Options (monthlies and dailies):

Observations:

1. HCN
2. VEMAP

Future (monthlies and dailies):

1. HCN monthly or daily data with GCM monthly deltas added (temperature) or ratios multiplied (precipitation). The resulting data has modern, observational variability, except if using the ratio approach for precipitation, which alters the variance of the time series by the square of the ratio (Mearns et al., 1996, 1997).
2. VEMAP: VEMAP monthlies maintain the GCM variability, though the data calculated by ratios will alter that variability, as discussed above. VEMAP dailies do not necessarily have the signature of modern, observational variability because they are generated values.
3. GCM: Maintains the original model variability for both monthlies and dailies.

Circumstances:

General: The VEMAP monthly dataset is probably fine for monthly/interannual variability examinations; the problem is with the dailies. To look at daily extremes, you are better off using the HCN data for the historical period and the HCN data combined with model deltas for the future.

1. Grid point vs. region: When looking at the trends at a given station or location, use the HCN-derived data for both observations and future. When looking at trends over a large region, use the VEMAP data for both observations and future. It is possible that trends at a point will approximate those for the VEMAP half degree grid, but that is not necessarily the case.
2. Daily autocorrelation: If you require synchroneity among the grid points for daily data, use GCM dailies. Since VEMAP dailies are generated by a weather generator, there is no correlation between the grid points in the daily data (only in the monthly data).
3. Daily extremes: Use direct GCM dailies for looking at extremes in temperature and precipitation. Since VEMAP dailies are generated by a weather generator, the values don't represent the daily extremes well. GCM dailies will incorporate variability changes due to global warming, though since the GCMs don't include all processes upon which daily variability depends, the results are uncertain and should be viewed as only one possible scenario. Ultimately you may want to look at both measure of variability.

Notes on Tmin, Tmax, Tmean calculations:

The atmospheric GCM computations occur at time steps on the order of 20-30 mintues, even though the actual values are saved only several times daily at most.  Temperature is computed as an instantaneous value at each model time step.  The minimum and maximum temperatures are therefore the minimum and maximum (screen) temperature values that occur amongst all the time steps between 0Z and 24Z at a given grid point.  The mean screen temperatures are the mean of the temperatures at each time step within this 24 hour period (in the case of CCCma, provided as two values between 0Z and 12Z and 12Z and 24Z).  Therefore the mean screen temperatures will not equal the mean of the minimum and maximum temperatures, even though this is the case in the real world where records of minimum and maximum temperatures are the primary temperature observations.  The average monthly temperatures (minimum, maximum, and mean) are calculated by averaging these daily values.

References:

Mearns, L. O., Rosenzweig, C., and R. Goldberg. 1997). Mean and variance change in climate scenarios: methods, agricultural applications, and measures of uncertainty. Climatic Change. 35(4): 367.

Mearns, L. O., Rosenzweig, C., and R. Goldberg. (1996). The effect of changes in daily and interannual climatic variability on Ceres-Wheat. a sensitivity study. Climatic Change. 32(3): 257.