1. Introduction
Analyses of water resources in many regions of the world are hampered by a lack of reliable data on precipitation spatial variability and change. Understanding the hydroclimate regime, comprising average precipitation, interannual fluctuations, and changes, is crucial for the planning and management of hydropower, agriculture, and water resources, as well as for grasping the patterns and implications of climate change. Mountainous areas especially have a significant data gap in precipitation information due to the terrain causing rapid spatial gradients in precipitation that might go unnoticed by limited rain gauge networks. The situation is even more pronounced in countries that were once part of the Union of Soviet Socialist Republics (USSR). Following its dissolution in 1989, economic downturns and internal political issues have drastically downsized the hydrometeorological monitoring network, as reported by the National Environmental Agency. This has obstructed hydrological evaluations and the analysis of water resources.
Water resource plans, including in the countries of the former USSR, are often informed by national-level water resources assessments. The Water Balance of Georgia (WBG; Vladimirov et al. 1974) describes and analyzes meteorological data (precipitation and evaporation) and presents precipitation and runoff maps of the country. WBG is still used today to determine the multiannual average flow of ungauged rivers for water resources planning. The data and runoff map are the base of information for “standards of construction” [Стоительные нормы и правила СНиП 2.01.14-83 (Construction norms and rules SNiP 2.01.14-83] and its auxiliary manuals [Л., Гидрометеоиздат (L., Hydrometeoizdat), 1984-448 с.] to determine the hydrological characteristics of ungauged watersheds. The information provided in WBG is based on approaches and data developed in the 1970s, which have not been updated since 1990 and so neglect recent climate change and additional data such as satellite-based precipitation. Outdated methods and data gaps negatively affect the resilience of hydro-development, especially hydropower and water supply systems. The actual electricity generation of many new hydropower plants in Georgia is lower than designed (https://esco.ge/en/electricity/electricity-trade), and many users, including in agriculture, industry, and urban areas, do not have an uninterrupted water supply due to conflicts of interest and high water demand (Organisation for Economic Co-operation and Development 2021). There is an urgent need to evaluate and update national water assessments in Georgia and in other countries of the former USSR.
Improvements in the spatial resolution of precipitation datasets are important for robust water resources assessments and to better understand regional controls on precipitation distribution. Conventional rain gauge networks are the primary sources for accurate point measurements of precipitation (Katsanos et al. 2016) but are often inadequate for regions with sparse gauge networks and complex topography, and frequent gaps in rain gauge data complicate the mapping of spatial patterns and trends (Mu et al. 2021). In the Caucasus region, it is notoriously difficult to capture the spatial patterns of precipitation from ground station data alone given the steep topography (Forte et al. 2016). In recent decades, several satellite-based precipitation products have been developed to estimate precipitation, each with different spatial coverage, data sources, spatial resolutions, and temporal coverage and latencies (Sun et al. 2018). In Nepal, local rain gauge networks combined with ground-based radar provide the best accuracy (Krakauer et al. 2013). Satellite products, such as Tropical Rainfall Measuring Mission (TRMM) and Climate Hazards Group Infrared Precipitation with Station (CHIRPS) data (Funk et al. 2015a), can be very useful where adequate ground data are not collected or available (Krakauer et al. 2013). Validation of CHIRPS has been conducted in several regions, including Nepal (Shrestha et al. 2017), Iran (Ghozat et al. 2021), the Amazon basin (Paca et al. 2020; Mu et al. 2021; Mu and Jones 2022), and the central Andes of Argentina (Rivera et al. 2018). The assessment of CHIRPS data for the Caucasus/Georgia region from 1981 to 2021 has yet to be conducted.
Satellite-derived precipitation can also be used to interpret regional precipitation regimes in complex mountainous regions, including nonunique elevation–precipitation relationships caused by multiple interacting mountain ranges located at different distances from the coast (Bookhagen and Burbank 2006). Regional precipitation maps, including WBG, often rely on assumptions about the relationship between elevation and precipitation, typically assuming orographic effects control precipitation as described in many case studies (Spreen 1947). Factors other than elevation, including distance from the ocean, mountain range orientation, and position relative to other ranges upwind, may dominate in some regions, with examples from sub-Saharan Africa (Hayward and Clarke 1996), the French Alps (Weisse and Bois 2001), and parts of the Himalayas (Bookhagen and Burbank 2006). Satellite-based precipitation maps can be used to assess national precipitation models and to improve understanding of the regional patterns and controls on precipitation distribution, which is important for anticipating the impacts of climate change.
In data-scarce regions, alternative information sources can also be utilized for water resource evaluations and national water model assessments. Georgia has relatively few stream gauges, while hydropower plants often have accurate records of the designed and actual production but have not been commonly compared with regional precipitation maps.
We evaluate precipitation patterns in the country of Georgia using satellite-based precipitation maps, rain gauges, the Water Balance of Georgia, and hydropower production data. Three satellite-based precipitation datasets were used, including CHIRP, CHIRPS, and CHIRPS calibrated with a dense rain gauge network (geoCHIRPS). All CHIRPS products have fine spatial (0.05°) and temporal (daily) resolution, spanning 50°N–50°S from 1981 to the present (Funk et al. 2014, 2015a). We aim to assess the accuracy of the CHIRPS product for monthly precipitation estimates and to develop and evaluate a new dataset by combining CHIRPS with a dense network of rain gauges in the country of Georgia and comparing it to precipitation estimated by the Water Balance of Georgia (Vladimirov et al. 1974) and to hydropower production. We then model precipitation as a function of elevation and distance from the ocean to test hypotheses about the regional controls on precipitation patterns. We aimed to answer the following questions: (i) How accurately do CHIRPS and geoCHIRPS estimate precipitation and its spatiotemporal trends over Georgia? (ii) How do geoCHIRPS and WBG estimates of precipitation compare? (iii) How important are distance from the Black Sea and orographic precipitation for precipitation patterns in both geoCHIRPS and WBG? (iv) How much do hydropower plants in Georgia generate below capacity, and does any underperformance correspond with errors in WBG?
2. Methodology
a. Study area
Georgia (also known as Sakartvelo) is situated between the Caspian and Black Seas and covers an area of 69 700 km2. Geographically, it is divided into two regions, namely, western Georgia and eastern Georgia, separated by the Likhi Range. Georgia encompasses the southern slopes of the Greater Caucasus toward the north, volcanic plateaus and mountain ridges in the south Lesser Caucasus, lowlands toward the west (Kolkheti lowland), and vast plains in the east (Fig. 1). The topography of the Greater Caucasus, Lesser Caucasus, and East Anatolian Plateau profoundly affects the local climate. The southwestern slope of the Greater Caucasus and the northwestern slope of the Lesser Caucasus act as a mountainous barrier that, along with the Siberian high, obstructs winter storms from advancing along the Black Sea coastline (Borisov and Halstead 1965). The annual average precipitation along the Black Sea coast ranges from 1500 to 2500 mm and the amount of precipitation steadily decreases toward the east, owing to the gradual decline of the primary source of precipitation originating in the Black and Mediterranean Seas (Lydolph 1977). In contrast, the climate in eastern Georgia is drier, with precipitation declining from west to east, with an average of 500–1000 mm yr−1 east of the Likhi Ridge. In addition, the driest region in the country is the Kvemo Kartli Plain in the east (Fig. 1), with average annual precipitation ranging from 250 to 500 mm (Kordzakhia and Javakhishvili 1961).
Selected rain gauges and their data availability on a topographical map of Georgia (Sakartvelo).
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
b. Water Balance of Georgia
During the Soviet Union era (1921–89), numerous studies focused on Georgia’s hydroclimate and water resources. Notably, WBG, as documented by Vladimirov et al. (1974), gathered data on precipitation and stream discharge. This collected information was then synthesized and distributed across the entire nation. Due to the complex topography, the country was divided into 49 hydrological regions and precipitation, evapotranspiration, and runoff maps were developed within each hydrological region using elevation–precipitation relationships (Vladimirov et al. 1974). This method was also used by Svanidze et al. (1987) who generalized hydrological data using the same method and assessed the country’s hydropower potential, which was the most recent countrywide hydroclimate assessment. The precipitation data and 49 regions given in WBG have been digitized to compare with CHIRPS (Fig. 2). WBG uses positive linear relationships between precipitation and elevation. As a result, precipitation distribution in WBG reflects the terrain and borders of hydrological regions.
Mean annual precipitation distribution over Georgia (WBG).
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
c. Satellite-based precipitation estimates
The CHIRPS (v2.0; Funk et al. 2015a) product combines data from meteorological stations and several different satellites and climate models. First, a monthly precipitation climatology (CHPclim) was developed based on interpolated station data and satellite imagery from several sources (see Funk et al. 2015a,b for details), which was then temporally disaggregated for each grid cell location into six pentads per month. Anomalies from the CHPclim were then calculated from a combination of satellite imagery and reanalysis climate datasets, including 1) thermal infrared (IR) satellite observations from all available geostationary satellites as compiled by the National Climatic Data Center (NCDC, 3 hourly, 8 km, 1981–2008) and the Climate Prediction Center (CPC, 0.5 hourly, 4 km, 2000–present), 2) the TRMM 3B42 product (0.25°, 3 hourly), and 3) modeled precipitation from the NOAA Climate Forecast System v2 (CFSv2). Pentad precipitation calculated from satellite and model data alone is the CHIRP dataset. Then, precipitation data from meteorological stations, including those compiled in the Global Historical Climate Network (GHCN) and the Global Summary of the Day (GSOD) dataset, are used to calibrate CHIRP, resulting in the CHIRPS product (Funk et al. 2014).
Existing CHIRPS data covering January 1981–December 2021 were downloaded from the CHIRPS website (https://www.chc.ucsb.edu/data/chirps). The period of available CHIRPS data overlaps the rain gauge data available for Georgia (1950–2021). However, the number of precipitation gauges used in the CHIRPS dataset has decreased considerably, both globally and in Georgia, since 1992 (Fig. 3). In 1981, CHIRPS utilized 95 gauges in Georgia, but this number decreased to only 15 in 2021. To assess and enhance the accuracy of the CHIRPS dataset, in this study, we will use 45 rain gauges in total.
Time series of the number of precipitation stations used in the blending process to produce the CHIRPS data for Georgia (Funk et al. 2015a).
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
d. Satellite-based precipitation estimates calibration
The National Environmental Agency (NEA) of Georgia provides daily precipitation data for Georgia. Out of the total of 45 rain gauges used, 27 have continuous records from 1981 to 2021. Of the 45 gauges, 18 were closed as of 2006, but all 45 stations could be used for the calibration and validation of CHIRPS. The density of rain gauges in Georgia (average of 0.65 gauges per 1000 km2) is higher compared to other regions where CHIRPS was implemented, including the state of Rondônia in the Brazilian Amazon (0.53 gauges per 1000 km2; Mu et al. 2021) and for the whole Amazon (0.11 gauges per 1000 km2; Paca et al. 2020). However, the distribution of gauges in Georgia is not spatially uniform, with more stations located in lowlands and populated areas, leaving most of the mountainous areas and ungauged river basins without rain gauges. The NEA conducted quality control on the precipitation data, and we received monthly precipitation data for each gauge.
The 45 rain gauges selected for calibration were used to create the geoCHIRPS precipitation dataset (0.05° × 0.05° resolution, 1981–2021) by blending the calibration gauges network into the CHIRPS product. This was done using the Geospatial Climate Data Management and Analysis (GeoCLIM) software (Bamweyana and Kayondo 2018; Mwesigwa et al. 2017; Funk et al. 2015b), which employs the Background-Assisted Station Interpolation for Improved Climate Surfaces (BASIICS) algorithm. BASIICS blends CHIRPS with additional gauge data using a modified inverse distance weighting (IDW) that borrows some concepts from simple and ordinary kriging (Pedreros and Tamuka 2022). The modified IDW interpolation method was used with specific parameters, including a 1.0 fuzz factor, which hides the location of the gauge by one pixel to avoid reverse engineering the gauge-based pixel value. The BASIICS algorithm carries out a least squares regression between precipitation from rain gauges and satellite values at the rain gauge locations (Pedreros and Tamuka 2022). Point-to-pixel comparisons were carried out to cross-validate the extracted values from the satellite data (CHIRPS; geoCHIRPS) at all valid gauge values (Fig. 4). This method has been widely used in evaluating satellite precipitation estimates (Shrestha et al. 2017; Cavalcante et al. 2020; Mu et al. 2021).
(a) Original CHIRPS data, (b) calibrated geoCHIRPS, and (c) a difference map, which indicates changes of precipitation values due to calibration.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
e. geoCHIRPS performance assessment
The performance of the satellite precipitation products was assessed using several metrics, including the Pearson correlation coefficient r, mean error (ME), normalized root-mean-square error (nRMSE), root-mean-square error (RMSE), ratio of the RMSE and standard deviation of measured data (RSR), percent bias (PB), and Nash–Sutcliffe efficiency (NSE) (Table 1). Trend magnitude (mm yr−1) at the rain gauges was determined using linear regression at each pixel.
Error statistics for CHIRPS and geoCHIRPS, compared with rain gauge data.
f. Hydropower plant performance
Since 2010, more than 50 run-of-river hydropower plants (HPP) were in operation in Georgia. The statistical reports from Georgia’s electricity market operator (ESCO) and Georgian State Electrosystem (GSE) indicate that 80% of HPPs are underperforming more than 10% and about 6 HPPs out of 10 are underperforming more than 20% compared to designed multiannual average electricity generation. We analyzed the monthly production of the Lakhami 1, Lakhami 2, Iphari, and Khelra hydropower plants, located in the same hydrological region (Fig. 5), and compared it with the designed capacity. We calculated precipitation depth over the basin of the selected hydropower plant using both WBG and geoCHIRPS data. It should be highlighted that (i) the HPP does not have a reservoir and is a run-of-river power plant, and hence, its production reflects streamflow, (ii) rivers’ hydrological characteristics and seasonality are dominated by precipitation, (iii) no glaciers are presented in the basin, (iv) hydropower plants were designed using WBG, and (v) they are located in the same hydrological region according to WBG.
Selected rivers and hydropower plants: Lakhami 1 (labeled 1), Lakhami 2 (labeled 2), Iphari (labeled 3) and Khelra (labeled 4).
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
g. Regional patterns and controls on precipitation
To quantify relationships between distance from the coast, topography, and precipitation, we analyze four swath profiles (Fig. 6) using geoCHIRPS and precipitation station data (1981–2021). At each point in the northeast, west–east, and southwest direction along the swath, we averaged data for geoCHIRPS, elevation, and precipitation station within a 5-km-radius window (Fig. 6). Ordinary least squares (OLS) regression was then used to evaluate the relationship between distance from the ocean, elevation, and precipitation.
Selected swath profiles and precipitation stations.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
3. Results and discussion
a. Error in CHIRPS and geoCHIRPS
The satellite precipitation product, geoCHIRPS (R2 = 0.86, r = 0.92), was more accurate than CHIRPS (R2 = 0.74, r = 0.86), in predicting monthly precipitation. geoCHIRPS performed slightly better than CHIRPS when including all months, and both were able to better represent the extremes of precipitation. However, both datasets oversmoothed the precipitation field, which led to an underestimation of high precipitation values. geoCHIRPS had a lower mean nRMSE and was highly correlated with gauge data in almost all months (Fig. 7). CHIRPS tended to overestimate low precipitation amounts and had higher and more variable mean ME and PB compared to geoCHIRPS. Overall, geoCHIRPS performed better in capturing extreme precipitation events during the dry season at the monthly scale (Table 1).
(a) Monthly precipitation from CHIRPS and rain gauge, (b) monthly precipitation from geoCHIRPS and rain gauge, (c) monthly percent bias, and (d) monthly mean error (mm month−1).
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
b. geoCHIRPS compared with Water Balance of Georgia
The time series of WBG (1950–68) and geoCHIRPS (1981–2021) do not overlap in time, so testing for changes in mean annual average precipitation at the rain gauges was important. No significant trends were documented for any station (except for Lagodekhi, slope 0.030 mm yr−1 per year, p value 0.0008; Zugdidi, slope 0.007 mm yr−1 per year, p value 0.03; and Khulo, slope 0.032 mm yr−1 per year, p value 0.03) (Fig. 8), so we could directly compare WBG with geoCHIRPS. In addition, based on the findings of the comparison, WBG overestimates precipitation more than geoCHIRPS, particularly in western Georgia, where the average overestimation is around 30%–35% compared to geoCHIRPS and ground-based stations (Fig. 8). WBG estimates annual precipitation exceeding 4000 mm at high elevations, but this high value is not confirmed by either ground-based stations or satellite estimations.
Comparison between selected rain gauge data, geoCHIRPS, and WBG. Positive values indicate where precipitation in WBG is greater than precipitation from geoCHIRPS.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
In general, WBG overestimates precipitation compared to geoCHIRPS and rain gauges (Fig. 8). However, WBG shows significantly lower precipitation than geoCHIRPS in the downstream reaches of river valleys and in low-lying areas, as well as in arid regions of Georgia (southeast).
c. Regional controls on precipitation patterns
To understand the regional effects on precipitation distribution and precipitation dynamics, we adopted two methods. We mapped precipitation along specific profiles to pinpoint the relationship between topography and precipitation. We then examined four swath profiles, each 10 km in width and about 150 km in length (Fig. 9). For every direction of the swath, we computed the average values using data from geoCHIRPS precipitation, ground station precipitation, and elevation (Fig. 9). When we combine data from all the swaths, several significant features of the precipitation–topography data become apparent: 1) zones of highest precipitation occur along the Black Sea coast and Kolkheti lowlands; 2) precipitation decreases gradually with distance from the coast, with minimal orographic effects; and 3) the mountain ridges surrounding Kolkheti lowland block the frontal band of high precipitation, creating rain shadows, which accelerates the constant decline of precipitation further east from the Black Sea (Fig. 10).
Swath profiles, relationship between topography (brown line), geoCHIRPS precipitation (blue line), and station precipitation (blue triangles). Distance on the x axis is distance from the Black Sea.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
Relationship between (a) precipitation and distance and (b) precipitation and elevation. Blue dots refer to the west part of Georgia, and black dots indicate the east part of the country.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
OLS also identifies two climatic zones in Georgia: the western part and the eastern part separated by the northwest Lesser Caucasus and Likhi Ridge (Fig. 1). In the western part, there is a significant reduction in precipitation as the distance from the coast increases, with the largest reduction observed beyond 50 km from the coast (Fig. 10). This decrease in precipitation is likely due to the reduction in moisture and to the complex topography, which creates shadow effects on the leeward slopes of the mountain ranges. There are also noticeable localized increases in precipitation due to elevation in some parts, particularly in the northwest, where the Greater Caucasus runs parallel to the Black Sea coast and comes close to it (∼33 km), resulting in increased precipitation due to orographic effects. The northwest part of the Greater Caucasus has a relatively low elevation, with an average height of about 2000 m MSL, and unlike the Central Caucasus, it cannot block frontal air masses coming from the Black Sea. In the eastern part of the county, an elevation–precipitation relationship is more noticeable (Fig. 10). During the summer, the increase in precipitation is due to orography and locally generated convection clouds rather than frontal masses coming from the Black Sea (Kordzakhia and Javakhishvili 1961). Hence, the transition period between spring and summer (May and June) is the wettest season in the east, in contrast with the western part of the county where the maximum is in October–January (Fig. 11). Therefore, the influence of the Black Sea is reduced with increasing distance inland, shifting to an orographic-dominated regime in the east.
Monthly precipitation distribution over Georgia by region from geoCHIRPS.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
d. Hydroelectric plant performance
For the hydroelectric plant performance, we evaluated Iphari, Khelra, Lakhami 1, and Lakhami 2 HPPs which are in the same hydrological region and were designed using WBG. For example, for Iphari hydropower plant, the designed yearly generation was 17.4 million kWh; however, according to ESCO, actual yearly generation is about 10.9 million kWh, which is ∼37% less than designed generation (Fig. 12). This is similar to the 38% difference between annual average precipitation in the watershed from geoCHIRPS (1400 mm yr−1) and WBG (precipitation 2270 mm yr−1), and similar results were documented at the other three power plants (Fig. 12). This preliminary comparison suggests that (i) HPPs designed using WBG-estimated precipitation have higher designed production than actual, (ii) the underperformance as a percentage of capacity (∼37%) corresponds with the percentage difference between the differences in precipitation products (∼38%), suggesting that other HPPs in Georgia may face similar problems with underproduction, and (iii) monthly distribution of generation estimated by extrapolating values from neighboring river increases errors, and in this case, rivers fed from glaciers were used to determine monthly flow distribution, causing a higher error in the warmer months (Fig. 12).
Comparison between the designed and actual (multiannual) electricity generation for selected HPPs.
Citation: Journal of Hydrometeorology 25, 4; 10.1175/JHM-D-23-0116.1
4. Summary and conclusions
We developed monthly satellite-based precipitation estimates over Georgia for 1981–2021, compared those estimates with the existing Water Balance of Georgia (WBG) and with hydropower generation, and used the satellite-based estimates to map regional controls on precipitation patterns. Satellite-based precipitation estimates (CHIRPS) underestimated multiannual precipitation depth during cold and wet seasons but overestimated precipitation in the dry season. geoCHIRPS, which is calibrated with a dense rain gauge network, had improved accuracy during all seasons. The commonly used WBG overestimates precipitation depth across the majority of Georgia by about 30%–45%. Interestingly, a hydropower plant displayed a similar underproduction rate of ∼37%. This implies that hydropower plants across Georgia, designed based on WBG data, could potentially be oversized and underperforming. Together with constant flow measurement, dense rain gauge networks incorporating satellite-based precipitation estimates (geoCHIRPS) are necessary to accurately document and evaluate the spatial patterns of precipitation depth and water resources in the country and its ungauged river basins, especially under climate change, which causes uncertainty and threatens Georgia’s sustainable agricultural and hydropower development. Our study provides valuable insights into precipitation patterns in Georgia and the limitations and systematic errors of national-level data. Stations used by the WBG were primarily concentrated in urban centers and might oversimplify the complex and diverse terrain found within a country. The complex terrain and diverse climatic conditions make it challenging to generalize the results beyond Georgia, limiting the broader applicability of the study. For improved evaluation of the country’s hydrological patterns, future research should focus on the estimation of other components of the water balance including evapotranspiration, discharge, snow cover, and land use, and their changes over time.
Acknowledgments.
This study was supported by the U.S. Department of State under the Supporting STEM Research and Graduate Education in Georgia project.
Data availability statement.
Sentinel-2 (ESA) image courtesy of the Copernicus SciHub (https://scihub.copernicus.eu/). The geoCHIRPS are publicly available on Zenodo (https://zenodo.org/records/10889291?token=eyJhbGciOiJIUzUxMiJ9.eyJpZCI6IjY0YjljYWNmLTk3MTItNDI0ZS04YWE5LTI2ZmE3NGRiNzMyYSIsImRhdGEiOnt9LCJyYW5kb20iOiI0MjBmY2UwZDAzYjRjY2YwZDQ4YjMyODFlYjg4YmNjMCJ9.d3KbHgmLTFoOUi_FQLe0h3YrsHfRHCMd0UP4AobOcSsYV2cdMtG7ODsQfnU-SjMOYltzZH2ZGGlpdH8XbAXYyw). CHIRPS data were obtained from https://www.chc.ucsb.edu/data/chirps. Other supporting data can be requested from vazha.trapaidze@tsu.ge at Ivane Javakhishvili Tbilisi State University.
REFERENCES
Bamweyana, I., and L. M. Kayondo, 2018: Spatially explicit modelling of extreme weather and climate events hot spots for cumulative climate change in Uganda. South Afr. J. Geomatics, 7, 90–101, https://doi.org/10.4314/sajg.v7i1.7.
Bookhagen, B., and D. W. Burbank, 2006: Topography, relief, and TRMM‐derived rainfall variations along the Himalaya. Geophys. Res. Lett., 33, L08405, https://doi.org/10.1029/2006GL026037.
Borisov, A. A., and C. A. Halstead, 1965: Climates of the U.S.S.R. Oliver & Boyd, 255 pp.
Cavalcante, R. B. L., D. B. da Silva Ferreira, P. Rógenes Monteiro Pontes, R. Gonçalves Tedeschi, C. P. Wanzeler da Costa, and E. Barreiros de Souza, 2020: Evaluation of extreme rainfall indices from CHIRPS precipitation estimates over the Brazilian Amazonia. Atmos. Res., 238, 104879, https://doi.org/10.1016/j.atmosres.2020.104879.
Forte, A. M., K. X. Whipple, B. Bookhagen, and M. W. Rossi, 2016: Decoupling of modern shortening rates, climate, and topography in the Caucasus. Earth Planet. Sci. Lett., 449, 282–294, https://doi.org/10.1016/j.epsl.2016.06.013.
Funk, C., and Coauthors, 2014: A quasi-global precipitation time series for drought monitoring. USGS Data Series 832, 12 pp., https://doi.org/10.3133/ds832.
Funk, C., and Coauthors, 2015a: The climate hazards infrared precipitation with stations—A new environmental record for monitoring extremes. Sci. Data, 2, 150066, https://doi.org/10.1038/sdata.2015.66.
Funk, C., A. Verdin, J. Michaelsen, P. Peterson, D. Pedreros, and G. Husak, 2015b: A global satellite-assisted precipitation climatology. Earth Syst. Sci. Data, 7, 275–287, https://doi.org/10.5194/essd-7-275-2015.
Ghozat, A., A. Sharafati, and S. A. Hosseini, 2021: Long-term spatiotemporal evaluation of CHIRPS satellite precipitation product over different climatic regions of Iran. Theor. Appl. Climatol., 143, 211–225, https://doi.org/10.1007/s00704-020-03428-5.
Hayward, D., and R. T. Clarke, 1996: Relationship between rainfall, altitude and distance from the sea in the Freetown Peninsula, Sierra Leone. Hydrol. Sci. J., 41, 377–384, https://doi.org/10.1080/02626669609491509.
Katsanos, D., A. Retalis, and S. Michaelides, 2016: Validation of a High-Resolution Precipitation Database (CHIRPS) over Cyprus for a 30-year period. Atmos. Res., 169, 459–464, https://doi.org/10.1016/j.atmosres.2015.05.015.
Krakauer, N. Y., S. M. Pradhanang, T. Lakhankar, and A. K. Jha, 2013: Evaluating satellite products for precipitation estimation in mountain regions: A case study for Nepal. Remote Sens., 5, 4107–4123, https://doi.org/10.3390/rs5084107.
Kordzakhia, M., and A. Javakhishvili, 1961: საქართველოს ჰავა (Climate of Georgia). Science Academy of Georgian SSR, 246 pp.
Lydolph, P. E., 1977: Climates of the Soviet Union. Elsevier Scientific Publishing Company, 443 pp.
Mu, Y., and C. Jones, 2022: An observational analysis of precipitation and deforestation age in the Brazilian Legal Amazon. Atmos. Res., 271, 106122, https://doi.org/10.1016/j.atmosres.2022.106122.
Mu, Y., T. Biggs, and S. S. P. Shen, 2021: Satellite-based precipitation estimates using a dense rain gauge network over the southwestern Brazilian Amazon: Implication for identifying trends in dry season rainfall. Atmos. Res., 261, 105741, https://doi.org/10.1016/j.atmosres.2021.105741.
Mwesigwa, J. B., G. O. Ouma, J. O. Ouma, I. M. Lutta, S. G. Mwangi, and L. A. Ogallo, 2017: Tools and procedures for seasonal downscaling of climate forecasts for use by farmers over the greater horn of Africa: A case study for western Kenya. Climate Change Adaptation in Africa: Fostering Resilience and Capacity to Adapt, Springer, 393–411, https://doi.org/10.1007/978-3-319-49520-0_24.
Organisation for Economic Co-operation and Development, 2021: Developing a Water Policy Outlook for Georgia, the Republic of Moldova and Ukraine: OECD Studies on Water. OECD Publishing, 118 pp., https://doi.org/10.1787/512a52aa-en.
Paca, V. H. D. M., G. E. Espinoza-Dávalos, D. M. Moreira, and G. Comair, 2020: Variability of trends in precipitation across the Amazon River Basin determined from the CHIRPS precipitation product and from station records. Water, 12, 1244, https://doi.org/10.3390/w12051244.
Pedreros, D., and M. Tamuka, 2022: GeoCLIM manual. Family Early Warning System Network, U.S. Geological Survey and UC Santa Barbara, 97 pp., https://edcintl.cr.usgs.gov/downloads/sciweb1/shared/fews/geoclim/QGIS3/GeoCLIM_3.0.19_QGIS_Manual_English.pdf.
Rivera, J. A., G. Marianetti, and S. Hinrichs, 2018: Validation of CHIRPS precipitation dataset along the Central Andes of Argentina. Atmos. Res., 213, 437–449, https://doi.org/10.1016/j.atmosres.2018.06.023.
Shrestha, N. K., F. M. Qamer, D. Pedreros, M. S. R. Murthy, S. Md Wahid, and M. Shrestha, 2017: Evaluating the accuracy of Climate Hazard Group (CHG) satellite rainfall estimates for precipitation-based drought monitoring in Koshi basin, Nepal. J. Hydrol. Reg. Stud., 13, 138–151, https://doi.org/10.1016/j.ejrh.2017.08.004.
Spreen, W. C., 1947: A determination of the effect of topography upon precipitation. Eos, Trans. Amer. Geophys. Union, 28, 285–290, https://doi.org/10.1029/TR028i002p00285.
Svanidze, G. G., V. P. Gagua, and E. V. Sukhishvili, 1987: ВОЗОБНОВЛЯЕМЫЕ ЭНЕРГОРЕСУРСЫ ГРУЗИИ (Renewable Energy Resources of Georgia). USSR State Committee of Hydrometeorology and Control of the Natural Environment, 176 pp.
Sun, Q., C. Miao, Q. Duan, H. Ashouri, S. Sorooshian, and K.-L. Hsu, 2018: A review of global precipitation data sets: Data sources, estimation, and intercomparisons. Rev. Geophys., 56, 79–107, https://doi.org/10.1002/2017RG000574.
Vladimirov, L. A., D. I. Shakarashvili, and T. I. Gabrichidze, 1974: ВОДНЫЙ БАЛАНС ГРУЗИИ (Water Balance of Georgia). Metsniereba, 181 pp.
Weisse, A. K., and P. Bois, 2001: Topographic effects on statistical characteristics of heavy rainfall and mapping in the French Alps. J. Appl. Meteor. Climatol., 40, 720–740, https://doi.org/10.1175/1520-0450(2001)040<0720:TEOSCO>2.0.CO;2.
