Remote Sensing for Science, Education, Rainer Reuter (Editor) and Natural and Cultural Heritage EARSeL, 2010 Estimates of Yield Reduction Caused by Drought Katarzyna DABROWSKA-ZIELINSKA, Andrzej CIOLKOSZ , Alicja MALINSKA, Jedrzej BOJANOWSKI, Wanda KOWALIK, Maria BUDZYNSKA, and Maciej BARTOLD Institute of Geodesy and Cartography, Remote Sensing Centre, 02-679 Warsaw, Poland Abstract. The great demand on quick actual information on crop growth conditions and yield forecast for cereals caused the development of the method, which is based on indices created only from remotely sensed data. NOAA operational polar-orbiting satellites provide regular repetitive observation. The station is situated in the Institute of Geodesy and Cartography. Normalized Vege- tation Index (NDVI) for each decade of the year has been calculated. The AVHRR data collected in the first two channels have been used to calculate NDVI and the data obtained in the channels 4 and 5 have been used to calculate surface temperature (Ts). Three indices - Vegetation Condition Index (VCI), Temperature Condition Index (TCI) and Cumulated NDVI (CNDVI) have been computed for agriculture area for each ten-day period of the year. Every year differs the time when vegetation starts to grow. The delay or early start most often influence the time of different vegeta- tion phenology. The best period of time for crop yield prognosis was described by Cumulated NDVI (CNDVI) values as the measure of crop stage. It was found that the dominant information for the prognosis was CNDVI equal to 0.5 and 4. Then the indices VCI and TCI best characterized the future yield. Drought periods were computed and the reduction of yield due to drought have been performed and compared with data from Statistical Office. The methodology shows that the prognosis of biomass can be done with the error of 5-10%. Keywords. Drought, indices, yield reduction cumulated NDVI, modelling Introduction Proper management of agriculture requires rapid data about crop it’s growth conditions and progno- sis of yield reduction due to drought. For this objective remote sensing methods play a significant role, because of spatial and temporal coverage capability. Spectral reflectance signatures taken in the optical spectrum are very useful for such applications, and a variety of information from optical sensors can be applied for estimating soil moisture and vegetation growth conditions. However, the acquisition of optical sensor data is often hampered by unfavourable weather and that’s why com- position of indices for every 10 days of vegetation growth diminish the influence of atmosphere ef- fect causing significant error. Applications of NOAA/AVHRR satellite data play a significant role, because of spatial and temporal coverage. This paper demonstrates methods for forecast the yield reduction due to drought using only re- mote sensing information. The present results show that this method gave good results for yield prognosis and for such important information about the yield forecast. 1. Materials and methods From NOAA/AVHRR data the Normalised Vegetation Index (NDVI) for each pixel of arable land for 10 day period of the year has been calculated. It is well known that the NDVI fluctuates due to favourable or unfavourable weather and environmental conditions, Kogan (1997), Dabrowska– Zielinska et al. 2002. These variability of NDVI were estimated relative to the maximum and mini- mum (max/min) intervals of NDVI and named the Vegetation Condition Index (VCI):