A COMPARISON 01- SAMPLING HliTIIODS I'OK PERCENT COVER IN TUNDRA VEGETATION

I. A <:>•.::;•. iri son of Sampling Methods for Percent Cover in Tundra Vcgclat ion Line intercept and several random point techniques are compared. Random point techniques arc found preferable to line intercept in providing error estimates and speed of execution. Several lines with random points selected from them is a good compromise between a simple random sample of an area and the speed possible by restricting attention to a line. A technique is given for sequentially drawing a simple random sample from a line. II. A Cluster Analysis of Satell i te (ERTS) Data Data from the Earth Resources Technology Satellite (ERTS) arc examined fo r a "na tu ra l " c lass ifica t ion . C lass ifica t ion c r i te r ion involves commonness of classified spectral signatures and rarity of intermediates. Such a "natural" classification does seem to exist. The method makes no assumptions about underlying distribution of data. It involves order N calculations where N is the number of recognized d i f fe rent spect ra l s ignatures .