Detection of precipitation phase in hourly and daily intervals in northwest Iran

One of the significant impacts of climate change is the change in the type of precipitation that occurs in temperate and mountainous areas. Liquid or solid precipitation in form of frozen rain or snow can stop ground transportation, cause power outages, and determine the speed of the basin's response to floods and potential dangers to water resources in drainage basins especially in mountainous areas and, in particular, impose freshwater resources on rivers and glaciers. The type of precipitation can be just as important as the intensity and the amount of precipitation in the seasonal hydrological cycle and the health of ecosystems in high latitudes. The potential benefits and disadvantages of precipitation prediction depend on the form and intensity of precipitation, and the incorrect forecast of precipitation phase can cause problems in managing many areas, including water storage, air and soil moisture, land albedo, and surface currents. Changes in precipitation have more direct impact on society than changes in other meteorological elements. However, it is very difficult to determine the precipitation characteristics of an area due to its temporal and spatial fluctuations. Therefore, an important issue in modern hydrology is to determine the effects of climate change on the share of liquid and solid precipitation and its statistical distribution. Although many studies have been conducted so far to examine precipitation characteristics, changes in the precipitation phase have received less attention and studies in this field have often used daily data to detect and predict the precipitation phase. Atmospheric conditions on a scale of less than one day can also change the phase of precipitation. On the other hand, incorrect forecast of the precipitation phase can affect the environmental dimension of an area, especially its hydrological and climatic conditions. Relying on a specific element to detect the precipitation phase can lead to uncertainties due to the complexity of the precipitation process. Therefore, the present study was conducted to identify the precipitation phase in 19 stations in northwestern Iran with KSS model. The data used include the average of temperature, relative humidity, snowfall, and rain in two time intervals of 3 and 24 hours during the statistical period of 1951-2018. The model was executed for hourly and daily data and the accuracy of the model was evaluated with POD, CSI, PC, TSS, FAR, and FBI indices. The results were divided into two groups of solid and liquid precipitation and then the accuracy of the model was evaluated using 6 indices. Based on the indicators used, the model's performance in detecting both types of precipitation in the region is very high to the extent that the average evaluation of indicators in most cases is over 90%. The results of this study are consistent with the studies of Koistinen and Saltikoff, where the POD values for liquid and solid precipitation are 0.81 and 0.97 and Gjertsen & Ødegaard for solid and liquid precipitation are 0.97 and 0.84, respectively. Another important goal of this study was to compare the effect of time scale data on the accuracy of the model. The results showed that the use of 3-hour data in detection of the liquid phase of precipitation increases the performance of the model. Based on POD, CSI, PC, and TSS indicators, the model accuracy is 0.92 with hourly data and 0.85 with daily data. While the detection of the solid phase of precipitation in some cases offers contradictory results compared to liquid precipitation and the performance of the model with 24-hour data does not affect the accuracy of the model. Even in the POD index, a slight increase is seen compared to hourly data. The impact of data scale reaches its highest value with FAR index so that using daily data, -350 and 40% change is observed in detection of solid and liquid precipitation compared to the hourly data. On the other hand, the behavior of the model was different in some stations such as Sardasht and Khalkhal stations where the performance of the model was increased or decreased compared to the change of data time intervals, while no change was observed in Sarab station. Another result of the study shows that according to POD, FBI, and CSI indicators, the performance of the model in detecting hourly liquid rainfall is higher than solid hourly rainfall. Moreover, POD, FBI, and TSS metrics also showed an increase in the model's ability to detect solid daily precipitation relative to daily liquid precipitation. Because the study area is one of the areas affected by local masses and short time scale, its prevailing precipitation is in form of liquid precipitation in spring and autumn, while solid precipitation is more the result of migratory and permanent air masses in the region; therefore, based on the results of model evaluation, it can be concluded that the performance of the model in this area of Iran is quite acceptable.