State Institute of Statistics (SIS) has exclusive responsibility for producing and disseminating official statistics in Turkey. It conducts, among others, agricultural census once in ten years and publishes yearly agricultural production statistics which are based on data collected by the regional offices of Ministry of Agriculture. These figures are found to be not reliable. Because of the discrepancies between the yearly grain production figures of Ministry of Agriculture and results of Agricultural Census, State Planning Organization (SPO) has delegated SIS to conduct the "Estimation of Grain Production by Remote Sensing Project" in 1991. The objective of the project is to use advanced technology with minimal human intervention for better estimation of the cereal production in Turkey.

The project has national interest. SPO strongly supports the projects and coordinates the activities with the Ministry of Agriculture. Due to unreliable forecasts, there are times that Turkey is in a position to import wheat even though she is one of the major wheat producing country.

SIS has started the work by reviewing the studies in similar subjects throughout the world. It has been observed that the MARS project of the European Communities has similar objectives and it was decided to transfer the expertise and know-how accumulated in EC countries.

In 1992, two RISC based workstations, one PC, Image Processing and GIS software (Erdas and Arc/Info) were purchased. A formal training program for both Erdas and Arc/Info is provided by the supplier firms. Two experts from Scot Conseil of France have conducted a week length course on MARS Action IV, and with their consultance, three sites (40 km x 40 km) were chosen mainly based on the official statistics.

In 1993, satellite scenes were purchased for the selected sites (two for each) and the field work was conducted as required by the MARS methodology. The images are analyzed and the crop acreage values were achieved for the sites. Deciding that Action IV alone is not a complete solution for the requirements of the SIS project, it was decided to have some experience on Action I as well since it has been envisaged that the final approach to be followed for Turkey will be a combination of Action IV and I.

Consultants from GAF of Germany have visited SIS and gave seminars for three weeks about Action I methodology and its probable application in Turkey. In order to form a basis for the site selection in future activities, LANDSAT-MSS scenes covering the whole country have been purchased and photographic prints at 1:250.000 scale have been obtained. By visual interpretation of these images, the country has been stratified in eight classes. Also in 1993, one RISC workstation and two PC's have been purchased along with Erdas and Arc/Info software.

As the result of the experience achieved in 1993, it was decided to conduct both Action I and Action IV studies in selected provinces of Turkey. By comparing the results of these studies with the official statistics produced by Ministry of Agriculture, it would be decided on the optimal model to be applied throughout the country for the coming years. The aim was also to calculate the error component of the estimation methodologies. Unfortunately, as the result of economical crises, severe austerity measures have been taken for government spending in 1994 and the planned studies had not been conducted. Only the follow up of the three selected sites have been completed. Field work have been conducted and single image for each site has been obtained.

The activities planned for 1994 was carried out in 1995. Images covering the selected regions have been purchased. Intensive field work (based on 1% sampling) have been conducted in the selected regions. The results of Action I and Action IV activities in the selected regions were compared and a methodology to be applied in Turkey has been constructed. The main objective of the methodology was the announcement of early forecasts on acreage as early as June each year. Sites representing the agricultural field in Turkey were selected.

Consultance and training services have been acquired from NRSC in order to review the achievements and revise the future plans. A consultant from Joint Research Center (JRC), Ispra where the methodology of the MARS project has been developed have visited SIS (Mr. J. Gallego) to review the activities. 24 sites having 42 x 42 km dimensions were selected based on the stratification conducted by using MSS images. Different number of samples was drawn from each of the four strata. Field work could not be performed in one of the blocks, and for three sites, cloud free images could not be obtained. Therefore, the results for 1996 are based on 20 sample sites. Estimation of cereals acreage was calculated by using classification results of TM images with two different acquisition dates.

For the year 1997, the activities were conducted on the same 24 sites selected previously. Two images were acquired for each site. LISS images were substituted for the sites where TM image was not available. The cereal acreage estimation was calculated based on the classification of the acquired images.

The main condition for the project to be successful is the continuity of funding. Experience in the last four years showed that due to economical instability in Turkey, it is not possible to ensure appropriate continuos funding of the project. The scope of this proposal is to obtain international funds for the equipment, satellite data, consultance and training services. The funding of the personnel (3 Statisticians, 2 Geologists, 2 Agriculturists, 1 Environmental Engineer, 2 Cartographs, 2 Engineers, 1 Mathematician, 1 Geomorphologist) working in the project and expenses of the field study will be funded by the local sources of SIS.