So far, I've worked with the assumption that the time of the measurement is known (5 pm in SmallScaleHourly.ipynb), but part of the problem is that there is uncertainty as to the measurement time. I do think there is information about this in the data, from the combination of nearby hourly measurement and the T_n and T_x measurements. So from a mathematical and statistical point of view, it should be possible to add a parameter for this in the model, and infer the measurement times. However, the computational implications might be unpleasant.
So far, I've worked with the assumption that the time of the measurement is known (5 pm in SmallScaleHourly.ipynb), but part of the problem is that there is uncertainty as to the measurement time. I do think there is information about this in the data, from the combination of nearby hourly measurement and the T_n and T_x measurements. So from a mathematical and statistical point of view, it should be possible to add a parameter for this in the model, and infer the measurement times. However, the computational implications might be unpleasant.