Open Collections

Remote sensing applications in an alfalfa capability assessment of saline soils

University of British Columbia

In the Cariboo-Chilcotin region of British Columbia, two types of soil salinity have been recognized; natural salinity (Solonetzic soils) and secondary salinity. Soil salinity of either type can be a significant limiting factor in alfalfa (Medicago sp.) production. Stress-related changes in individual plant colour and plant morphology and invasion of the stand by salt-tolerant forbs and grasses affect the amount and direction of the radiation reflected and/or emitted by growing plants. Remote sensing techniques are capable of measuring that radiation and, therefore, offer the potential for quantitative assessment of plant stress caused by saline soils. A remote sensing study was initiated to develop a computer-assisted classification of categories of salt-affected soils. Site biophysical data and reflectance data from digitized infrared diapositives collected from a 50m x 50m grid of "training sites", were integrated to produce a series of digital numbers representative of different classes of salt-affected soils. The possible confounding effects of exposure, processing, geometric and atmospheric effects, and image window position on the interpretation of reflectance patterns were first evaluated. Alfalfa exposure fell on the linear portion of the Density-log Exposure curve indicating a satisfactory exposure. The IR balance of the film (22) was more appropriate to higher altitude use (12000 ft asl) than the altitude of the photographic flights (5500 ft asl) resulting in a slightly narrower range of near infrared reflectance values. Variation between four distinct images was not significantly different (P<0.05) so the complete data set was used for analysis. Sites positioned near photograph borders were avoided as decreased reflectance was noted at these locations. Parameters used in classification were selected on the strength of their correlation with other parameters in the system. The spectral-plant-soil model was developed using a correlation threshold (r > +0.35) to select parameters followed by a morphological systems analysis to devise the component systems. The key system parameters were; Spectral: Yellow(Green), Magenta(Red), Cyan(NIR), Ratio R3(M/C) and Ratio R4(M/Y); Plant: ALF(alfalfa), SALT(salt-tolerant grasses), GRASS(domestic grasses) and PUNU (Puccinellia nuttalli); Soil: PHW(pH water), EC (electrical conductivity), ECa (exchangeable Ca), ENa (exchangeable Na⁺), EK (exchangeable K⁺) BD (Bulk density) and E (elevation). Spectral parameters C and R4 were most strongly correlated with ALF (r = +0.40 and r = -0.35, respectively) and were selected as grouping parameters for cluster analysis. The training sites were clustered into three groups. The spectral data for these groups was used to devise digital numbers in the Y,M and C dye-layers for use in the Meridian-PC image analysis system. The plant and soil parameter trends (x + 1 sd and % CV) were used to determine continuous classifications representing LOW ALF (low alfalfa capability on saline and/or sodic soils), MED ALF (medium alfalfa capability, including areas of developing salinity with intermediate values for soil parameters) and HIGH ALF (high alfalfa capability which may include areas with high populations of domestic grass species in lieu of alfalfa which had declined for reasons unrelated to salinity). Parameter means were significantly different (P<0.05) for LOW ALF vs MED ALF, LOW ALF vs HIGH ALF and in some, but not all cases for MED ALF vs HIGH ALF. The BIOM (biomass) data set reflected higher production in the HIGH ALF class. Total nitrogen (TN) and organic matter (ORGC) showed negative correlations with saline soil indicators and positive trends with alfalfa. Total N levels were low to medium in comparison to common ratings for effective alfalfa production. More available phosphorus (PHTOS) was present in the LOW ALF than in the HIGH ALF, although all areas would require supplemental phosphorus to improve productivity. The spectral signatures supplied to the Meridian-PC image analysis system were used to produce capability maps of the study site depicting three alfalfa production classes and an unclassified category comprising values outside the spectral ranges established for the three classes. The training sites were grouped into LOW ALF (18%), MED ALF (43%) and HIGH ALF (37%). Computer-assisted supervised classification of the data improved the discrimination of the groups to produce a distribution of LOW ALF (31%), MED ALF (42%) and HIGH ALF (28%). This study successfully discriminated digital numbers which were indicative of classes of vegetation which are, in turn, reflective of gradations in saline soil conditions. Computer-based analysis of digitized CIR aerial photographs may, therefore, be a valuable tool in the identification and evaluation of the impact of saline soils on crop productivity.

Text

2010-09-09

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/28330

Plant Science

University of British Columbia

Graduate