Climate Data to Predict Geometry of Cracks in Expansive Soils in a Tropical Semiarid Region

The nonlinear dynamics of the determining factors morphometric characteristics cracks in expansive soils make their typification a challenge, especially under field conditions. To overcome this difficulty, we used artificial neural networks to estimate crack Vertisol From July 2019 June 2020, soil (area, depth and volume), environmental (soil moisture, rainfall, potential evapotranspiration water balance) were monitored evaluated six experimental plots tropical semiarid region. Sixty-six events measured each plot calibrate validate two sets inputs multilayer network model. One set was comprised with significant correlations soil. other included only those high very correlation, reducing number variables by 35%. showed greater accuracy predicting characteristics, implying that it is preferable have fewer higher correlation than more lower Both data good performance area clay content above 30%. highest dispersion modeled over predicted values for all sand 40%. model successful evaluating from within its limitations may support decisions on watershed management view climate-change scenarios.