Evaluation of the effects of minimum winter temperatures on changing the type of cultivation and changing the use of agricultural lands. (Case study: Urmia county)

Land use and land cover change (LULC) and climate change are among the major threats to the global environment. Assessing the causes of land use change and its relationship with climate change is one of the important issues that understanding its process can help better human interaction with the environment. Therefore, the purpose of this study is to evaluate the effects of climate change on changing land use and land cover. The indicators used to achieve the mentioned goal are: ((average minimum winter temperatures), Number of days ( ≤ 0°C), Number of days ( ≤ -10°C) and 30-year Landsat satellite images)), CMIP5(CanESM2) model was used to predict temperature changes and CA-MARKOV model was used to predict land use changes and nally Pearson correlation coecient test was used to measure the correlation. The results of the study indicated that, there is a direct relationship between changes of minimum winter temperatures and changing the type of cultivation and land use in Urmia city. Also, simulation of temperature changes showed that there is the highest (> 0.8) correlation between rcp4.5 scenarios and land use changes, which indicates a high probability of changes in the specied time period (2018–2033). into reflectance using Fast Analysis for 2018 resolution


Introduction
Over the past century, global warming has been closely related to human activities, Climate change is expected to increase future temperatures, potentially resulting in change and reduced crop production in many key production regions. Temperature changes are one of the most important factors that affect climate changes; eventually, this also impacts land use/land cover (LULC) changes (Zoran, 2010). Change detection is useful in many applications related to land use and land cover (LULC) changes, such as shifting cultivation and landscape changes (Imbernon, 1999;Serra, Pons, & Saurí, 2008), land degradation and deserti cation (Adamo & Crews-Meyer, 2006;Gao & Liu, 2010), Changes in LULC are one of the fundamental concerns associated with changes in the global environment and sustainable development. Rapid worldwide population growth, accompanied by economic activities causing urban agglomeration and subsequent construction and land expansion, has led to rapid LULC changes (Guan et al., 2011;Halmy et al., 2015;Zheng et al., 2015). Although, different studies have been conducted in the eld of climate change and land use change, in most of these studies, the most important factor of land use change (climate change) has been given less value or most studies have focused on the effects of land use change on small-scale temperature changes. For example, (Darvishi 2018) examines the relationship between land use change and land surface temperature in Marivan and shows that increasing the area of residential land and reducing vegetation and agricultural land has increased the temperature in the study period. (pielke et al., 2011;Oleson et al., 2004). Other studies assessed land use change scenarios and modeling (Gollnow et al., 2018;Nazar Nezhad et al., 2018;Rajabzadeh, 2016;Prestele et al., 2017) or scenarios of climate change using simulation models (KyunPark et al., 2015;Huang et al., 2015;(Weinberger et al., 2017;Fix et al., 2018;Barredo et al., 2018). As mentioned, so far there have not been many specialized studies on the effects of climate change on changes of agricultural land use. Climate change and its effects on Urmia city, especially Lake Urmia and also increasing the area of orchards in the region are among the most important reasons for this study. Therefore, the present study is based on the hypothesis that the reduction of glacial periods and winter frosts has led to the expansion of orchards.
Accordingly, the purpose of this study is to evaluate the rate of winter temperature changes ((1. Minimum winter temperature (Average) 2. The number of days (≤ 0°C) 3. The number of days (≤-10°C)) changes in agricultural land use and type of cultivation in the town, simulate the future of these changes and nally to determine the degree of correlation between agricultural land use change and climate change.

Study Area
The Urmia County is located between longitude 37° 04 'N and 38° 01'N, and latitude 44° 36'E and 45° 16'E in West Azarbaijan Province in Iran. (Enayat Allah, 1988). In terms of its natural features, Urmia County encompasses an area of 1851.551 km2, which is about 15.8% of the province. Its altitude ranges from 1000 m to 3500 m, and its average slope is 9%. Its political and administrative divisions are as follows: The Urmia county located is south of the city of Salmas, north of Naghda, east of Lake Urmia, west of Turkey, and southwest of Iraq. Urmia County is divided into ve regions: Central, Anzal, Silivana, Somay Beraydost, and Nazlou; it has 20 rural districts (Ghesmati et al., 2015). Finally, according to statistics released by the Iranian Statistics Organization, in 2016, the county's population was 1,040,565 (www.amar.org.ir) (Fig. 1).

LULC data
Five Landsat images of different Sensor (TM, ETM+, OLI) for study area were downloaded from the US Geological Survey Earth Explorer site (https://earthexplorer.usgs.gov).The acquisition dates with clear sky conditions for the Landsat data are presented in Table1. We chose one scenes study area close to summer to consider seasonal effects, such as the phenology of vegetation, and to increase classification accuracy, as found by Bechtel et al. (2015). All Landsat images were first clipped covering study area and then atmosphericcorrected into scaled reflectance data using ENVI Fast Line-of-sight Atmospheric Analysis of Hypercubus (FLAASH). Nine of the 11 bands (bands 1-7, 10, and 11) in each Landsat 8 scene were used as input data. Based on the statistical characteristics of the different bands of the Landsat (TM, ETM+, OLI) images and a visual evaluation of selected band combinations, a false-color composite of (TM bands 5-4-3 and 4-3-2 (RGB)) and (ETM+ bands 3-4-7 and 4-5-7 (RGB)) was produced for the 1985-2010 images, also Seven of the 11 bands (bands 1-7) in Landsat 8 scene were used as input data for the 2018 image. Bands 1-7 in Landsat 8 sensor were the 30 m resolution Operational Land Imager (OLI) spectral bands.

LULC classification
Classification using satellite images is one of the current challenges in remote sensing. The use of remote sensing science has received a great deal of attention in areas where access to real data is difficult and impossible (Nogueira et al., 2017;Penatti et al., 2015;Xia et al., 2016). Recently, new algorithms and techniques have been introduced to categorize satellite imagery. It is important to examine and compare different classification algorithms in order to identify a suitable classification method in a particular region (Lu et al., 2007: 827). Currently, the support-vector machine (SVM) model is one of the methods most often used to classify images (Wang et al., 2013;Peng et al., 2017;Mirunalini et al., 2017). The SVM model finds a linear separating hyper plane with the maximal margin in this higher dimensional space. C >0 is the penalty parameter of the error term. Furthermore, K (xi; xj) = ∅(xi) T -∅(xj) is known as the kernel function. Although researchers are proposing new kernels, the following four basic kernels are listed in SVM books: In this current study, a communally used radial basis function (RBF) kernel was recognized during image classification (Rimal et al. 2018). Finally, we used the Kappa coefficient to determine the accuracy of the classification. 3.1.2. LULC Simulation Simulation of land use change is a process that predicts changes in the future based on older changes. There are several models for predicting future in land use changes. The most recent of these is the CA-Markov model (Rimal et al. 2018;Hernández-Guzmán et al. 2019;Surabuddin Mondal et al. 2016). While the Markov model can quantitatively predict the dynamic changes of landscape patterns, it is not good at dealing with the spatial patterns of landscape change. on the other hand, the Cellular Automaton (CA) model has the ability to predict any transition among any number of categories (Gil et al., 2005). The important property of the CA model is that it demonstrates spatial and dynamic processes, which is why CA has been broadly used in land use simulations (Ye et al., 2008). Moreover, the state of each cell depends on the spatial and temporal state of its neighbors (Reddy et al., 2017). Combining the advantages of CA theory and the space layout forecast component of Markov theory, the CA-Markov model performs better than the Markov model or the CA model for modeling land cover changes for both time and spatial dimensions. At present, IDRISI software is one of the best platforms on which to run the CA-Markov model. That software was developed in the United States (US) by Clark Labs. Therefore, in the present study, the ca-Mrkov model in IDRISI software was used to predict the rate of land use change. Land uses are classified into seven categories (sandy soils, rocky soils, saline soils, residential and built-up lands, pastures, rainfed agriculture, irrigated arable lands, and orchards). The 18-year period of 2000-2018 was chosen to better and more accurately predict the amount of change in the period 2018-2033.

Historical climate data
To determine climate change we used the minimum temperature index. The minimum daily temperature data was obtained from the Urmia Synoptic Station for the period ranging from 1977 to 2017 (Mehrizi Zarea, 2018;Yousefi et al. 2018). Evaluating the region's Feature climate data was based on three indicators: 1. Minimum winter temperature (Average) 2. Number of days (≤ 0 °C) 3. Number of days (≤-10 °C).

Climate change simulation
General circulation models are the most important methods for forecasting global and continental global climate variables. This study used the general circulation model to predict the climate variables of the Canadian model (CanESM2). This model is the only model that has the daily data available for SDSM software for three scenarios: RCP2.6, RCP4.5, and RCP8.5 (Thomson et al., 2011;IPCC, 2014) (Table 2). Therefore, Canesm2 model in sdssm5.1 was used to predict changes in the minimum winter temperature in the period (2018-2033).

Correlation coefficient
A correlation coefficient is a numerical measure of some type of correlation, meaning a statistical relationship between two variables (Hoseini, 2003). Thus, we used the Pearson correlation coefficient in SPSS software In this study, satellite images from 1985,1992,2000,2010,2017 for seven land use categories: Sandy soils, Rocky soils, saline soils, built-up land, Pastures, rainfed agriculture, Orchards and Irrigated arable lands (Figure 3).
In the process of conducting the research study, the overall accuracy coe cient and the Kappa coe cient were used to determine the accuracy of the image classi cation (Table 3).
As seen in the information presented in Table 3, the overall accuracy of the classi cation and Kappa coe cient for all the studies years is >80%, and that result is con rmed. Accordingly, the rate of land use change during the studied years is shown in table 4.
The results of the land use change assessment show that most of the changes are related to Sandy soils and Orchards uses; the Sandy soils area decreased from 864 km 2 in 1985 to 460 km 2 in 2018, which represents a decrease of -0.077% of the area over a period of 33 years. In contrast, the use of Orchards increased from 131 km 2 in 1985 to 295 km 2 in 2018 (an increase of 0.032%in this area over a period of 33 years) Also, according to the classi cation, land use area of the built-up land and rainfed agriculture has increased, and Rocky soils and saline soils have decreased.

LULC Simulation
The CA-Markov model was used for land use and land cover changes simulation. A Transitional probability matrices produced by Markov model that prepare details on the probability of transition between LULC types during the periods 2000-2018 and 2010-2018 (Table 5). For example, the results of the model show that the probability of converting irrigated agricultural lands to orchards in the period 2000-2018 is about 25%, which has decreased by 2% in the period 2000-2018 to about 23%, and this shows that in the long run, the conversion rate of irrigated agricultural lands into orchards has been higher.
According to the prediction analysis, Rainfed agriculture areas will occupy 195.48 km 2 in 2025 and 196.74 km 2 in 2035. Rocky soils area is estimated to decrease to 176.54 km 2 , and 166.36 km 2 , in 2025 to 2033 (Fig 4). The results also show that Sandy soils and saline soils has been reduced by 28.20% and 1.23%, and the increased (Pastures, Built-up Land, Orchards, Irrigated arable lands) by (14.42%, 6.56%, 10.47%, 6.97%), respectively, in the period 2025-2033.

Simulation minimum temperature change
After checking and controlling the quality of observational data, predictive CanESM2 variables were selected which most closely correlated with observational data.

Correlation test between climate variables and land use and land cover change
According to Fig 8, there is a correlation between land use changes and variables (Minimum winter temperature (Average), number of days (≤ 0 °C), and number of days (≤-10 ° C)). More precisely, there is the highest correlation coe cient between changes in land use Rainfed agriculture, Orchards) and climate variables (Minimum winter temperature) with correlation coe cients (0.96%, %.941%) and sig (0.001%, 0.002%) respectively. Also, there is an inverted correlation between the variables (Minimum winter temperature) and land use (Sandy soils, saline soils) with correlation coe cients (0.861%, 0.843%) and with sig (.013%, .017%) respectively. Finally, there is the highest correlation by correlation coe cient (%.867) and sig (%`.012), between land use (Rocky soils) and (number of days (≤-10 ° C) variable.

Conclusion
This study was conducted with the aim of evaluating the relationship between climate change (minimum winter temperatures) and change in type of cultivation and agricultural land use, using satellite images classi cation and temperature data of Urmia Synoptic Station in the period  and simulating the rate of changes of agricultural land use and temperature patterns in the period (2018)(2019)(2020)(2021)(2022)(2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030)(2031)(2032)(2033). The results showed that the most land use changes include: increasing the land use areas of orchards and rainfed agricultural lands and reducing the land use areas of sandy and rocky lands. The study of temperature change scenarios also showed an increase in the average minimum winter temperature, so that the Rcp4.6 scenario had the highest correlation with a coe cient of more than 0.8% with land use changes. According to the proposed scenario, while the minimum winter temperature (average) has increased, the land use level (sandy, rocky, and saline soils) has decreased and the use of rainfed arable lands and orchards has increased. The increase in the average winter temperature in the study area has caused a large part of irrigated agricultural lands to be turned into orchards. Meanwhile, the average winter temperature has increased and the number of frost days (number of days (≤ 0 °C)) has decreased, which has led to a tendency to cultivate fruit trees, especially apple trees in the study area.
In addition, increasing the minimum winter temperature reduces the density of mountain glaciers and increases the permanent evolution of the snow line, and as a result, in the period 1977-2018, a large area of mountainous lands (pastures) turned into rainfed agricultural lands. Therefore, the most important results of surveys and forecasts of land use change and temperature change in the study area are: 1. Increasing the area under cultivation of orchards 2. Increasing the minimum winter temperature 3. Change in rainfall regime 4. Reducing the density of mountain glaciers Therefore, in the near future, these changes are expected to lead to water shortages and water stress in the area. Hence, it is recommended to prevent oods and soil degradation by using sustainable agricultural patterns such as: drip irrigation, and river boundary determination.

Declarations Con ict of Interest
It is the policy of the Journal Theoretical and Applied Climatology to ensure balance, independence, objectivity, and scienti c rigor in the Journal. All authors are expected to disclose to the readers any real or apparentcon ict(s) of interest that may have a direct bearing on the subject matter of the article. This pertains torelationships with pharmaceutical companies, biomedical device manufacturers or other corporations whoseproducts or services may be related to the subject matter of the article or who have sponsored the study.The intent of the policy is not to prevent authors with a potential con ict of interest from publication. It ismerely intended that any potential con ict should be identi ed openly so that the readers may form theirown judgements about the article with the full disclosure of the facts. It is for the readers to determinewhether the authors' outside interest may re ect a possible bias in either the exposition of the conclusions presented.
The corresponding author will complete and submit this form to the Editor-in-Chief on behalf of all authors listed below.

Corresponding author
Correspondence to Ghasem Farahmand.

Funding Statement
The author(s) received no speci c funding for this work.

Data and code availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
The data for this project are con dential, but may be obtained with Data Use Agreements with Email to these project researchers. Researchers interested in access to the data may contact [Ghasem Farahmand] at [farahmand.geo@gmail.com], It can take some months to negotiate data use agreements and gain access to the data. The author will assist with any reasonable replication attempts for two years following publication.
Ethical Statement I testify on behalf of all co-authors that the article was submitted: this material has not been published in whole or in part elsewhere; the manuscript is not currently being considered for publication in another journal; all authors have been personally and actively involved in substantive work leading to the manuscript, and will hold themselves jointly and individually responsible for its content. In addition, I consent to: audio-visual recording of any part of or all research activities (if applicable) publication of results from this study on the condition that my identify will not be revealed.      Figure 1 Location of study Area. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 2
The framework of the study Figure 3 Land-use/land-cover (LULC) Map from 1985 to 2018. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.