Comparing the Köppen-Geiger, Feddema, and UNEP Climate Classications: An Application to Iran

: 38 This study introduces the climates of Iran defined by Köppen- Geiger, Feddema’s, and 39 UNPEP classifications that applied to a high-resolution ground-based gridded data set relative to 40 the 1985 – 2017 period. Ten Köppen-Geiger climate types were found for Iran, from which Bwh, 41 Bsk, Csa, Bsh, and Bwk cumulatively account for more than 98% of the territory. Likewise, from 42 36 possible Feddema’s climate types, Iran possesses fifteen climate types from which the Dry 43 Cool, Semiarid Torrid, Semiarid Hot, Semiarid Warm, Dry warm, Semiarid Cool, and Moist Cool 44 climates collectively occupied approximately 93% of the country. Similarly, arid, semi-arid, 45 humid, and sub-humid UNEP climate types characterized more than 98% of Iran. A few other 46 vertically stratified climates appeared at the highlands of Iran just because of changes in elevation 47 and slope aspects of the mountains. The combined effect of topography and vicinity to sea also 48 creates very distinct climate types in northern Iran. The climate maps of the three used methods 49 reflect the joint effects of topography, latitudinal variation, and land/sea surface contrast on the 50 climate of Iran. A pairwise comparison made between the three classifications showed a 51 satisfactory agreement between the three schemes in representing the main climate types of Iran. 52 53


Introduction
Climatologists have long been used climate classification systems to classify the earth 68 into different climatic zones based upon important climatic variables like precipitation 69 and temperature. The modern climate classifications rooted back to Köppen (1900), De 70 Martonne (1926), and Emberger (1932) who have considered precipitation and air 71 temperature as the key variables for climate classification. Undoubtedly, the climate 72 classifications proposed by Köppen (1900) and Thornthwaite (1948) are the most 73 physically based classification systems. 74 Based on monthly total precipitation and average temperature and their seasonal 75 variations, Köppen (1900) has divided the Earth into five major climatic regions as  Geiger (1930), Trewartha (1980), Guetter and Kutzbach (1990), and Stern et al. (2000) 88 have proposed several modifications to Köppen classification. 89 Many scientists like Thornthwaite (1948) sought to provide a better way to classify 90 climates so that they would have both a stronger physical basis and a more accurate 91 demarcation of climatic zones. Thornthwaite (1943) first used the concept of the moisture 92 index to develop a classification system structured around the moisture factor and then 93 modified it with improved water balance metrics (Thornthwaite 1948). He also made the 94 system more "rational" by using even class intervals (Feddema 2005). Although 95 Thornthwaite climate classification has a very high scientific validity, the Köppen 96 classification method is still the most popular climatic classification method for drawing 97 climate atlases at major centers for climatology around the world (FAO 1977). Köppen 98 and Geiger (1930) manually drawn the first Köppen-Geiger climate classification map 99 that was used in many textbooks of climatology to introduce the climate types of the 100 world and teach students how to classify climates (Kottek et al. 2006). The need to update  107 Köppen-Geiger method. Rubel and Kottek (2010), using the CRU monthly temperature 108 and GPCC monthly total precipitation data, presented another map of the  climates for the globe and examined the displacement of the world's climatic regions due 110 to climate change. 111 The spatial accuracy of each of the aforementioned maps on a regional scale depends 112 on the quality of the data and the number of stations used in that area. Due to the low 113 spatial resolution of the data used in the above studies, it is necessary to use a denser 114 network of meteorological stations to provide a more accurate map of  climate classification at the regional and national levels. In this regard, Stern et al. (2000) 116 have provided a climate classification map for Australia using the data of the Australian  Unlike the Köppen classification system, the Thornthwaite climate classification has 128 never been used for producing a world map despite John "Russ" Mather frequently 129 discussed the idea of producing such a map with the Thornthwaite climate classification 130 (Feddema 2005). It is only applied to the United States, partly because Thornthwaite did 131 not have access to sufficient climate data to do a global study (Feddema 2005), and since 132 the Thornthwaite full classification was too complex none of his associates or other 133 researchers attempted to produce a Thornthwaite global climate classification map. a less complex and more systematic climate classification approach that is easier to 136 interpret and convey to students in a classroom setting (Feddema 2005  Daily total precipitation, as well as maximum and minimum air temperature relative 171 to a varying number of meteorological stations regularly distributed over Iran, were used 172 for gridding the data into regular grid points with 0.25 × 0.25 degree spatial resolution.  and very useful when there is not strict confidence on the data measurements (Raziei and 194 Pereira 2013b). It also produces visually appealing surface maps from irregularly spaced 195 data to express existent trends in the dataset, so that high points might be connected along 196 a ridge rather than isolated by bull's-eye type contours (Raziei and Pereira 2013b). Since

Köppen-Geiger climate classification 228
The Köppen (1900) climate classification system that was modified by later  Although the Köppen classification passed several revisions, including its final version 240 provided by Geiger (1961), its main shortcoming is that the boundaries between certain 241 climate types do not correspond with the observed boundaries of natural landscapes. This 242 led Trewartha (1968) and Trewartha and Horn (1980) to introduce more realistic criteria 243 to distinguish between the B and C climate types to match better the climate types of the 244 mid-latitudes with the vegetative zones. Although Trewartha's scheme is more reflective 245 of ecosystem variations, the Köppen-Geiger scheme is widely used because it is a 246 simplified version of its original classification. Therefore, the Köppen-Geiger 247 classification briefly described below is used herein to divide Iran into different climates.
248 Table 1 presents the first two letters used in the Köppen-Geiger classification and the 249 criteria with which the main climates are defined (Kottek et al. 2006). In Table 1, Tann is  Table 2 wherein the mean monthly temperature is denoted 262 by Tmon. Based on the criteria listed in Table 1 and Table 2 where P is total annual precipitation and PET is total annual potential evapotranspiration.

277
The Im values range from −1 to +1, where a value of 0 indicates that the annual moisture 278 supply (P) is equal to the annual moisture demand (PET) (Elguindi et al. 2014).

279
Monthly PET required for estimating Im in Eq. 2 is estimated using the well-known 280 Thornthwaite method (Thornthwaite 1948 To account for variable day and month lengths, the computed PET is adjusted as in Eq.  classes that range from low to extreme (Table 4). Since the Im varies as a function of both 307 PET and P variability, it simultaneously reflects the inherent seasonality of both variables.

309
In this classification, the seasonal moisture variation is estimated by calculating the 310 annual range of precipitation variability defined as the difference between maximum monthly precipitation (Pmax) and minimum monthly precipitation (Pmin), i.e., Pmax -Pmin.

385
Considering that the combined effects of precipitation, temperature, and PET are the Iran where the altitude of the locations substantially increases. Figure 5 also  The present climate classification is highly consistent with that created by Raziei        The UNEP climate classification of Iran shown in Figure 8 is very similar to 536 Feddema's climate classification illustrated in Figure 7b. This is not a surprise if one notes 537 that both the UNEP and Feddema's climate classifications are based on the water balance 538 concept computed with the Thornthwaite (1948) method. However, Feddema's climate 539 classification presents more details than the UNEP classification since it benefits the 540 combined effects of thermal and moisture factors to categorizes the climate of the world 541 into 36 possible categories. Table 9 presents the areal extents of the UNEP climate types 542 of Iran identified in the present study. Based on the UNEP climate classification more 543 than half of Iran is arid (52.85%), 34.26% is semi-arid, and 6.95% has a dry climate. The 544 humid and sub-humid climate types also occupy 3.98% and 1.65% of the country.  Arid climate type mostly agrees with Bwh climate but it may find the Bwk and Bsh 578 climate types as its counterpart at less than 30% of the grid points. The semi-arid climate 579 also mostly coincides with Bsk but it may match Bsh or Csa at less than 30% of the grid 580 points. As is seen, the Dry and Semi-dry climates of the UNEP well coincide with the 581 Csa climate type of the Köppen-Geiger climate classification whereas its Humid climate 582 type mostly concords to Csa (>80%); however, it matches the Csb climate type at less 583 than 20% of grid points. climates also correspond to the UNEP Arid climate type in most of the grid points they 588 occupied but they match with Semi-arid climate at only less than 30% of the grid points. 589 Feddema's Dry Torrid climate type mostly agrees with the Semi-arid climate of UNEP 590 but corresponds to the Arid climate at less than 50% of the grid points. The Dry Hot   Due to the complex topography and wide latitudinal and longitudinal extent of Iran (Figure 1) a dense network of meteorological stations is required for creating a spatially informative climate classi cation for the country. By using a gridded data set created based upon a dense network of regularly distributed stations over the complex topography of Iran (Figure 1), the present study aims to identify the climate types of Iran with the Koppen-Geiger, the UNEP, and the modi ed Thornthwaite climate classi cation schemes proposed by Feddema (2005). The rst two climate classi cations are used to update the climate classi cation maps of Iran created by Raziei (2017) and Raziei and Pereira (2013a); (Raziei and Pereira 2013b) while the third climate classi cation is used herein for the rst time to create a new climate classi cation map for Iran based on improved water balance metrics (Thornthwaite 1948).      shows the spatial variation of the annual total precipitation over Iran. As depicted, the highest annual total precipitation is observed over the coastal areas of the Caspian Sea and some parts of the Zagros mountains in western Iran while central, eastern, and southern Iran is characterized by far lower annual total precipitation. While the total annual precipitation of central-eastern Iran is almost less than 250 mm, it is between 350 and 950 in the mountainous areas of western Iran and over 1000 mm in the central and western parts of the coastal areas of the Caspian Sea. Figure 4a also shows the signi cant role of topography and vicinity to water bodies in shaping the spatial pattern of precipitation over Iran. Figure 4b also shows the spatial variability of PET over Iran, computed with the Thornthwaite method for the period 1985-2017. This map also clearly shows the role of topography in representing the spatial distribution of PET over Iran. As is seen, southeastern and southwestern Iran show the highest PET values that range between 2200 and 3400 mm, followed by central-eastern Iran wherein the PET is between 1300 and 2200 mm. The lowest PET values ranging between 400 and 1300 mm correspond to the mountainous areas of western and northern Iran that are characterized by the lower annual average temperature, annual maximum temperature, and annual minimum temperature as depicted in Fig.4a through Fig.4c).

Figure 5
According to Figure 5, the temperate Csa climate characterized by dry and very hot summers dominates a large part of the mountainous region of Alborz in the north and Zagros in the west of the country. Dsa climate type characterized by cold and snowy winters and dry and very hot summers appears at the highlands of Zagros in the west of Iran where the altitude of the locations substantially increases. Figure   5 also shows that some parts of the Alborz ranges in the north have a Dsb climate that has a cooler summer compared to Dsa climate type observed over Zagros ranges. A small part of the western coastal areas of the Caspian Sea has Csb temperate climate type that is hot and dry in summers. Likewise, the Cfa climate type which is a temperate rainy climate with no dry season but hot summers characterized two spots areas over the Alborz ranges in the north. According to Figures 5, from 31 possible Köppen-Geiger climate types that can be found over the earth, ten climate types were found in Iran, from which only Bwh, Bsk, Csa, Bsh, and Bwk with 35.98%, 23.69%, 17.03%, 15.70%, and 5.94% occupied area cover very large parts of the country (Table 7). The Cfa, Cfb, Csb, Dsa and Dsb climate types with less than 1% areal coverage (Table 7) are created under the strong in uence of altitude, latitude, vicinity to water bodies or a combination of them.      Feddema's Semiarid Torrid, Semiarid hot and Semiarid Warm climates match with Bwh climate type whereas a minor proportion of the grids correspond to Bwk or Bsh climate types. The most proportion of Feddema's Semiarid Cool climate (50%-60%) corresponds to Bwk climate while the rest (40%-50%) assigned to Bsk climate. The Feddema's dry torrid climate mostly matches with Bsh (>80%) while a minor proportion corresponds with Csa climate. With a relatively similar proportion, the Dry Hot climate is either match with Bsh or Csa climate types. Dry Warm climate mostly agrees with Bsk climate but a noticeable proportion of it may either match with Bsh or Csa climate types. Feddema's Dry Cool and Dry Cold