Current Suomi NPP Research Team:
Bryan A. Baum (PI, SSEC), W. Paul Menzel (Co-I, SSEC), Elisabeth Weisz (SSEC), and
Irina Gladkova (Co-I; City College of New York)
Goals
The goals of our team changed from evaluating VIIRS cloud products generated by contractor algorithms to contributing to a new process to generate cloud products for NASA's effort to continue climate data records from the Earth Observing System Aqua and Terra platforms into the future. The philosophy of the Suomi-NPP program to date has been that cloud properties are derived from the imager (VIIRS), while atmospheric profile products are derived from the IR interferometer (CrIS). Our philosophy is that the cloud products can be much improved by merging VIIRS with CrIS, or more generally, an imager with a sounder if they reside on the same platform.
Recent Advances:
The basis for our work came on a breakthrough achieved by Prof. Irina Gladkova and her graduate student (James Cross III) at the City College of New York to merge high spectral resolution IR data (specifically AIRS) with broadband imager data (specifically MODIS). This work is detailed in Cross et al. (2013 - see references link above). They constructed a high spatial resolution (1 km) 13.3-µm CO2 channel for MODIS by building a relationship between the MODIS 11- and 12-µm window radiances at high spatial resolution and the same radiances averaged to low spatial resolution coincident with the AIRS fields of view (FOV). Each high spatial resolution pixel was associated with five nearby low spatial resolution FOVs. Based on this relationship, they demonstrated that it was possible to construct high spatial resolution 13.3-µm radiances from AIRS spectral data. Since MODIS measures radiances at 13.3 µm (band 33), it serves as an ideal platform to test and analyze their method. The “fusion” constructed radiances were within 1% of the measured radiances where MODIS and AIRS both collect measurements. Cross et al. (2013) provided results for a MODIS granule (5 min) and demonstrated a positive impact on cloud top height retrievals using the Heidinger optimal estimation method (Heidinger et al. 2010). The initial application of this approach was to build such a channel for VIIRS, which lacks any IR absorption channels.
When Dr. Elisabeth Weisz began working with us in January, 2017, we began to make some really serious progress. The entire data fusion software package was completely reworked and underwent extensive testing on global MODIS and AIRS data. MODIS and AIRS provide an ideal testbed because MODIS has the IR absorption bands that we want to construct for VIIRS. In fact, we had one of those rare moments of scientific discovery when we came across something quite remarkable. While our initial goal was to use imager-sounder data fusion to construct a high spatial resolution (750 m) 13.3-micron band for VIIRS that was similar to MODIS Aqua Band 33, we found out that we could construct other MODIS IR bands too. We are now able to construct the IR absorption bands for VIIRS that are similar to those from MODIS: Bands 23, 24, 25, 27, 28, 30, and 33-36. This methodology was tested extensively with MODIS and AIRS, so that we could compare the fusion-based radiances to those actually measured by MODIS. The results indicate that radiances are accurate within about 1%, or brightness temperatures within 1-3K. These radiances may not have enough accuracy for some applications, but that cannot be determined without testing on global data. A full day of fusion radiances have been constructed for both MODIS+AIRS and VIIRS+CrIS, and are in the process of being tested for potential use in the derivation of total precipitable water, atmospheric stability, polar winds, volcanic ash plume tracking, smoke plume tracking, cloud-top height, and cloud thermodynamic phase.
To take this a step further, we can construct any additional IR bands for which the CrIS hyperspectral sounder has measurements.
Further details may be found in our Algorithm Theoretical Basis Document (ATBD). Additionally, a journal article on this work was just published (Weisz et al. 2017); and some examples are provided in the Data Fusion links at the top of this page.