WSP10301 - Laserfiche WebLink

Home Browse Search

28 Estimating Economic Impacts of Salinity,.. l'') Co:) ...... en leveL These data problems were reduced by selectively dropping certain aberrant data sets from the regression analysis. In deciding which data to retain and which to drop, primary em- phasis was given to data from Imperial County, California, because it represents a very large share of aU salinity damages (93 percent of all damages at 800 mg/L according to Kleinman and Browo).ll A computer program was created specifi- caUy to calculate a series of regression formulas from experimental data on agricultural crop yields against salinity leveL The computer pro- gram allows the user to enter up to 39 data sets, i.e., coordinates, and then determine a series or polynomial regression formulas, ranging through polynomial 1 (linear), polynomial 2 (quadratic), polynomial 3 (cubic), polynomial 4 and polyno- mial 5. In each case, the program calculates the regression curve of best fit, then calculates a Goodness of Fit Criterion to permit the user to compare the various regression curves to deter- mine which polynomial curve best fits the data sets. The program displays a plot of each curve and, with an on-line printer, provides a printout of the regression formula, the Goodness of Fit Criterion, and the plot of the regression curve. An example of a regression curve is shown in Figure 6. The plotted curves extended only through the range of TDS levels in the experimental data, i.e., 611-892 mg/L. There is no reason to assume that the curve formulas could be extrapolated at either end of this range, so a judgmental extrapo- lation was used to extend the computer-plotted curve through the entire range of 200-1200 mg/L TDS. The curves tended to be flat in the range from 200 mg/L to the approximate threshold of salinity damage determined by E. V. Maas, then dipped in the range from the damage threshold to 1200 mg/L. This nonlinear nature of the salinity function had been expected, based on the earlier work of Kleinman and Brown as weU as the work of Gardner, who noted that Klein- man and Brown found average salinity damages per mg/L in the 900-1400 mg/L range to be ORANGES YIELD 3 COUNTIES , 181'". : ',- . . ~ '........, . "- 145 " " : ',... . ~ . ". . . '". . : '.. . .... lAQ " --i . - . . . . . . . 74 . . . . . . : . 38---------+---------+---A-~--+----- 611. 9 691. 3 771. 6 -851. 9 TDS . . . QJ '- U <0 ---- ..... ~ u ~ -0 ~ QJ . .~ >- . . . . . .!, . . a -'-"'_ . --..~- --.......-........ . ---------- . . . - . . Figure 6. - Regression curve of yield of oranges/tangerines vs TDS level. l1Colorado River Salinity, p. 8.