There has been a problem with the EDDI product over the northeastern and north-central regions of CONUS during the winter of 2022/23 (and perhaps still at the longer timescales), with EDDI indicating extensive and severe dry conditions (ED1-ED4) over much of the region during this period. This was due to an error with the NLDAS-2 variables that drive EDDI. Since July, 2022 the NLDAS-2 forcing data (at NASA GES DISC) was ingesting snow-depth data at an incorrect precision, which was leading to gross overestimates of surface temperatures. This became a problem for EDDI as soon as snow-cover was established during the winter—around October/November, 2022—leading to EDDI showing extensive ED4 (98%ile drought) over much of northern and northeastern CONUS that was not reflected in the USDM or in other non-NLDAS-driven products.
While NASA implemented a correction to new data from July, 2023 forwards, given the memory of the land-surface as modeled by NLDAS-2 reanalysis system, the effects of this error will take some time to disappear from the reanalysis. Data for the affected period is not accumulated into the climatology of reference ET that is used to generate EDDI, so the affected EDDI is limited to the period of July, 2022 to when these lingering effects “wear out” of the reanalysis.
The long and short of this is that, from now (December 8, 2023) short-term EDDI (say, 1-week to 3-month timescales) is no longer affected by this error and may be used with confidence. As we proceed through the coming winter, this timescale threshold will grow. By the end of the winter, we believe that EDDI at all timescales will be free of the effects.
Hopefully, there will be a new reprocessing of NLDAS-2 for the affected period. At that point, we will re-run and re-post EDDI at all timescales.