Vegetation in the microwave regime can be modelled by replacing vegetation scattering elements (stems, leaflets, etc.) by their dielectric equivalents. Unlike most vegetation, the scattering elements in the alfalfa canopy are not independent of one another at L-band and they also exhibit a unique geometrical structure. This structural feature of the alfalfa canopy has been incorporated into a modelling procedure to predict radar response from the canopy. An alfalfa plant is modelled as a conical clump of stems that are clustered with leaflets. The scattering amplitude of this structural unit (clump) is calculated by the coherent addition of amplitudes from the individual stems. The calculations are done for L-band, where the leaflet size as well as the average distance among the leaflets are small compared to the wavelength. Careful ground and canopy geometry data were collected during the WASHITA'92 experiment conducted in Chickasha, Oklahoma in June 1992. The clump characteristics such as the number of stems with a common root, clump geometry, clustering factor of the leaves, etc., were measured in addition to sizes, density distributions of the stems and the leaves. These measured parameters of the alfalfa canopy have been used in a distorted Born approximation to compute the backscattering coefficient. For comparison, the backscatter from the alfalfa canopy is also computed using a conventional modelling procedure. In this approach, the vegetation is modelled as a random, independent, non-clump distribution of stems and leaflets. The theoretical results with and without the clumping structure are compared with one another and also with the experimental data obtained by the NASA truck based radar. There is a reasonable agreement between the clump model and the data. More importantly, the present study emphasizes the importance of coherence effects among different scattering elements in a canopy and the need to model them by identifying a respective structural unit for a particular canopy.