitations and assumptions of any study, including periods of data coverage.
-Jesse