MODAPS PCR 21-025
| PCR Number | 21-025 |
|---|---|
| Date | 2021-09-02 |
| Initiator | Sudipta Sarkar |
| Abstract | PGE23 V6.1.13 -- Modified the NRT procedure: make a forced 9-day magnitude inversion after the full inversion. |
| Problem Statement | PGE23N lags by 7 days, when compared with OPS PGE23 at same day of interest (DOY). This is because the DOY is the last of the 16-day input period for PGE23N, and the 9th for PGE23. As the BRDF shape is determined largely by the set of input observations so, differences in the input observations created fundamental differences in the BRDF from NRT and OPS. |
| Description of Change | In order to minimize these differences between the NRT and OPS BRDF, due to differences in the set of input observations, the NRT BRDF code has been modified to allow for an extra round of magnitude inversion, following a full inversion using the full set of inputs. This extra magnitude inversion will only use the set of 9 days that are overlapping between OPS and NRT, with the highest weight being assigned to the last day. As an example: NRT 2021016 uses input days 1-16; OPS 2021016 uses inputs 8-23. For 2021016, day 8-16 are the common inputs for NRT and OPS. Hence, with this revision, the NRT code will now follow-up the first round of full inversion using days 1-16 with a second round of magnitude inversion using only these overlapping days 8-16. The day 16 will be assigned the highest weight. Through the extra magnitude inversion, we adjust the NRT BRDF using the common inputs with OPS and assign the highest weight to the last day (which is the center day for OPS). This way the lag will be more or less reduced, but the OPS and NRT aren't expected to show perfect match because of differences caused by the 7-additional days in NRT and OPS |
| Products Affected | Daily BRDF/Albedo (MCD43A1N/A2N/A3N/A4N) in the NRT processing stream |
| Software Affected | PGE23 V6.1.13 |
| Processing String to Receive the Change | Terra and Aqua Forward Processing. Terra and Aqua Reprocessing. NRT. |
| Downstream Product Effects | All C61 L3 daily BRDF/Albedo products (MCD43A1N/A2N/A3N/A4N) in the NRT processing stream. |
| Data Granules to be Used for Science Testing | A science test was run in AS 1858 from nrt4 and confirmed thoroughly by the science team and LDOPE. |
| Justification | The C61 PGE23N V6.1.13 will diminish any differences caused by the differences in the input days, when compared to the OPS C61 PGE23, by forcing magnitude inversion after full inversion, with a subset of days that are common between OPS and NRT. While it's not possible to completely remove this difference, the differences are expected to come down a bit, following this approach and has been confirmed by the science team. |
| Effective Date for Implementation of Change | The revised PGE23N V6.1.13 was inducted for C61 NRT at leading edge for all C61 MCD43* daily L3 products and ancillary DB and update files, starting 09/07/2021. |
| Status | Major algorithm development completed. Sadashiva Devadiga approved on 09/12/2021 Gang Ye approved on 09/12/2021. |
