Model to Estimate Available Water for Coffee Crop in Colombia.

The coffee area in Colombia comprises near to 880.000 hectares distributed along the three mountain ranges that cross the country from south to north. One characteristic of cropping in these areas is the rainfed condition. Knowledge of soil moisture distribution in the coffee area in Colombia is very important because water stress is necessary for floral bud induction and conditioning  previously to anthesis, besides  the bean filling requires high soil water availability. Soil moisture distribution is also important for fertilizer and other agronomic practices scheduling. Although, climatic water balances have been estimated for Colombian coffee areas, they do not provide precise estimation of the available water for crops because they do not consider processes like runoff, rainfall interception, crop evapotranspiration, and soil moisture distribution. Several field studies were conducted from 1997 to 2008 to measure these components and this information was utilized to develop a water mass balance based agrometeorological model. The model includes the soil physical characteristics like field capacity, wilting point and porosity, rainfall interception, runoff and crop evapotranspiration (ET_c) estimations. A meteorological model for the reference evapotranspiration was previously calibrated. The model  considered  an adjustment at ET_c by two possible ways: adjusting the surface resistances in the Penman-Monteith model  with the previous day moisture or using an adjusted  crop stress coefficient (K_s) in the FAO´s  two step approach. A variable crop coefficient (K_c) was also considered, according to  the age of the crop and plant density. The model include a water deficit and excess derivation with the aim to predict magnitude of the floral bud induction and the susceptibility to the water deficit and excess.   The model was tested with field moisture measurements using reflectometry (TDR) and oven dry samples in two experimental fields (Cenicafé and El Jazmín) during the periods 1997-1998 and 2003-2004.  The model performance was evaluated with the Willmott index (D) and linear parameters. At the Cenicafé plot, the D-index was 0.78 for 1997 and 0.85 for 1998, and in the El Jazmín plot, the average 2003-04 D-index was 0.86. The advantage of this model is that it includes the soil variability, crop and meteorological conditions, and can be applied to all Colombian coffee regions to monitor soil moisture seasonal changes.