# Is there an archived surface upward longwave radiation field in ERA5, or should it be derived as STRD − STR? 

**URL:** <https://forum.ecmwf.int/t/is-there-an-archived-surface-upward-longwave-radiation-field-in-era5-or-should-it-be-derived-as-strd-str/15360>\
**Category:** Other - Datasets and Usage\
**Tags:** c3s, era5\
**Created:** [28 September 2026 08:25 UTC](https://forum.ecmwf.int/t/is-there-an-archived-surface-upward-longwave-radiation-field-in-era5-or-should-it-be-derived-as-strd-str/15360 "2026-09-28T08:25:33Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Cristiam\_Portilla](https://forum.ecmwf.int/letter_avatar_proxy/v4/letter/c/85f322/32.png) [@Cristiam\_Portilla](https://forum.ecmwf.int/u/Cristiam_Portilla)\
**Post date:** [28 September 2026 08:25 UTC](https://forum.ecmwf.int/t/is-there-an-archived-surface-upward-longwave-radiation-field-in-era5-or-should-it-be-derived-as-strd-str/15360/1 "2026-09-28T08:25:33Z")

</div>

Our team is working on the Fondecyt Regular Project N° 1260452 (focused on projecting marine heatwaves and cold spells off the coast of Chile).

**Context of the problem:**

We are comparing the agreement between net shortwave and net longwave radiation derived from ERA5 against the same variables derived for 16 global climate models (CMIP6), over the historical period 1995-2014. Since CMIP6 does not directly provide the net variables for our workflow, we derive them by subtracting their components (downward − upward) for each model.

When comparing the net shortwave radiation derived from each GCM against that of ERA5, we obtain a very good fit. However, when performing the same comparison for net longwave radiation, using exactly the same pipeline, we find a much more significant mismatch — even though the individual component (downward) shows high correlation separately with the respective observations.

**Figure 1** (below) shows the daily series for both components: the Downwelling Longwave Radiation panel (left) shows good agreement between ERA5 and the 16-GCM ensemble, while the Net Longwave Radiation panel (right) shows a much weaker fit, with ERA5 falling largely outside the ensemble range.

**Figure 2** (below) shows the same comparison for the seasonal cycle: again, Downwelling Longwave Radiation (left) tracks the ensemble closely, while Net Longwave Radiation (right) shows a visibly poorer match.

**Our question:**

In order to independently recalculate ERA5’s net longwave radiation (by subtracting its own components, rather than using the already-computed net product), we need the upward component. In the CDS catalogue we only found the downward component available (“Surface thermal radiation downwards”, STRD) and the net (“Surface net thermal radiation”, STR).

Is there, in any other ERA5/ERA5-Land dataset or product, an archived field for surface upward longwave radiation? If it is not available, is it correct to assume that the standard way to obtain it is as a residual (STRD − STR), as reported in some radiation validation studies (where this derived variable is referred to as SLUR)? And if so, is there any known consideration regarding the accumulation convention of STRD and STR (for example, recent changes in how the CDS delivers hourly accumulated fields) that could introduce inconsistencies when deriving this difference, and that could explain part of the mismatch we are specifically observing in the net longwave component?

**One more thing:** if anyone happens to know a possible cause of our problem, or a solution to it, we would greatly appreciate your input and support.

Thank you in advance for any guidance.

**Figure 1**

 ![daily_series_comparable_1995](https://forum.ecmwf.int/uploads/default/original/2X/5/5ee5dcb76b8c1b4d2f5a475bb56f6dd769fdf89c.png)

**Figure 2**

 ![ciclo_estacional_comparable_1995](https://forum.ecmwf.int/uploads/default/original/2X/7/76eef65182ee035d6ec2222cd294aaa947809647.png)