Combinação de modelos de balanço hídrico no solo e sensoriamento remoto para o monitoramento de áreas irrigadas

Abstract

This work uses remote sensing (RS) and geographic information system (GIS) techniques for the support of irrigation management and crop monitoring of center pivot irrigated areas with particular emphasis on the estimation of the basal crop coefficient (Kcb). Besides the traditional meteorological approach, it can be seen that the use of information from moderate spatial resolution sensors (TM/Landsat) is coming forward, especially in the last two decades. The remote sensing information, implemented in energy balance models of the soil surface as SEBAL or METRIC, or assimilated and correlated with soil water balance models as SIMDualKc can provide estimations of crop coefficient (Kcb) and evapotranspiration (ET) which are closer to local conditions, due to pixel level spatial information. The organization of information related to irrigation management in GIS environment provides the means for efficient visualization of the agricultural cycle dynamics in several levels. Cross-linked information allows the understanding of vegetation and soil characteristics, together with rainfall and irrigation effects and the crop water demand. The GIS database created in this work helped to identify: (i) rainfall spatial distribution; (ii) crops and phenological stages, by the normalized difference vegetation index (NDVI); (iii) land use; e (iv) erosion risk and agricultural potential. The NDVI showed a sensitivity of 0.02 units for the identification of phenological stages and crop cycle features for central pivot irrigated soybean and maize in the typical conditions of southern Brazil. The resulting FAO56-like crop growth stages for maize and soybean were, respectively: [0.0-0.4] and [0.0-0.3] for the initial period; [0.4-0.75] and [0.3-0.85] for the rapid growth period; [0.75-0.1] and [0.85-1.0] for mid-season period; [0.75-0.3] and [0.85-0.3] for late season period. The average relative error (ARE) was around 7%. The curves also showed a kind of fingerprint of the crop type and management practices in the region that could be associated with the phenological stages in the growing season, as a good tool for agricultural monitoring. The assimilation of NDVI data to Kcb was made through the correlation equation between the Kcb output of a FAO56-like soil water balance model (SIMDualKc) and the obtained a Kcb NDVI assimilated function. The actual irrigation coefficient Kcb act NDVI was obtained through the product of the assimilated Kcb potNDVI with the stress coefficient (Ks) output of the SIMDualKc model. The average relative error (ARE) between the assimilated general KcbNDVI curve and the individual pivot curves was lower than 30% for both potential and actual Kcb. The results showed that the assimilation of NDVI for the calculation of Kcb with the methodology proposed can potentially benefit the irrigation management with a better adjustment of the values to the actual condition of the crop during the growing season. This can be a useful tool for the determination of the water demand of soybean and maize in irrigated fields in Brazil. The methodology may also be adequate as a base to be adapted for unmanned air vehicles based monitoring with NDVI sensors.