Interventions may not be required in any one year (?). You want to use a SMART to decide the need for an intervention (?). You will already have a good idea when interventions might be required - You will have some idea on the pattern of nutritional status, infection, food availability, food prices, &c. throughout the year from (e.g.) previous surveys, surveillance data, clinic workload returns, agricultural calendars, market monitoring &c. You will usualy plan surveys for the "lean period", "hungry season", or whatever you want to call it. The basic approach is a two-stage screen. You have a sensitive but low cost first stage (mostly routine data and a few shopping trips for market data) followed a much more expensive but specific second stage (the SMART survey). You may be able to improve the first stage screen somewhat by adding indicators. In your example, you would probably want to add rainfall data or measures such as soil moisture or vegetative cover that can be obtained by remote (i.e. satellite) sensing or by observation by agriculture extension workers. You would collect and examine this basket of indicators to see if they indicated a problem. If they did then you would use a SMART survey (or similar) to confirm (or deny) the existence of a problem and to make a coarse estimate of need. I hope this is of some use.