Modeling objective - value of spatially-distributed vs. high-frequency data

[lekoenig]

Following up on a conversation @jds485 and I had with Joel Blomquist yesterday, it would be interesting to see how much process information the model learns from spatially-distributed (snapshot) samples versus the smaller number of high-frequency stations.

I find this question generally interesting when we're thinking about water quality prediction at unmonitored locations, but I also think there's utility in understanding whether our models or predictions are improved by the high-frequency monitoring through the NGWOS program, and/or informing where strategic placement of continuous stations would be most useful*.

*at least for basin- or regional-scale prediction or extrapolation