Improved retrieval of tropospheric temperatures from remote measurements of thermal radiation using the adiabatic lapse rate as a constraint

Space-borne radiometers now form an important part of the global network of atmospheric observing systems which provide data for weather forecasting and studies of climate change. However, achieving adequate vertical resolution remains a problem, particularly for the retrieval of profiles of temperature and of water-vapour concentrations in the troposphere. The problem of retrieving these profiles from radiance measurements is illposed in that there is no unique answer-solutions may be unstable and excessively sensitive to measurement noise. A technique known as regularization may be used to stabilize the solution by biasing the retrieved profiles toward an acceptable form. Similar inverse problems occur in many areas of science and engineering. Work on image enhancement has shown that regularization which includes a constraint, such as non-negativity of the solution, can provide retrievals with improved accuracy over other regularization methods. We note that the adiabatic lapse rate provides an effective constraint on tropospheric temperature profiles and we exploit this to develop a regularization scheme for the retrieval of atmospheric temperatures. We demonstrate that the retrieved profiles are indeed more accurate than those obtained without this constraint.