UBC Theses and Dissertations
Rapid forage inventory of Kenyan rangelands by double sampling for regression estimation : a feasibility study using an airborne digital radiometer Mwanje, Justus Inonda
There is a growing need for an inventory technique which can be readily applied to extensive and remote rangelands as current destructive and non-destructive sampling methods are slow and expensive. An attempt to resolve this problem is made. The major objective of this study, therefore, is to determine the feasibility of using airborne digital radiometers (biometers) to inventory range forage. Remotely sensed information of spectral radiance or reflectance ratio (RR) and local-level referenced biophysical characteristics variables from the climatically diversified Kajiado District, Kenya has been processed. A light-aircraft fitted with digital radiometer (biometer) was used in a double-sampling strategy for rapid acquisition of spectral data. Biomass clipping of sample microplots on selected transects provided local-level reference information. These data elucidated factors characterizing vegetation spectral response and hence behavior, limitations and applicability of digital radiometers as a tool for rapid inventorying of range forage (standing crop). Least-squares regression methods have been applied on the collected data. Statistically significant linear relationships have been found to exist between spectral radiance or reflectance ratio (RR) and biomass fresh weights (FW) (r = 0.703, P <0.01, df = 17), and leaf water content (LW) (r = 0.83, P <0.01, df = 17). Polynomial relationship between spectral reflectance ratio (RR) and biomass fresh weights (FW) was significant (r = 0.872, P <0.01), and exponential relationship between the same variables was also significant (r = 0.61, P <0.01, d.f. = 17). Comparing the spectral regressions demonstrated that the linear equations are the best, simple and ease to apply. Total ecosystem moisture capacity was found to influence plant canopy reflectivity on a large scale. The soil catena characteristics were identified as an influencing factor on transect spectral reflectance dynamics. Recommendations for further investigations, have been made and an improvement on the present sensor systems (radiometers) is desirable to optimize acquisition of spectral information. Percent vegetation greenness estimates were found to provide acceptable predictions of range forage in place of spectral data. However, such estimates need to be properly defined as they are primarily subjective. The digital radiometers have been shown to be a precise and expedient method for estimating standing crop. However, the present model (Tektronics J-16) should be used only in grasslands because its inability to penetrate the canopy gives inaccurate readings in brush-lands and woodlands. The use and reliance on remote sensing techniques show a great potential for monitoring tropical rangelands, in particular Kenya, for providing the necessary information for their ultimate resource development.
Item Citations and Data