Field experiments were conducted in moderately saline and saline soils duri
ng the 1996 dry and wet seasons and the 1997 dry season to document salt dy
namics and establish their relationship with local hydrology. Topsoil (0-15
cm) salinity in the dry season varied from 4.0 to 9.0 dS m(-1) in moderate
ly saline soils at Mirzapur and from 5.0 to 12.0 dS m(-1) in saline soils a
t Barodanga. In wet season, the corresponding figures were from 1.5 to 2.5
dS m(-1) and from 2.0 to 3.0 dS m(-1) respectively. Dry season cropping sig
nificantly reduced topsoil salinity at both the research sites. Overall pea
k salinity in non-plowed cropped lands was 25-38% lower than that of fallow
lands, and in plowed cropped lands it was about 30-40% less than the non-p
lowed cropped lands.
Multiple linear and non-linear regression models were developed to predict
topsoil salinity of the fallow land for both moderately saline and saline s
oils by using daily rainfall and evaporation as independent variables. The
prediction level was not significantly improved when a non-linear model was
employed in place of linear model. Therefore, a linear model may be used t
o predict topsoil salinity of the coastal ricelands of Bangladesh. (C) 2001
Elsevier Science B.V. All rights reserved.