AQUA; a model to evaluate water deficits and excesses in tropical cropping, 2: Regional yield prediction
Indices of water deficit (IDef) and of deficit + excess (IStres) produced with the water balance model AQUA were correlated against rainfed regional yields of one annual planting of rice and beans, and of two different planting dates of maize. Data used are from northern Costa Rica, with 12 years of rice yield data available and eight years for the other crops. The linear regression model best fitted the data, with all r values obtained significant at P<0.05 and most at P<0.01. Although water deficits explained most of the year-to-year yield variability, in all cases r exponente 2 values from using IStres were higher than those from IDef; this indicates that a portion of yield reductions are the result of water excess. Yields predicted for 2 years were not significantly different from actual yields, using either index. Yield variability explained through r exponente 2 by using IStres was 90 percent for rice, 70 percent for beans, and 61 and 83 percent for first and second maize plantings, respectively. After converting yields to a relative basis, both maize plantings were grouped in one regression analysis and 71 percent of yield variability was explained by using IStres. After converting yields of all three crops and two maize plantings to their respective relative values, one single regression was run using all values. The overall yield explanation thus obtained was of 76 percent using IStres and of 65 percent using only deficits through IDef. A simple water balance model like this, that can explain from 60 to 90 percent of regional yield variability of individual plantings, from 80 to 90 percent of variability of individual crops if the best planting date for maize is considered, and 76 percent of the variability of all three crops and planting dates when regressed together, can have many uses such as yield forecasting, crop zonification and regionalization, and a series of analyses of climates and expected responses of crops that at present have but a loose connection with yield.
Main Author: | |
---|---|
Format: | biblioteca |
Published: |
1990
|
Subjects: | BALANCE HIDRICO, RECURSOS HIDRICOS, ABASTECIMIENTO DE AGUA, SISTEMAS DE CULTIVO, TECNICAS DE PREDICCION, RENDIMIENTO, PROGRAMAS DE COMPUTADORA, MODELOS MATEMATICOS, APLICACIONES DEL COMPUTADOR, COSTA RICA, |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Indices of water deficit (IDef) and of deficit + excess (IStres) produced with the water balance model AQUA were correlated against rainfed regional yields of one annual planting of rice and beans, and of two different planting dates of maize. Data used are from northern Costa Rica, with 12 years of rice yield data available and eight years for the other crops. The linear regression model best fitted the data, with all r values obtained significant at P<0.05 and most at P<0.01. Although water deficits explained most of the year-to-year yield variability, in all cases r exponente 2 values from using IStres were higher than those from IDef; this indicates that a portion of yield reductions are the result of water excess. Yields predicted for 2 years were not significantly different from actual yields, using either index. Yield variability explained through r exponente 2 by using IStres was 90 percent for rice, 70 percent for beans, and 61 and 83 percent for first and second maize plantings, respectively. After converting yields to a relative basis, both maize plantings were grouped in one regression analysis and 71 percent of yield variability was explained by using IStres. After converting yields of all three crops and two maize plantings to their respective relative values, one single regression was run using all values. The overall yield explanation thus obtained was of 76 percent using IStres and of 65 percent using only deficits through IDef. A simple water balance model like this, that can explain from 60 to 90 percent of regional yield variability of individual plantings, from 80 to 90 percent of variability of individual crops if the best planting date for maize is considered, and 76 percent of the variability of all three crops and planting dates when regressed together, can have many uses such as yield forecasting, crop zonification and regionalization, and a series of analyses of climates and expected responses of crops that at present have but a loose connection with yield. |
---|