The assessment of PERSIANN model as an alternative for interpolation methods to evaluate wheat rain-fed yield (A case study for northeast of Iran)

Precipitation is the most important factor which affects the rain-fed wheat productivity. Precipitation characteristics, such as rainfall intensity and duration, usually exhibit significant spatial variation, even within small watersheds; while rain gauge network density could not provide desirable cover. Nearly all related researches use interpolation methods for places without rain gauge data. Many studies showed that the estimated error was usually high by usual interpolation methods. Employing satellite data with high spatial and temporal resolution could provide accurate precipitation estimation. PERSIANN (Precipitation estimation from remotely sensed information using artificial neural network) model works based on the ANN (artificial Neural Network) system which uses multivariate nonlinear input-output relationship functions to fit local cloud top temperature (Tb) to pixel rain rates (R). In this study, PERSIANN model and two interpolation methods (Kriging & IDW) were employed to estimate precipitation for northeast of Iran. The outputs from these three methods were used to evaluate wheat rainfed yield and compared. Results showed that correlation between PERSIANN model outputs and real yields records is significantly high in comparison with two other interpolation methods outputs.

The assessment of PERSIANN model as an alternative for interpolation methods to evaluate wheat rain-fed yield (A case study for northeast of Iran) Keywords:

PERSIANN model , Precipitation , Interpolation methods , Wheat rainfed yield