INTRODUCTION
There is a high spatial and temporal climatic variability in the Andean regions of Colombia, and it can be analyzed in the short term, when the information dates only a few years back, or the long term, when information for a longer period is available. Both types of analysis allow characterizing and knowing the patterns of climatic variables in space and time (Martínez & Salas, 2015). Knowing the spatio-temporal variability of the main climatic variables on its intra-daily and intra-annual scale in a tropical Andean city is an urban planning tool that strengthens decision-making. Precipitation is fundamental since it directly defines the water supply or availability and is considered one of the most important mechanisms in the global hydrological cycle (Marín et al., 2005). Therefore, knowing this variability becomes a water heritage management tool, where its proper management involves agricultural practices, human and hydroelectric consumption, the ecosystems, and other industrial or mining uses. Proper management and planning thus require implementing actions that minimize the impact of water use, so it is necessary to analyze and detect historical modifications on the spatial and temporal characteristics and behaviors of the pluviometric regime (Puertas, Carvajal, & Quintero, 2011).
Understanding the diurnal rain cycle is important in the short term for many reasons: the main one is because the amplitude and phase of this cycle are the result of the interaction between dynamic and radiative processes. The successful explanation of such processes is a useful measure to make sense of atmospheric physics at these time scales (Zuluaga, Poveda, & Mejia, 2004), which allows understanding regional patterns of climate behavior. This knowledge is basic for the physical parameterization of climate models (Gui-Ying & Slingo, 2001). Some factors that directly affect local atmospheric dynamics and cause changes in diurnal precipitation cycles are orography, relative humidity and thermodynamic processes (Poveda et al., 2002). Due to its importance, the diurnal cycle of precipitation has been studied for a long time (Albright, Mock, Recker, & Reed, 1981; Gray & Jacobson, 1977; McGarry & Reed, 1978). In Colombia, precipitation has been extensively studied on monthly or longer periods, while the daytime cycle has been little studied due to the scarce information available from ground stations until relatively recently. The studies carried out to date (Bedoya-Soto, Aristizábal, Carmona, & Poveda, 2019; Mapes, Warner, Xu, & Negri, 2003a, 2003b; Poveda et al., 2005; Snow, 1976; Suárez-Cobián, 1959; Trewartha, 1981; Trojer, 1959; Zuluaga et al., 2004) have allowed to recognize fixed patterns as a bimodal behavior in the diurnal cycle of precipitation in the western zone of the central mountain range of Colombia, with peaks in the afternoon and towards midnight, driest periods between 9:00 - 11:00 h, and highly variable cycles between contiguous stations.
To make up for the lack of ground stations that allow such research to be carried out, the latest studies have concentrated on the use of satellite or radar information on both cloudiness and precipitation. Thus, studies regarding the frequency, intensity and diurnal cycle of rain use satellite information from different sources, such as CMORPH (Climate Prediction Center Morphing technique), PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks), TRMM3B42 (Tropical Rainfall Measuring Mission daily and sub-daily (3hr) averages precipitation), which in turn have been compared to ground records. The results indicate that the spatial distribution is comparable on monthly scales, yet diurnal scales deviate in up to 30% (Dai, Lin, & Hsu, 2007). Rozante et al. (2018) reach similar conclusions, identifying that the satellites Itegrated Multi-satellite Retrievals for Global Precipitation Measurement (IMERG) and Global Satellite Mapping of Precipitation (GSMap) overestimate diurnal rainfall in northern Brazil when compared to TRMM, reason why they conclude that the GSMaP satellite could replace the TRMM which will soon go out of orbit (Rozante, Vila, Barboza Chiquetto, Fernandes, & Souza Alvim, 2018). Most of these studies have concluded that, in the tropics, maximum convection or precipitation tends to occur late at night or in the early morning (Gui-Ying & Slingo, 2001).
Finally, the intra-hourly distribution of precipitation is analyzed by means of a frequency analysis. The temporal distribution of events is important because it has been established that the hydrological response and the maximum peak of runoff flow depend directly on the maximum intensity of precipitation (Adams & Howard, 1986; Ball, 1994; El-Jabi & Sarraf, 1991; V. P. Singh, 1997; Šraj, Dirnbek, & Brilly, 2010). Therefore, adequately estimating the design storm allows to obtain design floods for hydraulic infrastructure. Design storms must be based on the dimensionless shapes observed in each particular region, which is why their analysis has been widely carried out in the USA (Chow, Maidment, & Mays, 1994) and in different parts of the world using the data from the registers, in order to replicate design storms for hydraulic infrastructures (Azli & Rao, 2010; Huff, 1967; S. C. Singh et al., 2017).
This research aims to characterize the rain events and the diurnal cycle of precipitation in the city of Manizales, a tropical Andean city, taking into account the hydrometeorological 2010-2018 data of the city stations of the network of the Risk Management Unit (IDEA UN Manizales & CORPOCALDAS, 2015), a network that belongs to the Caldas Integrated Environmental Monitoring System (SIMAC because of its name in Spanish), where an effort is being made to improve meteorological information that allows for this type of analysis. The SIMAC is a recent system and, although it now has 124 installed hydrometeorological stations, only 13 of these stations allow simultaneous analysis of the last 9 years. It is very important to highlight that the interannual and multiannual analysis of climate variability and climate change was not performed, basically because there are not yet long enough series for a complete and detailed analysis of this type of variability within Manizales. This analysis is still relevant given its contribution to the knowledge of short-term temporal-spatial variability in the city.