a a rt i S 
For the l-year, one-state (1978 N.D.) aggregation, no glaring deficiencies 
were noted. The attractive characteristics of low processing cost, 
objectivity, repeatability, modularity, and adaptability packaged within an 
automated framework make the outlook for meaningful advancements very 
optimi stic. 
B. SUMMER CROPS,.CORN, AND SOYBEANS EXPERIMENT? 
This experiment in summer crops, corn and soybeans proportion 
estimation technology represents our first attempt to take advantage of an 
improved systems approach to research. It allowed the effective utilization 
of the experimental environment for more iterations of development, testing, 
and feedback to research than heretofore achievable in a given time (such as 
6 months). 
Figure 6 shows our accuracy results (in terms of mean error and 
90 percent confidence limits) for the previously described data for crop years 
1978 and 1979. i 
The following are some additional observations: 
1. There is a very accurate estimate of crop group (summer crops) with 
relative mean error (RME) less than 2 percent in both years. 
2. For crop type (corn and soybeans) there was a significant bias (corn RME 
15 percent, soybeans RME -19 percent) in one of the years (1978). 
3. The crop type bias for the other year (1979) was not significant (corn RME 
6 percent, soybeans RME 3 percent) 
4. The direction of the bias for crop type appears to be consistent between 
years (corn tends to be overestimated, soybeans tend to be underestimated. 
5. The standard deviations for all estimates are relatively consistent and 
comparable to those achieved in previous "best" spring small grains 
technologies. 
6. Compared to a previous procedure for the 1978 crop years, CS-1 exhibited a 
significantly lower bias in estimating crop group (summer crop RME -1.5 vs. 
-16.4 percent) with lower standard deviations for both crop groups and crop 
types. 
In terms of the other preformance criteria the following was 
observed: 
1. Timeliness - processing to crop type is achievable after corn tasseling, 
which is 30-45 days prior to harvest. This is about the middle of August in 
the U.S. Corn Belt. 
2. The rate of processability of allocated segments with the CS-1 technology 
were quite high, typically 50 percent to 75 percent. 
Quick identification of subcomponent error contributors was 
achieved. These specific results have led to the earlier-than-planned 
development of a more-automated summer crop, corn, and soybeans proportion 
estimation procedure. The process of making the developmental modifications 
for a new version (CS-1A), conducting a verification test, then designing and 
implementing a semiautomatic version (CS-1B) has already been accomplished. 
Results of early verification testing over a sample (10 segments) of the 
1978-79 data set show excellent summer crop accuracy? and corn and soybeans 
accuracy of about 10 percent RME and standard deviations of 4 to 6 percent. 
374 
devel 
catio 
one-f 
are e 
1. MW 
objec 
2. W 
Was ni 
resoui 
8.. T 
subst 
probl 
criti: 
S E 
techn 
4. F 
crops 
non-U. 
B F 
quali: 
6. D 
perfo 
contr