Precipitable Water Variability Studies

A dataset of precipitable water over the US for the summers of 1987-88 is being developed at MSFC. This dataset is being generated using the Physical Split Window (PSW) technique for deriving Precipitable Water (PW). This technique utilizes differential absorption in the 11 and 12 10^-6m (or µm) regions to obtain an PW estimate (Guillory et al. 1993; Jedlovec et al. 1994). This study is using GOES-7 VAS infrared data for the summer of 1988 in the calculations. The domain of the dataset includes the continental United States, Mexico, extreme southern Canada, and portions of the Caribbean. Retrievals were made once per day at 1500 UTC for the period June 1 - August 31, 1988. Gridded daily values for this dataset were averaged over a smaller domain covering much of the eastern and central U.S. Temporal and spatial analysis of this dataset is planned as well as producing an equivalent dataset for the summer of 1987. GOES IR data have the potential to provide very high temporal and moderate spatial coverage of PW over the Americas for the period beginning in 1981 and extending beyond the year 2000. Products such as the PSW PW dataset from GOES can be used to study climate change and for GCM verification.

Objectives

The general research objective for this topic is to quantify regional-scale PW, its temporal and spatial variability, and its relationship to observed cloud cover, precipitation, storm formation, and surface features. This work has just begun and the specific tasks are listed below and supports NASA's GVaP and GCIP initiatives.

1.Generate accurate regional distributions of PW derived from geostationary satellite data for extend periods of time. The PSW algorithm will be used to derive PW at high spatial resolutions over the continental U.S. and surrounding areas. Satellite data sets will include historical data from the GOES pathfinder period and/or future data from the GOES I imager/sounder.

2.Available "ground truth" data (e.g., radiosondes, surface observations, special in situ or remote water vapor measurements, etc.) will be utilized to validate the accuracy and reliability of PW retrievals and the derived variability.

3.Regional water vapor variability will be quantified with structure function analysis and with other statistical procedures to document the variability over weeks, months, and seasons.

4.The time series of regional water vapor data will be correlated to other hydrologic and climate parameters to study the inter-relationship of these variables.

5.Higher resolution data sets from aircraft platforms will be used to study very fine scale variability and local sources of moisture.

Methodology and Results

The research described below focuses on the application of the PSW retrieval technique (Jedlovec 1987; Guillory et al. 1993) for deriving PW, or precipitable water, from the Geosychronous Operational Environmental Satellites (GOES). Results will be compared to the NASA Water Vapor Project (NVAP) PW dataset as well as radiosonde measurements.

GOES Pathfinder data from the Visible Infrared Spin Scan Radiometer (VISSR) Atmospheric Sounder (VAS) are used in this study. These data were obtained from the University of Wisconsin and are available for the benchmark period (May 1987 - November 1988). The data gathered for this study consisted of multispectral imaging (MSI) data for August 1988 which also included visible channel data.

A first guess field was derived from radiosonde observations for August 1988. These resulting data were objectively analyzed to a 4 x 4 grid encompassing the region from 16 to 52 N and from 68 to 124 W using a 2-D Barnes objective analysis scheme. Finally, these objectively analyzed data were put through the "forward" code of a radiative transfer model. This code takes a vertical profile of temperature and mixing ratio and simulates the channel transmittances and the observed brightness temperature at that point. This was done for the entire Barnes' grid. These analyses were performed using 0000 UTC radiosondes for the day after the retrieval day, i.e., a first guess of 0000 UTC 2 August 1988 was employed as the first guess for retrievals made on 1 August 1988. The 0000 UTC time was chosen over 1200 UTC to reduce the effect of surface-based inversions on the guess information (Suggs and Jedlovec 1996).

The retrieval algorithm employed in this study is essentially the same as that used by Guillory et al. (1993) with modifications described in Jedlovec et al. (1994). Retrievals were made at 1500 UTC for each day in August 1988 (except for August 10, 15, and 18). These retrievals were made at an approximate spacing of 120 km. Cloud filtering in the algorithm is presently limited to using a simple 11 m brightness temperature threshold. The retrieval data were then analyzed to a 1 x 1 grid using the Barnes scheme so that a comparison could be made with the NVAP and radiosonde products.

The NVAP dataset is a global PW dataset that utilizes as weighted merging scheme of Special Sensor Microwave/Imager (SSM/I) microwave retrievals, TIROS Operational Vertical Sounder (TOVS) infrared retrievals and quality controlled radiosonde retrievals. The final PW dataset was created by combining all three of these input datasets, using a hierarchical weighting scheme. This algorithm uses radiosonde data, when available as truth, and then applies a weighting scheme to the TOVS and SSM/I retrievals. Finally, linear and temporal interpolation is used to fill in missing data points (Vonder Haar et al. 1994). The NVAP data are on a 1 x 1 grid.

Examples of Results

• June 1988 PSW Mean

Last updated on: November 2, 1999