Discrepancy between gridded ERA5 vs DAYMET downwelling surface shortwave solar radiation

Question

I am looking at two datasets that estimate downwelling/incident surface shortwave radiation (or insolation), and noticed a consistent order-of-magnitude discrepancy between the two while playing with the data in Google Earth Engine.

I compared the following:

1. DAYMET V4 srad, which according to the docs is "Incident shortwave radiation flux density, taken as an average over the daylight period of the day."
2. ERA5-Land Hourly surface_solar_radiation_downwards, which the docs defines as "Amount of solar radiation (also known as shortwave radiation) reaching the surface of the Earth."

To do a 1-1 comparison, I converted the units of (1) to match (2), converting from W/m^2 to J/m^2 by multiplying by the daylength, in seconds.

var daymet_daily_j_per_m2 = daymet_image.select('srad').multiply(daymet_image.select('dayl'))

Since the ERA5 data is hourly, I computed the sum of incoming solar for each day:

var era_images_on_day = era5_hourly
  .select('surface_solar_radiation_downwards')
  .filterDate(date, date.advance(1,'day'))
// print(era_images_on_day) // 24 images, as expected

var era_daily_j_per_m2 = era_images_on_day.sum().set('date',date)

Then I directly compared the daymet_daily_j_per_m2 with era_daily_j_per_m2 images, by taking an average over a ROI and computing the ratio. The ratio between the two is consistently around 0.10 or so, varying a bit depending on the date and the region.

Why is there an order-of-magnitude discrepancy between the two datasets?

Link to code: https://code.earthengine.google.com/dfca0d838f4b1b1d6a088f9cbfbe8fb0

Here is a linear regression (blue line) for a set of plots in North America, comparing daily summer data from 2019 to 2022. The dotted grey line is the expected 1:1 ratio. The y-axis shows the surface shortwave radiation from ERA5, and the x-axis shows the same from DAYMET.

Answer 1

For historical reasons, source files are accumulating hourly values every day for flow bands. We say this in the band descriptions: "This variable is accumulated from the beginning of the forecast time to the end of the forecast step.", but I guess this is not prominent enough.

For ease of use, we add _hourly bands where the values are disaggregated. Use bands like 'surface_solar_radiation_downwards_hourly' instead of 'surface_solar_radiation_downwards'