A remotely sensed study of the impact of meteorological parameters on vegetation for the eastern basins of Afghanistan

Despite the importance of the Amu Darya and Kabul River Basins as a region in which more than 15 million people live, and its vulnerability to global warming, only a few studies addressed the issue of the linkage of meteorological parameters on vegetation for the eastern basins of Afghanistan. In this study, data from the MODIS, Global Precipitation Measurement Mission (GPM), and Global Land Data Assimilation System (GLDAS) was used for the period from 2000 to 2021. The study utilized several indices, such as Precipitation Condition Index (PCI), Temperature Condition Index (TCI), Soil Moisture Condition Index (SMCI), Vegetation Condition Index (VCI), and Optical Integrated Drought Index (OIDI). The relationships between meteorological quantities, drought conditions, and vegetation variations were examined by analyzing the anomalies and using regression methods. The results showed that the years 2000, 2001, and 2008 had the lowest vegetation coverage (VC) (56, 56, and 55% of the study area, respectively). On the other hand, the years 2010, 2013, 2016, and 2020 had the highest VC (71, 71, 72, and 72% of the study area, respectively). The trend of the VC for the eastern basins of Afghanistan for the period from 2000 to 2021 was upward. High correlations between VC and soil moisture (R = 0.73, p = 0.0008), and precipitation (R = 0.63, p = 0.0014) were found and also significant correlation was found between VC and drought index OIDI. It was revealed that soil moisture, precipitation, land surface temperature, and area under meteorological drought conditions explained 45% of annual VC variability. It was also found that the orography had a significant influence on both the spatial distribution of the LST and VCI, as well as the spatial correlations between VCI and meteorological parameters.


Introduction
Abnormal climatic conditions related to climate change have been associated with the effects of human activities over the past few decades. They lead to numerous environmental and ecological problems, such as air pollution, dust storm, biodiversity loss, soil erosion, and vegetation degradation (Li et al. 2022a;Zhao et al. 2022b;Zhou et al. 2021). Therefore, the knowledge of how climate change affects different ecosystems has an important role in the protection and management of vegetation cover (Tian et al. 2021b).
Vegetation occupies almost half of the planet and plays an important role in providing food, fiber, and fuel, supporting animal biodiversity, maintaining climate quality, and supporting ecological processes that preserve ecosystems and landscapes (Hu et al. 2022;Li et al. 2021;Li et al. 2022b;Yang et al. 2022). Vegetation is one of the important components of the terrestrial ecosystem, which plays an effective role in preventing desertification and also plays a key role in providing various ecosystem services to adapt and mitigate climate change; Additionally, every change in vegetation affects the climate of the region, especially temperature and air quality, through its influence on net radiation, energy partitioning, conversion of precipitation to runoff Liu et al. 2020;Mansourmoghaddam et al. 2022b;Zhu et al. 2022), soil moisture (Zhang et al. 2019b), evaporation, and transpiration (Najmuddin et al. 2017). Since global climate change has become a major topic of discussion today, the relationship between vegetation and meteorological factors is of great importance in ecological studies (Gao et al. 2012).
Remote sensing can continuously and systematically deliver information on the water cycle and vegetation variations and therefore, remote sensing drought indicators can be used for spatial and temporal drought monitoring (Mansourmoghaddam et al. 2022a;Wang et al. 2021;Xie et al. 2021aXie et al. , 2021bYin et al. 2022b). Remote sensing is of particular importance in applications requiring actual and constantly updated information. Due to various spectral ranges and data availability, the use of remote sensing data is one of the best ways to obtain vegetation (Tian et al. 2021a;Yin et al. 2022a) and soil moisture (Zhao et al. 2021;Zhao et al. 2020) maps.
The NDVI index has become one of the most popular and commonly used indicators to monitor vegetation due to its universality and simple mathematical formula (Rousta et al. 2020a). According to Huang et al. (Huang et al. 2021b), the number of articles using the NDVI index to monitor changes in vegetation increased from 795 in 1990795 in , to 3361 in 2000795 in , and 12,618 in 2010795 in (Huang et al. 2021a. The NDVI index is widely used in studies related to vegetation classification, and soil erosion risk assessment, because soil erosion decreases with increasing vegetation cover (Zhongming et al. 2010). By correlating NDVI data with the meteorological parameters using the long-term time series for the specific study area it can be checked how climate change affects the growth of vegetation (Li and Kafatos 2000). Also, such studies can be performed to check whether persistent drought conditions occur in a given area and how they affect vegetation.
The period of instability from dry weather conditions, which leads to water scarcity, is simply known as drought (Xu et al. 2022b). Drought is a very complex and not wellunderstood phenomenon. It causes social and environmental problems, and it leads to immeasurable economic losses (Zhao et al. 2022a). Drought is a serious natural hazard, especially in regions with arid and semi-arid climates (Ali et al. 2019). Compared to other natural phenomena, drought affects wider areas over a longer period, thus causing much more damage than other natural disasters such as floods and earthquakes . The study of climate change and the identification of years of drought are valuable for the management of water resources and vegetation, especially in areas with dry spell occurrence (Jawadi et al.) Afghanistan is a mountainous country with spatially and temporally varying ecological conditions. Mountainous areas are prone to the effects of climate change, which intensifies the pressure on natural and human systems (Omerkhil et al. 2020). Climate changes have caused short-term and long-term droughts that have severely affected Afghanistan's economy. According to the International Disaster Management Agency (IDD), droughts accounted for only 5 percent of natural disasters but affected about 30 percent of the population (Ji and Peters 2005). There are two main types of drought in Afghanistan: meteorological drought (usually accompanied by a lack of rainfall) and hydrological drought (usually associated with a lack of surface and groundwater flow, potentially originating in the wider river basin region) Liu et al. 2022;Munyasya et al. 2022;Quan et al. 2021;Zhang et al. 2019c). These issues may also be combined with land and crop management practices, leading to agricultural drought. Currently, Afghanistan is facing significant drought issues that have a direct impact on the livelihoods and the economy of the country.
The vegetation in Afghanistan has been severely affected by human activities, climate change, and drought, which resulted in the naturally occurring vegetation being preserved intact only in a few high mountain areas and abnormally dry deserts (Rousta et al. 2020a). Such a situation additionally contributes to Afghanistan's vulnerability to the effects of climate change (Akhundzadah et al. 2020). In Afghanistan, the combined effects of climate change and four decades of civil war have destroyed vegetation and infrastructure, leading to the underdevelopment of the 1 3 country. The high dependence of the majority of the country's people on small and large-scale agriculture means that due to the country's dry climate and the low adaptation capacity of farmers climate change creates major problems to deal with (Xu et al. 2022a). The arid and semi-arid climate of the eastern basins of Afghanistan implies that this area can be strongly affected by short and long-term fluctuations of meteorological parameters, which as a result will endanger human living conditions (Aich et al. 2017). In Afghanistan, where a large part of the population is engaged in the agriculture sector, assessing the impacts of climate change and drought on vegetation is crucial for the implementation of sustainable agricultural practices. This is especially important for different crops that are grown annually and seasonally, for example, wheat produced in the north, northeast, and eastern regions of the country (Shroder 2014).
In Afghanistan, due to security problems and the lack of stations monitoring weather, not many studies have been performed on the correlation between meteorological parameters and vegetation, and only a few research were done using remote sensing data (Rousta et al. 2020a). Therefore, the investigation of the impact of weather and climate change on vegetation for proper management and ensuring the stability of vegetation, being of particular importance for the eastern basins of Afghanistan, is still required and expected.
The presented study has been conducted to monitor the fluctuations of vegetation conditions and to assess their relationship with meteorological parameters and drought conditions in the period of 2000-2021 in the eastern basins of Afghanistan. The main objectives of this study were a) to determine the trend of vegetation changes between the years  in the eastern basin of Afghanistan, b) to analyze the past trends in drought from the perspective of meteorology, and c) to determine the relationship between vegetation, drought and meteorological parameters for the eastern basins of Afghanistan. The results of this study can be used by governmental agencies, such as the Ministry of Agriculture, to identify dry and wet years, as well as to determine the trend of changes in meteorological parameters and vegetation for the period 2000-2021.

Study area
The study has been performed for the eastern basins of Afghanistan ( Fig. 1), namely the Amu Darya Basin (ADB) and the Kabul River Basin (KRB), with a total area of 163,840 km 2 . The Amu Darya Basin, with an area of 90,692 km 2 , is bordered by Tajikistan from the north, and Pakistan from the southeast. The total annual water flowing through this basin is 82 billion m 3 , of which 61% comes from Tajikistan, 30% from Afghanistan, and the remaining 9% from Uzbekistan and Turkmenistan. Important Amu Darya tributaries include the Wakhan, Kokcha, Kunduz, Andarab, Khenjan, and Punjab rivers in Afghanistan. The population living in this basin was reported to be about 4.5 million people in 2015. According to the division of the Ministry of Energy and Water, it is divided into 7 sub-basins: Upper Five, Lower Five, Kokcheh, Taloqan, Upper Kunduz, Lower Kunduz, and Lower Amu (Maharjan et al. 2021).
The Kabul River Basin, with an area of 72,843 km 2 is located in the eastern part of Afghanistan and is part of the Indus River Basin, which is common between Afghanistan and Pakistan. A part of this basin includes the Kabul and Kunar river basins, with an area of 53,832 km 2 . Kabul River Basin is the second largest basin in Afghanistan after the Amu Darya and is divided into 13 sub-basins: Upper Panjshir, Lower Panjshir, Ghorband, Central Kabul, Maidan, Logar, Laghman, Lower Kabul, Kunar, Parun, North, Khorram, and Gomel. The population living in this catchment area is estimated to be about 12.1 million in 2015 (Maharjan et al. 2021).
Most of the Amu Darya basin is mountainous with snow/ glacier-fed rivers flowing in steep terrain. The rangeland on the mountainside is grazed by farm animals and nomads during the warm summer months (Cesaro et al. 2019). The northern part of KRB contains mountains with high elevations (Najmuddin et al. 2018). The lowlands of KRB are the most suitable land for agriculture. The farms are mainly located in the central and eastern parts of the KRB. Agriculture and livestock are the main source of livelihood in the KRB. Agriculture is largely irrigated and shares around onefifth of the total irrigated land in the country. The river basin also holds around two-thirds of the total forest resources in the count (Najmuddin et al. 2017).
Climate change has already affected agriculture and vegetation in the ADB and KRB. Due to earlier snow melting spring floods increase in size and the risk of the shortage of water occurs more often in summer and early autumn in years of drought (Klemm and Shobair 2010). In 2013, annual precipitation in the KRB was 327 mm downstream, with a usual fringe effect of the Indian monsoon coming from the South Asian Himalayas, and around 418 mm upstream. The mean annual temperature at the central upstream and downstream locations were 13℃ and 23 ℃, respectively (Akhtar et al. 2018). Table 1 and Fig. 2 present the land cover types of ADB and KRB watersheds, together with their area (km 2 ) retrieved from MODerate resolution Imaging Spectroradiometer (MODIS) MCD12Q1 images according to the International Geosphere-Biosphere Programme (IGBP) classification (Loveland et al. 1999). The figure shows that most of the study area is covered by croplands and grasslands (18 and 26% of the study area, respectively).

Data
In the present study, the vegetation coverage variability for the eastern basins of Afghanistan was investigated for the period 2000-2021, and the impact of such factors as land surface temperature (LST), precipitation, soil moisture, and drought on vegetation coverage was assessed using regression methods. The summary of the sources of the remote sensing data used in this study is provided in Table 2, while the flowchart of data processing is presented in Fig. 3. All satelliteborn statistics of the surfaces were obtained in cloud-free conditions from Google Earth Engine (GEE) platform.

Normalized Difference Vegetation Index (NDVI) data
The Normalized Difference Vegetation Index (NDVI) is one of the most important and widely used vegetation indicators and its application in satellite assessment for global vegetation monitoring has been well-proven in the two last decades (Leprieur et al. 2000). It is commonly used as a detector of surrounding greenness areas and in epidemiological studies to investigate the health effects of green space in urban  (1) where Rred and Rnir represent surface reflectance averaged over visible (RED) (λ ~ 0.65 μm) and near-infrared (NIR) (λ ~ 0.85 μm) regions of the spectrum (Mansourmoghaddam et al. 2022d). The range of NDVI values is between -1 and 1, with the vegetation having NDVI between 0.2-1.0, while values lesser than 0.2 indicate areas without vegetation cover, usually barren, or with rock, snow, water, or ice (Mahmood et al. 2022;Rousta et al. 2022b;Rousta et al. 2020a;Rousta et al. 2020b).  In this study, the time series of the NDVI 16-Day L3 Global 250 m from MOD13Q1 MODIS product (Testa et al. 2014) for a period from January 2000 to December 2021 (22 years, 528 images in total) have been downloaded using the Google Earth Engine (GEE) platform. The data was converted to a spatial resolution of 1 km using the bicubic method (Keeratikasikorn and Trisirisatayawong 2008;Yuan et al. 2007).
To obtain the yearly values of the NDVI the data were averaged as: where i is consecutively numbering the timely ordered images from a specific year from 1 to N.
Based on the NDVI, vegetation coverage was calculated as the area with any type of vegetation by summation of the number of pixels with NDVI > 0.2 and multiplying their number by the area of one pixel. The uncertainty estimation of the area of vegetation cover in this research was mostly related to the conditions in the study area. As the study area is mountainous with unpredictable weather, the major contribution to NDVI uncertainty comes from topographic and atmospheric factors (Borgogno-Mondino et al. 2016). Additional errors, especially for the areas with highly heterogeneous vegetation coverage, are associated with two assumptions made. The first one is that only pixels with NDVI > 0.2 are covered by vegetation, while the second is that all the area of a pixel with NDVI > 0.2 is treated as the area covered by vegetation. To significantly reduce such errors, satellite images with a high spatial resolution (250 m x 250 m) were used for NDVI and VC calculation.

Land Surface Temperature (LST) data
In the study, the time series of the LSTDay-8Day-1 km from MOD11A2 MODIS product with a spatial resolution of 1 km and temporal resolution of 8 days data was used. The data from 2000 to 2021 (22 years, 1056 images in total) were downloaded using the Google Earth Engine (GEE) platform and then averaged to yearly values using the: where i is consecutively numbering the timely ordered images from a specific year from 1 to K.

Precipitation data
The precipitation was derived from the Global Precipitation Measurement (GPM) product. It is an international satellite mission to provide next-generation observations of precipitation and snow worldwide every three hours (Huffman et al. 2015). The GPM data were obtained using the Google Earth Engine (GEE) platform and then averaged to yearly values.

Soil moisture data
The purpose of the Global Land Data Assimilation System (GLDAS) was to employ a source of data for the assessment of Fig. 3 Flowchart of the data processing the environmental and food security in developing countries, such as Afghanistan, that do not have access to terrestrial data (McNally et al. 2017). The overall goal of the GLDAS model was to drive multiple offline LSMs and integrate large amounts of observation-based data, to be implemented globally with high resolution. GLDAS offers a product with a spatial resolution of 0.25° and 1° and a temporal resolution of 3 h. The data is available from January 1948 up to the present (Bi et al. 2016). In this study, the Global Land Data Assimilation System (GLDAS) data was used to obtain information on soil moisture from a depth of 0-10 cm. To match the same spatial resolution as for the other data, the bicubic method has been used to re-sample the soil moisture data to a 1 km grid. Such resampling was needed to integrate soil moisture data with other data into the Optical Integrated Drought Index (OIDI) (they had to have the same spatial resolution). The GLDAS dataset was accessed using the GEE. Methods.

Vegetation Condition Index (VCI)
Since 2014, Kenya's National Drought Management Authority (NDMA) uses the vegetation condition index (VCI) as the basis for providing disaster contingency funds to counties in drought conditions (Mansourmoghaddam et al. 2022c). VCI is a normalized pixel-based NDVI to separate long-term ecosystem changes from short-term climate-related NDVI fluctuations and to reflect relative changes in vegetation conditions from very poor to optimal (Liu and Kogan 1996). VCI compares the current time vegetation with the minimum long-term NDVI and shows how close the current time step is to the long-term minimum NDVI, taking into account the difference between the maximum (indicating the best conditions of vegetative growth) and minimum values (indicating the worst conditions of vegetative growth), which reflect somehow the conditions of the local vegetation (Karnieli et al. 2006). The range of VCI is between 0 and 1, with smaller VCI values indicating worse vegetation growth conditions and, at the same time, higher degrees of drought. Based on the literature regarding aridity classification standards, VCI lower than 0.5 indicates drought conditions (Sha et al. 2013). VCI is defined as: where NDVI min and NDVI max are the multiple-year minimal and maximal values of the NDVI for a pixel, respectively, and 'i' denotes the current time step.

Temperature Condition Index (TCI)
The Temperature Condition Index (TCI) is one of the indicators of drought, which assumes that the occurrence of drought phenomenon reduces soil moisture and creates (4) VCI = NDVI i − NDVI min NDVI max − NDVI min thermal stress on the surface of the earth, which results in the monthly LST in the year of drought greater than for the same month in normal years (Du et al. 2013). It is calculated as: where 'min' and 'max' are the multiple-year minimal and maximal values of the LST for a pixel, respectively, and 'i' denotes the current time step.

Precipitation Condition Index (PCI)
Precipitation Condition Index (PCI) was used to evaluate the variation of precipitation and drought conditions from GPM DATA (Wang et al. 2019). Many previous studies used PCI for monitoring drought instead of the Standard Precipitation Index (SPI) and Standardised Precipitation Evapotranspiration Index (SPEI) (Baig et al. 2020;Han et al. 2020;Wang et al. 2019), and indicate that it is a very reliable index. It is defined as: where 'min' and 'max' are the multiple-year minimal and maximal values of the precipitation for a pixel, respectively, and 'i' denotes the current time step.

Soil Moisture Condition Index (SMCI)
Soil moisture data (GLDAS) was used to drive the soil moisture condition index (SMCI) as: where 'min' and 'max' are the multiple-year minimal and maximal values of the soil moisture for a pixel, respectively, and 'i' denotes the current time step.

Optical Integrated Drought Index (OIDI)
The MIDI is a reliable index for monitoring drought, which integrates the precipitation condition index (PCI), soil moisture condition index (SMCI), and temperature condition index (TCI) (Zhang et al. 2019a). These indices are linearly scaled between 0 and 1 using the absolute maximum and minimum values for the same month based on microwavederived precipitation, soil moisture, and land surface temperature (LST), respectively. Microwave Integrated Drought Index (MIDI) integrates the PCI, TCI, and SMCI indices with flexible weights α, β, and γ (Wei et al. 2021 where α + β + γ = 1. In this study the data from the optical range was used as an input to the MIDI, thus introducing a new index called Optical Integrated Drought Index (OIDI), which is calculated the same way as MIDI.
Based on the literature recommendations, in which the best correlation with the short-term Standard Precipitation Index (SPI) was obtained (Zhang et al. 2019a), weights α = 0.5, β = 0.3, and γ = 0.2 were used. These values of the weights were tested in arid and semi-arid climates and selected as suitable for monitoring drought in such climatic zones (Baig et al. 2020). The range of OIDI values is between 0 and 1, where the value between 0 to 0.1 indicates extreme drought conditions, the value in the range from 0.11 to 0.2 indicates severe drought conditions, from 0.21 to 0.3-moderate drought conditions, from 0.31 to 0.4-low drought conditions and from 0.41 to 1.0 indicates that area under consideration is not experiencing drought.

Z-score calculation
Z-score, also known as the standardized anomaly, informs how large the deviations of the quantity under consideration are. The Z-score is calculated using the formula (Zhao et al. 2019): where i represents the assessed period and j stands for the time scale, Xij is an analyzed parameter in a given year, U represents the mean value for the analyzed statistical period, whereas σij indicates the standard deviation. Positive values of the standardized anomaly indicate that the values under consideration are larger than the mean, the negative values of the standardized anomaly indicate that the values are smaller than the mean, and the values >|2| indicate that the result is abnormal (Wu and Onipchenko 2007).

Correlations calculation
The correlation coefficients were calculated using Pearson correlation. It measures the strength of the linear relationship between two (dependent and independent) variables (Zhou et al. 2016). It has a value between -1 to 1, with -1 meaning a total negative linear correlation, 0 meaning that two quantities are not correlated, and + 1 meaning a total positive correlation. The Pearson correlation is the first formal measure of correlation and it is still one of the most widely used indicators of relationships between variables (Lee Rodgers and Nicewander 1988): (8) MIDI = PCI + SMCI + γTCI, where X denotes the mean of x, ȳ denotes the mean of y. Spatial correlation coefficients were calculated between the average annual VCI (dependent variable) and the annual accumulative PCI, yearly average TCI, yearly average SMCI, and yearly average OIDI (independent variables) using the Pearson correlation method (Zhou et al. 2016). The significance of the correlation coefficients was judged at the confidence of 95% level. Figure 4 shows the average intra-year vegetation coverage of the eastern basins in Afghanistan throughout the study period. The VC had a low value in late winter (~ 11% of the study area was covered in vegetation, 18,063 km 2 ). A rapid monotonic increase up to a maximum in late spring and early summer can be observed (46% of the study area, 75,635 km 2 ), followed by a slow monotonic decrease from summer up to early winter (13% of the study area, 20,679 km 2 ). In wintertime, VC decreases very slowly. From the above results, we can conclude that the peak of VC in the eastern basins of Afghanistan is observed in May and June. Figure 5 presents the relationship between annual vegetation coverage and the annual mean of the area affected by drought conditions for the eastern basins of Afghanistan during the studied period . The annual mean of the area affected by drought conditions was calculated using the percentage range value of the VCI index. If the value was between 0 and 50% it indicated that the area (pixel) under consideration had bad vegetation growth conditions and was affected by drought conditions (DAV), whereas values from 50.1 to 100% indicated good vegetation growth conditions, and that the area was not affected by drought conditions (NDAV). It is worth mentioning here that DAV can take larger values than VC, because for the calculation of the area the values of NDVI < 0.2 (barren land, rocks, buildup areas) are also taken into account.

Analysis of VC variations
study area, with 133,221, 134,110, and 134,016 km 2 , respectively). The relationship between VC and DAV assessed with the use of the linear regression model was significant at the 95% confidence level (R = 0.78, p-value < 0.05). Figure 6 shows the maps of vegetation coverage in the eastern basins of Afghanistan Afghanistan for the years with the lowest (2000 and 2008) and the highest (2016 and 2020) vegetation coverage. Better vegetated areas were observed in the northern and northeastern areas of ADB and the eastern and southeastern areas of the KRB, whereas in the eastern, southeastern, and southwestern areas of the ADB, and the western and southwestern areas of the KRB vegetation occupied a much smaller area.
The areas with changes in vegetation are shown in red color (compared to the year with the highest vegetation during the studied period, which was 2020). From Fig. 2 it results that they are mostly seen in agricultural lands, which is probably connected with human activities.

Annual variations of OIDI
In Fig. 7 the maps of the spatial variations of OIDI in the eastern basins of Afghanistan are presented separately for each year from the studied period (2000-2021), whereas in Fig. 8 the same information is aggregated into a column plot for better comparison of temporal changes. In 2000, which can be recognized as the year affected by extreme and severe drought to the highest degree among the years analyzed, most of the studied area (32%, 51,968 km 2 ) was affected by severe drought conditions. Severe drought affected most of the southwestern and western areas of the KRB, and the northwestern and western areas of the ADB in the Kunduz sub-basin, while the extreme drought conditions were affecting some parts of the KRB in the Gomal sub-basin only (~ 1% of the study area, 1218 km 2 ). For most of the central and northeastern areas of these two basins (20% of the study area, 34,121 km 2 ), no drought conditions were observed. In 2001, most of the southwest and west areas of KBR, and northwest of ADB were affected by moderate drought (33% of the study area, 54,596 km 2 ). In 2002, areas without drought, mild drought, and moderate drought almost have the same size (27, 29, and 29% of the studied area, 44,778, 48,098, and 47,449 km 2 , respectively), and severe drought had affected very few areas of the southwest KRB (6% of the study area, 9829 km 2 ). In 2003, most of the central areas had no drought conditions (39% of the study area, 63,928 km 2 ), however, some areas in the southwest of KRB and the northeast of the ADB had been affected by moderate drought (26% of the study area, 42,272 km 2 ).
In 2004, most of the northern, northeastern, and central areas of the basins experienced no droughts conditions (35% of the study area, 56,889 km 2 ), in turn, the southwest areas of the KRB had been affected by severe drought (15% of the study area, 24,850 km 2 ). 2009 was one of the years least affected by the effects of extreme, severe, and moderate drought conditions from the studied period, and also had the highest area that hadn't experienced drought (55% of the study area, 90,785 km 2 ). Only some areas in the northeast and northwest of the ADB and the southwest of the KRB had been affected by mild, moderate, and severe drought (32, 7, and 2% of the study area, 52,499, 11,948, and 2474 km 2 , respectively). In 2011, most of the central northeast, and southeast areas of the basin were under mild drought conditions (39% of the study area, 63,939 km 2 ), and the northwest areas of the ADB and the southwest areas of the KRB were under moderate drought (39% of the study area, 64,043 km 2 ). In 2016, most of the northwestern and southwestern areas of the study area and some southern areas of the ADB were affected by moderate drought (35% of the study area, 57,937 km 2 ), and most of the central and southwestern areas of the study area hadn't experienced drought (25% of the study area, 42,253 km 2 ). 2019 and 2020 were the second and third years of the studied period with the highest area that hadn't experienced drought. In 2019, 51% of the study area (83,682 km 2 ) was under no drought conditions, except for some areas in the southwest of the KRB and the east of the ADB, which were affected by mild and moderate drought (12% and 40% of the study area, 20,650 and 65,654 km 2 , respectively). In 2020 only some areas in the southwest of the KRB and the south of the ADB were affected by mild and moderate drought (30% and 17% of the study area, 48,654 and 27,409 km 2 , respectively), while 48% of the study area (79,342 km2) was without 1 3 Fig. 7 The maps of the spatial variations of meteorological drought expressed by the annual Optical Integrated Drought Index in the eastern basins of Afghanistan for each year from the study period  1 3 drought conditions. From the temporal changes of meteorological drought in the eastern Basin of Afghanistan during the study period shown in Fig. 8, it results that the areas affected by extreme, severe, and moderate droughts had a downward trend, whereas the trends for the areas affected by mild drought, and with no drought conditions were upwards.
According to Fig. 7, severe meteorological drought has occurred in the northeastern, northwestern, and southern parts of the ADB, and in the southwestern and southern areas of the KRB, almost in all of the studied years. Conversely, according to Fig. 6, these areas were either without vegetation cover or are covered by vegetation to a small extent in 2000, 2008, 2016, and 2020. Figure 9 shows the annual anomalies of VC, precipitation, soil moisture, LST, and OIDI for the eastern basins of Afghanistan during 2000-2021. 2000, 2001, 2016, 2017, 2018, and 2021 were the years with the highest LST (22.3,18.6,18.8,18,18,and 18.4℃ on average, respectively) during the study period. In turn, 2012, 2019, and 2020 had the lowest LSTs (15.2, 16.2, and 16.6℃, respectively). 2009, 2012, 2013, 2014, and 215 had the highest precipitation (525, 569, 574, 590, and 675 mm, respectively) during the study period, whereas 2000, 2001, 2017, and 2021 had the lowest precipitation (211, 217, 330 and 269 mm, respectively). For 2005For , 2009For , 2012For , 2015, and 2019 the highest soil moisture was recorded (23, 22, 24, and 23.4 m 3 m −3 , respectively) during the study period, and conversely, for 2001,2002,2008, and 2021 the lowest soil moisture was observed (17, 18, 19.4 and 18.6 m 3 m −3 , respectively). 2000, 2001, 2004, and 2008 had the lowest vegetation coverage (91,747, 91,847, 98,750, and 90,576 km 2 , respectively), while 2005, 2010, 2013, 2016, and 2020 were the greenest years with the highest vegetation coverage (113,894, 116,570, 116,718, and 118,840 km 2 , respectively). Meteorological drought conditions calculated with the use of the OIDI indicated that in 2000, 2001, 2010, 2011, and 2021 the most area had been affected by meteorological drought (133,049, 133,554, 145,906, 146,870, and 135,398 km 2 , respectively). In turn, in 2005In turn, in , 2009, 2019, and 2020 the smallest area was affected by drought (91,950, 76,475, 83,511, and 87,939 km 2 , respectively). In 2005, precipitation was close to the normal value (precipitation Z-score was close to 0), LST and OIDI were below the normal value, but the soil moisture and vegetation cover were above the normal value. Almost the same can be observed for 2020, in which the precipitation, OIDI, and LST were below the normal value, but the soil moisture and vegetation coverage were above the normal value. It strongly suggests that soil moisture was one of the key parameters controlling the LST and had the highest impact on the variations in vegetation coverage. The decrease in the annual mean LST for the eastern basins of Afghanistan in the studied period was -0.06℃, while an increase in the annual mean precipitation was 6.9 mm yearly. Annual mean soil moisture also had an increasing trend, whereas the area with meteorological drought had a decreasing trend during the study period. Figure 10 and Table 3 show the relationship between the annual mean of the vegetation cover and assessed meteorological parameters, such as precipitation, soil moisture, LST, and drought-affected area calculated on the base of OIDI (OIDI area), for the studied period. A positive relationship was observed for VC and precipitation, and VC and soil moisture, whereas a negative relationship was seen for VC and LST, and VC and OIDI area. It was found that the relationships between VC and precipitation, VC and soil moisture, and VC and LST were significant (R = 0.63, p = 0.0014; R = 0.73, p = 0.0008; and R = 0.57, p = 0.04, respectively), Fig. 10 The scatter plots of the time series presenting the relationships between vegetation coverage and precipitation, soil moisture, LST, and OIDI for the eastern basins of Afghanistan during 2000-2021 Table 3 The correlation (R) and determination (R 2 ) coefficients and p-value for relationships between annual vegetation coverage and precipitation, soil moisture, LST, and OIDI for the eastern basins of Afghanistan during 2000-2021 calculated using the linear regression method * denotes that the correlation was significant (p-value = 0.05) whereas the relationship between VC and OIDI area was not significant (R = 0.36, p = 0.126) at the 95% confidence level. For the variations of VC, the multiple regression equations taking into account the relationships between VC, precipitation, soil moisture, LST, and OIDI area for the eastern basins of Afghanistan during 2000-2021 were calculated for the yearly values (Table 4). These equations allow for estimating the projected value of VC. The obtained multiple regression and determination coefficients indicate that precipitation, soil moisture, and LST explained about 45% of the yearly VC variations.

Spatial variability of the analyzed variables in the period 2000-2021
In Fig. 11 the spatial variations in the mean precipitation, LST, soil moisture, vegetation cover, and meteorological drought index for the period of 2000-2021 are presented. The highest precipitation occurred in the flat areas of north, northeast, south, and southeast of the KRB area and the north, south, and central areas of the ADB, The smallest amounts of precipitation were observed for the areas with high altitudes. Generally, the yearly precipitation sums in these basins increased during the study period. In the southern, southeastern, and southwestern areas of the KRB and the western, northwestern, and southwestern areas of the ADB, LST was high. Generally, the LST decreased in the study area during the study period. LST had an inverse relationship with the orography (Table 5), as, for example, it could be observed for the part of Wakhan district of Badakhshan province, which has high elevation and low LST. Soil moisture depended mostly on the amount of rainfall in the area.
The correlation coefficients (R) between the mean VCI, and the mean PCI, TCI, SMCI, and OIDI were calculated separately for each pixel to determine the spatial relationships between vegetation conditions and climatic variables. The correlations between VCI and precipitation, temperature, soil moisture, and drought, presented in Fig. 12, were dependent on the orography. In the majority of flat areas, the correlation between VCI and PCI index was positive, with very high, up to 0.84, values, whereas in mountainous areas negative correlations were observed. The correlation between the TCI index and VCI was negative in mountainous areas and positive in some flat areas. The correlation between VCI and SMCI index was positive in most flat and low-altitude areas and negative in some mountainous areas. The correlation between the VCI index and OIDI in most of the areas with vegetation was negative, whereas it was positive for the areas with a lack of vegetation. Generally, the high absolute values were observed in most of the flat (characterized by the same height) areas.

Discussion
While Afghanistan's natural ecosystems have already been destroyed during the country's many years of civil wars, unsustainable management, and over-exploitation, literature reports indicate that Afghanistan will face a wide range of new and increased climate risks (Kimura 2020). The worst adverse effects of climate change on Afghanistan are related to drought, including these leading to desertification and land degradation. Severe drought conditions occurring for prolonged periods can lead to aridification in droughtaffected regions (Ma and Fu 2007). Drought is estimated to be the norm by 2030, not a periodic event (Kimura 2020). Currently, Afghanistan is facing significant drought issues that have a direct impact on livelihoods and the economy. The drought that occurred in 2011 has pushed millions of people into food insecurity and poverty (Ranghieri et al. 2017). Although few studies have determined the impact of drought events, there is still a need to assess its impact on various aspects, like vegetation coverage dynamic, especially for longer periods.
In this study, the relationship between vegetation coverage dynamic and meteorological parameters (precipitation, soil moisture, and LST) and meteorological drought conditions was assessed for the eastern basins of Afghanistan for the period 2000-2021. Despite the climate changes that are occurring in Afghanistan (Rousta et al. 2020a Fig. 11 Spatial distributions of the mean precipitation, temperature, soil moisture, and drought index for the period from 2000 to 2021 and significant correlation between the annual mean of VC and the area affected by drought conditions expressed with the use of VCI (the values of the VCI = < 50%) was found (R = 0.78, p = 0.000014). Obtained results were in line with other research made for the whole Afghanistan territory (Rousta et al. 2020a), in which it was found that the vegetation coverage was increasing in the period 2001-2018. The authors found that the correlation between NDVI and VCI was high, whereas the correlation between NDVI and LST was low. Additionally, it was stated that in 2000 and 2008 the lowest vegetation coverage was observed, while in 2010 and 2016 the highest vegetation coverage was recorded.
In general, the area with meteorological drought conditions (OIDI = < 0.4) had a decreasing trend in the study period. The area decreased from 2000 to 2005, and then it increased from 2005 to 2007. Similarly, from 2010 to 2020 a downward trend was observed for the area with meteorological drought conditions. The areas under extreme, severe, and moderate meteorological drought were decreasing, while the area with mild drought conditions was increasing during the study period. Most areas were affected by moderate and mild droughts. These results were in line with the other research Table 5 The correlation (R) and determination (R 2 ) coefficients and p-value for relationships between elevation and PCI, SMCI, TCI, and OIDI for the eastern basins of Afghanistan calculated using the linear regression method

R R 2 p-value
Elevation -PCI -0.036* 0.001 0.04 Elevation -TCI 0.9* 0.8 0 Elevation -SMCI 0.019* 0.03 0 Elevation -OIDI 0.38* 0.14 0 Elevation -VCI -0.6* 0.3 0 Fig. 12 Spatial distribution of the correlation coefficients between the yearly mean of VCI and yearly means of the TCI, PCI, SMCI, and OIDI made for the Kabul river basin (Baig et al. 2020), in which it was stated that 2000 and 2004 were the years with the worst meteorological drought conditions from the period from 2000 to 2018 and that the trend of meteorological drought changes in KRB was downward. During the studied period most of the northwest, southwest and some eastern areas of the eastern basins in Afghanistan had been influenced by drought. The highest value of the area under meteorological drought was observed in 2000, 2001, 2007, 2010, 2011, and 2021 (81, 82, 85, 90, and 83% of the study area affected by drought, respectively). Observed variations in annual VC were related to the changes in meteorological parameters. For example, in 2000 and 2021 annual VC was below the normal value, simultaneously with annual precipitation, and soil moisture, whereas LST was above the normal value. In 2015 and 2019, annual VC was above normal value, simultaneously with annual precipitation, and soil moisture, whereas annual LST was below the normal value. Obtained results indicated that the correlation between VC and precipitation was positive and significant (R = 0.63, p = 0.0014), and the total annual precipitation had an upward trend during the study period (Yu et al. 2020). The correlation between VC and soil moisture was positive and significant (R = 0.73, p = 0.0008), and the annual mean soil moisture had an upward trend during the study period (Pei et al. 2018). The correlation between VC and LST was also positive and significant (R = 0.57, p = 0.04), and the annual mean LST was decreasing during the study period. The correlation between VC and metrological drought was not significant (R = 0.36, p = 0.126) at a 95% confidence level. Obtained results are somewhat in line with the other research made for Kabul River Basin in Afghanistan , in which the vegetation coverage dynamics and its relation to atmospheric patterns were investigated (Quan et al. 2022). It was found that the vegetation dynamics in KRB were impacted by both precipitation and LST, however, the magnitude of this impact depended on the season. During the winter LST had a greater impact on VC variation than precipitation, and conversely, in summer, precipitation impacted vegetation to a higher degree than LST. In another study, the vegetation dynamics and its relationship with climatological factors for Caspian Sea watersheds in Iran were analyzed (Rousta et al. 2022a). It was found that the correlations of vegetation coverage with ET and LST in winter were positive and significant (R = 0.46 and 0.55, p-value = 0.05, respectively), while the correlation with the precipitation was not significant. In the spring, the correlation between VC and precipitation was positive and significant (R = 0.55, p-value = 0.05), but the impact of LST on the vegetation coverage was negligible when the precipitation was abnormally high. In the summer, the correlation between VC and LST was negative and significant (R = -0.45, p-value = 0.05).

Conclusions
In the present study, the impact of meteorological parameters and meteorological drought on vegetation coverage in the eastern basins of Afghanistan has been investigated using remote sensing data. It was found that soil moisture had a high impact on VC, and the LST impacted VC to the slightest extent from the studied meteorological parameters. The relationship between VC and the area under meteorological drought was insignificant. The correlations between VC and precipitation and soil moisture were positive and significant (R = 0.63, p = 0.0014, R = 0.73, p = 0.0008, respectively), whereas the correlation between VC and LST was negative and significant (R = 0.57, p = 0.04). It was revealed that precipitation, soil moisture, LST, and area under meteorological drought conditions explained about 45% of the yearly VC variation in the eastern basins of Afghanistan.
The results of this research indicated that the changes in the vegetation coverage in the eastern basins of Afghanistan during 2000-2021 had an upward trend. VC increased slightly from 2000 to 2005 and decreased slightly from 2005 to 2008, with 2008 being the year with the least vegetation during the studied period. From 2008 to 2021, VC generally increased, however, a slight downward trend was observed between 2016 and 2018. Annual mean LST had a downward trend, whereas total annual precipitation had an upward trend during the study period. In most parts of Afghanistan, the vegetation depends on the winter rain, however, in the south winter rains are often irregular. Rainfall increases to the north and east resulting in better vegetation conditions in these parts. The eastern parts additionally receive some monsoon rains in summer (Breckle 2007). Annual mean soil moisture had an upward trend, and the areas under extreme, severe, and moderate meteorological drought conditions were declining during the studied period. In turn, the areas with mild meteorological drought conditions had an upward trend in the study period.
It was also revealed that the orography had a significant influence on both the spatial distribution of the LST and VCI, as well as the spatial correlations between VCI and meteorological parameters.
Funding This research was supported by Vedurfelagid, Rannis and Rannsoknastofa i vedurfraedi.
Data availability Not applicable.

Institutional review board statement Not applicable.
Informed consent statement Not applicable.

Competing interests
The authors declare no competing interests.

Conflicts of interest
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.