SIMULATION OF IMPACTS OF CLIMATOLOGICAL MSU DATA-PROCESSING METHODS USING NCEP NCAR REANALYSIS DATA/

The Microwave Sounding Units (MSUs) aboard TIROS-N/NOAA satellites hav e played an important role in providing long-term global climate :info rmation. However, mainly due to the ''limb-correction'' and the ''intr ascan averaging'' procedures in the data processing, changes to the ch aracteristics of the data may have occurred. In this paper we investig ate this possibility in two parts. In the first part we have chosen to revisit the radiative transfer theory and to re-derive the (effective ) radiative transfer equation for the processed MSU data under more ge neral conditions. We find that the weighting function of the processed MSU deviates from the original at nadir and that the normalization co ndition, which ensures the total weight of contributions to the total observed radiance from different sources sums up to unity, is breached . In the later part of the paper we generate simulations of climatolog ical MSU-2 before and after data processing using a line-by-line radia tive transfer code of Shah and Rind [1995] with the monthly mean outpu t of the National Centers for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) reanalysis project [Kalnay et rrl., 1996] for the period 1982-1994. These simulations are examined a nd compared with the processed MSU-2 observations available in the pub lic domain. We find that a systematic reduction of contrast between la nd and ocean MSU-2 brightness temperatures can result from the limb-co rrection procedure. The temporal variances are affected by the greater -than-unity limb-correction coefficients, especially in the tropical r egion. A reduction of the spatial variances over the land region, as w as reported earlier by Basist et al. [1995], can be caused by the intr ascan averaging treatment in the processing. We suggest that a better approach in future MSU data processing may be to use the nadir data di rectly without averaging with the other off nadir data and, in additio n, that the off-nadir data should be ''navigated'' and be processed in to separate data sets.