throttling technique for data cube (space + time) requests

[bbest]

ERDDAP servers have limits. Ideally, we have a simple threshold like max_points = 1,000,000, and then the metadata for the gridded dataset gets interrogated so we:

1. Create the bbox grid: n_points_grid
2. Time slices requested based date_beg, date end and given the dataset's frequency: n_slices_time
3. Batch the rerddap::griddap() requests based on n_batches = n_points_grid * n_slices_time %/% max_points + 1 (assuming a remainder)

[bbest]

Coral Reef Watch (CRW) is a good test case b/c has daily data

[bbest]

For n_points_grid, see generation of r_template in #3

[bbest]

Let's try this new data cube throttling approach with this salinity dataset having x,y,z,t dimensions:

• ERDDAP - ECCO ECCO2 cube92 salt - Data Access Form

[7yl4r]

Hi Ben, I know things are already well along but I came across Roy Mendleson's script griddap_split and I wonder if there is anything there worth copying over here or vice-versa?