Discussion 



and F-3, showing a 

 complex set of frontal features. 

 The NESDIS worksheet and a 

 SLAR interpretation are super- 

 imposed in Figure F-3, showing a 

 very close match of the features. 



In comparing the two images in 

 Figure F-3, the similarities are 

 apparent. The two fronts that 

 converge to the east, the 

 transverse north-south feature to 

 the west and the area of sharp 

 curvature on the western part of 

 the southern front are present in 

 both images, but differ somewhat 

 in spatial orientation. The SLAR 

 fails to detect the warm-core eddy 

 shown on the final NESDIS chart 

 (probably off the edge of the film), 

 but SLAR shows an additional 

 small-scale (1 km) eddy along 

 the front. The differences 

 between the SLAR interpretation 

 and the NESDIS worksheet do 

 not appear to be due to 

 navigational displacement or 

 rotation between the two, and are 

 probably due to movement of the 

 feature during the two day span 

 between images. 



It is significant that by using the 

 apparent SLAR 

 temperature/backscatter 

 relationship, the gradatbn of 

 temperature from south to north 

 is the same for both SLAR and 

 AVHRR, i.e., a large area of warm 

 water to the south (the Gulf 

 Stream), a narrow band of cool 

 water, a band of warm water and 

 finally a band of cool water. 

 Perhaps more important is tfiat, 

 giving the good match of location, 

 shape and apparent temperature 

 gradients across fronts, both 

 SLAR and AVHRR appear to be 

 detecting the same features. 



The SLAR and satellite infrared 

 imagery from 28 April and 26 

 April, respectively, show a good 

 match of the features detected, 

 both in location and overall 

 shape. The particular SLAR 

 image is a good illustration of how 

 well the two sources can agree. 

 Over the three years of SLAR 

 operation at IIP, a large number of 

 SLAR images of fronts have been 

 collected. Of these, there have 

 been a number of cases in which 

 SLAR and AVHRR do not seem 

 to agree both in k>cation and 

 shape of features. Williams 

 (1985) examines the match and 

 mismatch of SLAR and AVHRR 

 images in eddies and associated 

 features in the IIP region. Most 

 frequently the difference seems 

 to be one of placement rattier 

 than shape, reflecting a 

 navigational discrepancy 

 between the two sources. 



Of the tvw), SLAR offers 



the greater positional accuracy. It 



makes use of the aircraft's Inertial 

 Navigation System (INS), yielding 

 an accuracy of ±5 km (Thayer 

 SLAR/LORAN,unpub.). 

 Positioning on the NESDIS chart 

 is done using visible known land 

 forms on the image, which may be 

 obscured by ctoud cover, making 

 it less accurate, with errors 

 possibly as rrxjch as 15-20 

 kilometers (personal 

 communication, Jennifer Clark, 

 NESDIS). Given the nature of the 

 NESDIS product, i.e., the large 

 area covered, more accurate 

 F>ositioning is unnecessary. 



There are other cases in which 

 there is considerable difference 

 in overall shape between SLAR 

 and the NESDIS product. This 

 usually occurs when the area is 

 obscured by ck>uds and NESDIS 

 is estimating the kxjation and 

 shape of features based on 

 information that is up to several 

 days okl. 



In worthing with the original 

 satellite imagery, the wori<sheets 

 produced from it, and the final 

 NESDIS product, it becomes 

 apparent that NESDIS is able to 

 take very complex, detailed 

 imagery and produce from K 

 remarkably accurate, coherent 

 informatton. The imagery 

 compared in Figure F-3 occupies 

 approximately 1 square 

 centimeter on the satellite image, 

 from whrch the NESDIS 

 interpreter was able to extract and 

 correctly interpret several 

 features. 



88 



