Evaluation of high-resolution precipitation datasets CHIRPS, TerraClimate and TAMSAT over the Enkangala Escarpment of South Africa

doi:10.21203/rs.3.rs-4365508/v1

This study evaluates the performance three high resolution rainfall products (CHIRPS, TerraClimate and TAMSAT) with reference to ground rain observations network of 25 weather stations data over Enkangala Escarpment of South Africa, for the period of 40 years. We used continuous, categorical, and volumetric indices, and at various elevations, and temporal scales (monthly, seasonal, and annual). CHIRPS have shown the best statistical scores at monthly, seasonal (DJF, SON, and MAM) and annual scales owing to it high r values, lower RMSE, higher IA and relatively low bias for the magnitude. The correlation analysis of elevation shows CHIRPS resolve problem of orographic rainfall better than TerraClimate and TAMSAT. Overall, the underestimation of rainfall by CHIRPS at monthly scale is approximately 1.6 mm, seasonal (6.5–15) mm and annual 29.47 mm while TerraClimate overestimate at annual scale (17.1) mm with higher RSME. Based on the categorical metrics it shows both data set can detect rainfall estimate at various scale but varies with increase in elevation. TAMSAT provide poor estimations at monthly and annual scale but render it not suitable for hydrological studies over Enkangala Escarpment. We recommend CHIRPS as alternative to station dataset.

Rainfall is an important component in the hydrological cycle and play important role in the modelling of rainfall, its variability, climate change and water resources management. Rainfall contributes to environmental and economic prosperity of human race(Mohsin et al., 2021; Hendrix and Salehyan, 2012). To keep track and record rainfall temporally and spatially has become an enormous task for institutions. In situ rainfall record are not usually available in most developing countries because of the high cost of maintaining hydrometeorological equipment associated centres. Missing record or complete lack records are impediment for robust climate modelling in low rainfall gauge region(Emanuel, 2021; Flato et al., 2014).

There is need to have complete dataset for estimation of rainfall since rainfall distribution has been altered by climate change (Flato et al., 2014; Mendez et al., 2020). Climate change is expected to be impacted more in low earning countries and rainfall dependent agriculture countries(IPCC, 2007). These countries are mostly in the Sub Sahara countries due to lack of capacity for mitigation and adaptation. Therefore, climate change and variability studies are incomplete with weak and missingness datasets. The advent of satellite data has provided opportunity for obtaining continuous climate datasets in the last four decade. The constant improvement of the sensors and computational abilities of the various remote sensing products are available at different temporal and spatial scale (Sishodia et al., 2020; Alvarez-Vanhard et al., 2021). The resolutions of such products have been the main issue on the accuracy and reliability of precipitation dataset across the globe.

High resolution precipitation datasets are scares which are tailored to specific region of the globe. The common dataset such Merged Analysis of Precipitation, Tropical Rainfall Measuring Mission, TRMM-3B43, CMAP; Climate Prediction Center morphing technique, CMORPH; Global Precipitation Climatology Centre, GPCC; and CPC are with resolution more than 10km on the horizontal. It became imperative to for researchers to evaluate the performance of the dataset against station datasets. In this paper three continuous satellite datasets, the Climate Hazards Group Infra-Red Precipitation with Station data (CHIRPS) at 0.05 km resolution TerraClimate datasets (TC) at 0.04 km resolutions and TAMSAT (TS) 0.0375 resolution u are used provide as proxies for the gauge-based precipitation measurements in Enkangala Escarpment. Using high-resolution datasets distributed spatially could enhance the evaluation of water security in different sectors like farming, ecosystem functions, flooding, and drought. (Cepeda Arias and Cañon Barriga, 2022; McNally et al., 2017).

There are series of validation of the CHIRPS dataset at global and regional scale. The short term validation on global scale (Verdin et al., 2020; Shen et al., 2020), over Africa continent (Mekonnen et al., 2023), East Africa (Dinku et al., 2018; Gebrechorkos et al., 2018; Ageet et al., 2022), West and Central Africa (Kouakou et al., 2023), and South Africa (Du Plessis and Kibii, 2021), Egypt (Nashwan et al., 2020; Nashwan et al., 2019), and Ghana (Atiah et al., 2020). Other basins like Nile basin (Abdelmoneim et al., 2020), Barada basin in Syria (Alsilibe et al., 2023) and Imperial River basin in Chile (Baez-Villanueva et al., 2018).

CHIRPS dataset is usually validated on global and regional level in countries such as Colombia, Afghanistan, Ethiopia, Sahel, and Mexico. They specifically examined satellite gridded data from CHIRPS, CHIRP, CFS, CPC, ECMWF, and GPCC in their study (Funk et al. 2015). There are some validations of CHIRPS in South Africa, CHIRPS was proven as alternative to observed rainfall data in the Pitman runoff modelling in Bereg Catchment area (Kibii and Du Plessis, 2023). CHIRPS was used for inter precipitation product comparison for South Africa (Cattani et al., 2021). Monthly CHIRPS product was use for rainfall estimate (Du Plessis and Kibii, 2021)

On local scale CHIRPS, TerraClimate and TAMSAT datasets, have not been evaluated in Enkangala Escarpment. Over the years CHIRPS dataset has not been thoroughly validated during the dry period in tropics and early rainfall season. The TerraClimate dataset are incapable of capturing temporal variability and orographic inversion and rainfall at a finer scale than its core dataset (Abatzoglou et al., 2018).

This paper focuses on the Enkangala Escarpment of South Africa which part of the Greater Drakensburg network of complex mountainous climate change vulnerable area. South African Weather Service (SAWS) rainfall stations data were analysed to test for the accuracy of CHIRPS, TerraClimate and TAMSAT datasets. This is for the period between 1984 to 2023 at monthly, seasonal, and yearly scales.

2.1. Study area.

The Enkangala escarpment, as defined for this project a 200x200 km2 area defined by a northwest corner at 26.965 S 28.991 E and a southeast corner at 28.791 S 30.908 E. This square includes the quaternary headwater of the Tugela, Pongola, Usutu, and Vaal rivers. It excludes the high Drakensberg along the KwaZulu Natal- Lesotho border, which provides water to the Tugela system, but below the abstraction points for the inter-basin transfer to the Vaal River system, which provides water to the power generation and industrial heat of South Africa. The range of the elevation in the study area is 345 m to 2900 m. The mean annual rainfall of Enkangala Escarpment is based on the dataset from SAWS 883 mm. Moreover, about 49% of the rainfall is from December to February, and 29% is between September to November.

Table 1

List of the synoptic stations and their characteristics
Station	Lat	Lon	Alt	Duration % of missing
Athole	-28.57	30.48	1557	853.4	3.3
Bergville	-28.73	29.35	1145	781.6	1.1
Blaauwkop	-28.5	30.27	1676	736.8	0.9
Cavern farm	-28.64	28.96	1490	985.8	1.9
Chelmsford dam	-27.95	29.95	1219	814.9	3.5
Dannhauser	-28.01	30.06	1350	875.4	1.4
Dundee	-28.16	30.23	1247	590.3	4.2
Glencoe	-28.18	30.15	1311	517.7	1.1
Harrismith	-28.26	29.14	1650	816.2	3.9
Hattingspruit	-28.08	30.12	1310	635.4	3.9
Memel	-27.68	29.57	1728	789.3	0.8
Moorside	-28.4	29.61	1219	1016.	0.9
Ncome	-27.93	30.65	1295	909.9	0.9
Nerston	-26.57	30.79	1525	753.4	0.9
Rocco	-27.73	29.17	1844	672.1	1.14
Roseleigh	-28.61	29.55	1145	936.3	1.96
Royal Park	-28.69	28.95	1392	740.0	1.3
Swartwater	-28.36	30.36	1372	665.2	3.59
Tafelkoppies	-26.88	30.62	1372	524.8	3.43
Tygerfontein	-27.59	29.45	1852	732.4	0.98
Verkykerskop	-28.92	29.28	1830	770.6	0.98
Warden	-28.85	28.96	1631	976.5	0.98
Waterval	-27.79	30.25	1190	625.4	0.98
Watervaldrift	-26.89	30.61	1380	657.4	7.07
Zaaiplaats	-27.3	30.94	1052	826.8	5.39

2.2. Observational data from gauge stations

We analysed historical rainfall records between 1984 and 2023 from twenty-eight meteorological station found with the Enkangala Escarpment. The South Africa Weather Service (SAWS) operates and maintain distributed network of meteorological stations across the country. The temporal scale varies from location to location based their history. In this paper, SAWS stations are used for evaluation of the high-resolution products over the Enkangala Escarpment. Henceforth, the SAWS dataset would be referred as Stations dataset from the period from 1984 to 2024. This is adopted to accommodate the temporal scales of the CHIRPS, TerraClimate and TAMSAT datasets. A total of 25 stations are carefully selected for this research based on the robustness and completeness of the dataset from the stations as shown in Fig. 1. The SAWS stations datasets are subjected to internal quality checks which are publicly available upon request. We used only stations with less than 7% missing values for our study period The gauge rainfall data were subjected to quality assurance test such as homogeneity, outlier test and infilling were necessary. Outliers’ tests are important because it can affect the overall mean which may lead to wrong decision. The homogeneity test helps us to understand the spread and uniformity of the dataset’s analysis included outlier detection, infilling of missing data. Outliers were detected using the Grubbs and boxplot diagrams. We used an open-source statistical software R (version 4.1.2) for most of the analysis.

2.3. Satellite data

2.3.1 CHIRPS

When creating satellite and reanalysis rainfall products, some products utilize gauge data to adjust for biases in rainfall estimates during the validation process. CHIRPS v2 (referred to as CHIRPS hereafter) is a dataset that covers nearly the entire globe and offers estimates on a daily, five-day, and monthly basis (Funk et al., 2015). CHIRPS utilizes data from National Agencies like Global Surface Summary of the Day (GSOD), Global Historical Climatology Network (GHCN), and other in situ rain gauge data. CHIRPS utilizes a worldwide monthly precipitation climatology (CHPclim), thermal infrared (TIR) cold cloud duration (CCD), and daily and monthly RG data. The TIR CCD data is locally calibrated with TMPA 3B42 pentadal precipitation at a constant CCD temperature threshold of 235 K. The pentadal estimates are then used to calculate Climate Hazard Infrared Precipitation (CHIRP) by multiplying them with their corresponding CHPclim estimate. CHIRP data is combined with gauge data to generate CHIRPS data at pentadal and monthly intervals that is used to produce high resolution CHIRPS dataset at 0.05 resolution on the horizontal (Funk et al., 2015). CHIRPS V.2 datasets satellite-based monthly rainfall is also available online at http://chg.geog.ucsb.edu/data/chirps/).

2.3.2 TerraClimate

TerraClimate is a detailed dataset for rainfall, minimum and temperature at high resolution (4 km) on the horinzonal which also provide water balance information monthly for terrestrial surfaces worldwide from 1958–2024 on monthly basis. ion provides extremely valuable contributions for Ecological and hydrological research on a global level necessitates high spatial resolution and constantly changing data. The data is updated monthly. This rainfall product is product climatically assisted interpolation by merging detailed climatological normals from WorldClim database with spatial dynamic data information from the Gridded Time Series (CRU Ts4.0) data collected by the Climatic Research Unit and Japanese reanalysis data (JRA55) (Harris et al., 2014; Fick and Hijmans, 2017; Cepeda Arias and Cañon Barriga, 2022). TerraClimate can be accessed for free on GEE and from Climatology Lab (https://www.climatologylab.org/terraclimate.html) for 1985 to 2024

2.3.3 TAMSAT

Several research studies have highlighted that rainfall data generated using the Tropical Applications of Meteorology Using Satellite Data and Ground-Based Observations (TAMSAT) group at the University of Reading in the United Kingdom is more suitable for research in Africa compared to other alternatives(Tarnavsky et al., 2014; Nashwan et al., 2020; Dube et al., 2023; Balti et al., 2020). This is due to the accessibility of precipitation estimates, the accuracy of TAMSAT data, their spatial resolution, and their frequent use for trend analysis. The TAMSAT precipitation products use thermal infrared (TIR) observations from Meteosat and data from precipitation stations to create gridded precipitation data. In addition, TAMSAT precipitation data provide a useful basis for studying long-term precipitation patterns spanning from February 1983 to the present, covering the entire African continent at a spatial resolution of 4 km. The TAMSAT information was created by combining Meteosat data with ground data using an exploratory calibration method (Tarnavsky et al., 2014). Before 2009, TAMSAT data were limited to North, South and certain parts of East Africa, but updated data covering the period 1983 to the present, including calibrated measurements for the entire African continent, are now available(Seyama et al., 2019; Dinku et al., 2018). A methodical statistical evaluation shows that the TAMSAT daily datasets can be used with great precision for a variety of applications as they are like other precipitation datasets obtained through remote sensing. While the reliability of TAMSAT data has been demonstrated in numerous studies, there are cases where rainfall is either over- or underestimated. Possibly because TAMSAT precipitation estimates use infrared data instead of SAR data. Additionally, Palmer et al. (2023) suggests that TAMSAT data is suitable for assessing risks and examining precipitation trends over time. Henceforth, this research will incorporate the monthly TAMSAT precipitation estimates from 1984 to 2023 as one of three high resolution datasets for evaluation Enkangala Escarpment of South Africa

2.4. Methodology

2.4.1 Observations and datasets comparison

Point to pixel comparison was adopted to evaluate the obtained value from SAWS operated stations within Enkangala Escarpment with the three high resolution rainfall products (CHIRPS, TerraClimate and TAMSAT). A total of 26 stations fairly distributed across the Escarpment were carefully and systematically for this task. The study area, complex landforms, and varying precipitation make it difficult to accurately estimate rainfall in gridded data. Because daily records are not widely available for most rain gauge data. We decided to utilize monthly rainfall of the selected stations from 1984 to 2023. This allows us to have common temporal scale for both the stations data and grided high resolution datasets. The chosen timeframe depends on finding a consecutive 7-year period with no breaks in recent years. Based on WMO (2018), a duration of five years is deemed suitable for carrying out this type of research.

2.4.2 Evaluation of Accuracy of the datasets

Monthly, seasonal (MAM, JJA, SON and DJF) and annual accuracy of the high-resolution datasets were quantitatively. We used continuous statistical measurements and categorical metrics for the evaluation (Abdelmoneim et al., 2020; Liu et al., 2020).The continuous statistics measures are the root mean square error (RMSE), bias, Kling–Gupta Efficiency (KGE) and index of Agreement (IA). These statistical measures used in this study have been extensively applied to evaluate relative accuracy of various rainfall estimation products in other studies(Abatzoglou et al., 2018; Atiah et al., 2020; Palmer et al., 2023; Dinku et al., 2018; Du Plessis and Kibii, 2021; Ogbu et al., 2020; Cattani et al., 2021). The monthly, seasonal annual difference was evaluated across the stations with datasets (CHIRPS, TerraClimate and TAMSAT). The tendency of the over/underestimate are determined by assessing bias, while the root mean square error (RMSE) was employed to measure the average error magnitude between the two values. The KGE, is a strong, impartial measurement that characterizes and assesses time-series' general fitness by combining correlation, bias, and variability (Gupta et al., 2009). It varies from negative infinity to positive infinity with an ideal value of 1.

We used categorical indices to determine the capacity of the datasets (CHIRPS, TerraClimate and TAMSAT) to detect estimates above a given threshold (Degefu et al., 2022; Cepeda Arias and Cañon Barriga, 2022). These indices are e Probability of Detection (POD), which refers to the proportion of precipitation events that were correctly detected (i.e. 1 means all precipitation events were detected by the grid data, 0 means none the events were detected). was recognized correctly). The false alarm ratio (FAR) represents events that were identified by the grid data but not confirmed by the rain gauge observation (i.e. 0 perfect agreement of the rainfall events in both the observation and the grid data, 1 means that all detected events were not confirmed by the measuring devices are confirmed). The Critical Success Index (CSI) is the fraction of the rain event that was correctly detected by the grid data, including missed rain events (i.e. 1 means no rain event occurred in both the false alarm and false alarm categories).

Consequently, the correctly/incorrectly recorded precipitation volume fraction is still unknown. This margin for bias (overestimation or underestimation) can be important in a context where rainfall varies greatly in frequency, duration, and intensity. Therefore, the categorical metrics were extended to volumetric measures to quantify the error size between the observed and estimated variables. Volumetric indices include the volumetric hit index (VHI), which defines the correct amount of precipitation captured by the satellite product relative to the amount of precipitation captured by the satellite and missed in the observation.

3.1. Monthly evaluation

The calculated monthly rainfall for the SAWS stations and their corresponding basic statistical values for presented in Table 2. TerraClimate dataset has the highest mean value (66.93 mm), followed by Stations dataset (with a mean of 66.54 mm), followed by TAMSAT dataset (with a mean of 62.7 mm) and lastly the CHIRPS (with a mean of 60.7 mm. Regarding the deviation of the monthly rainfall, TerraClimate and Stations datasets have similar value 69.9 mm and 68.9 mm respectively. TAMSAT has 59.1 mm while the lowest is CHIRPS with standard deviation (STD) value of 53.9 mm (Table 2). The scatter plots of the CHIRPS, TerraClimate and TAMSAT datasets against stations showed a high agreement CHIRPS, TerraClimate and TAMSAT r (0.96, 0.92, 0.84) respectively (Fig. 2).

Table 2

Descriptive statistics of observed and satellite datasets for the period 1984 to 2023 over Enkangala Escarpment of South Africa.
Dataset	Mean	Min	Max	Median	STD	CV	Skewness
TerraClimate	66.93	0	339.9	50.7	61.9	0.91	1.05
Stations	66.88	0	396.2	47.3	68.9	1.04	1.33
TAMSAT	61.85	0	251.2	48.4	58.6	0.95	0.68
CHIRPS	62.9	3.4	304.5	46.3	55.6	0.84	0.94

The level of agreement between observed data and the three datasets is assessed in their calculated Pearson correlation coefficient (r). The r values for both dataset is high greater than 0.7, with CHIRPS dataset has the highest r value 0.79, TerraClimate 0.76 and TAMSAT 0.7 of their r values respectively (Table 3). The results of continuous and categorical metrics used for the monthly assessment is presented in Fig. 3. There is an indication in the continuous statistical measures for the datasets that, CHIRPS (CH) presented in the boxplots in Fig. 3 has outperformed, TerraClimate (TC) and TAMSAT (TS) based on their calculated RMSE. The median of CH RMSE is approximately 40 mm/ month across the stations while TC and TS have their RMSE greater than 50 mm/mon. The continuous metric of KGB which measure overall fitness of the time series, it is drive by integrating r values correlation, bias, and variability integration of r values. It shows TC dataset have overall high goodness of fit shows but skewed to right as result of outliers. The median value of TC is 0.73 against CH 0.68 and TS 0.63 respectively.

The three datasets have high values for IA above 80% across the stations over Enkangala Escarpment. In terms of the bias CH has a median value of 0.06 mm while TC has 1.58 and TS has − 4.17. This show considerably good performance of CH dataset against the others. CHIRPS overestimation is less than 1% while TerraClimate overestimated about 15% and TAMSAT proved to be underestimating the stations rainfall. The last continuous metric SS used in the evaluation of the datasets shows a very good performance.

The POD of the monthly data show that the CH have perfect score of 1 across all the stations which is followed by TC which has median value of 0.99 while TS has POD of 0.86. The FAR for TS is closer to zero than CH and TC likewise their values for VFAR. The CSI and VCSI for TC performed better than CH which is also better than TS (Fig. 3). The VHI of the datasets shows TS is better than CH and TC which are overestimating rainfall between 1.2 mm to 1.3 mm/mon. Overall, the volume of rainfall missed is not more than 3% of threshold allocated for the datasets during the calculation of the VHI.

3.2 Seasonal Evaluation

The seasonal evaluation of the three high resolution rainfall products and the data rich SAWS network of stations as our observation is shown in Table 3 and Fig. 4. During December, January, and February season (DJF) all the three datasets products overestimate rainfall over Enkangala Escarpment. The values their BIAS suggests a range between 112 t0 146 mm, CH datasets 112 mm, TAMSAT 113 mm and TerraClimate 146 respectively during the three-month period. The magnitude of error captured by the RMSE show that CHIRPS performed better with RMSE values 177.16 mm against 210.77 mm and 191.19 mm produced by TerraClimateS and TAMSAT per season. The KGE which captured r residuals are all positive values CHIRPS (0.32), TerraClimate (0.23) while TAMSAT have (0.21) during DJF. The satellite datasets are s are moderately in agreement with the observational dataset based on the range of the IA values. In MAM season, TAMSAT has negative KGE =-0.17 against the station’s dataset while CHIRPS and TerraClimate both have a moderate positive KGE values of 0.37 and 0.34 respectively (Table 3). During MAM, CH and TS dataset underestimated the MAM season with CH (4.4 mm) and TAMSAT 41.97 mm. The RMSE scores help to show the difference in the intensity CHIRPS and TAMSAT above 60 mm per season while TerraClimate dataset yielded 73.35 mm per season (Fig. 4).

All the datasets shown capacity to accurately detect the rainfall by scoring high POD and almost perfect score for FAR. The categorical indices aligned with the volumetric indices.

Table 3: Result for seasonal and annual evaluation metric over Enkangala Escarpment.

Seasons	Datasets	RMSE	KGE	IA	BIAS	POD	FAR	VHI	CSI
MAM	CHIRPS	66.98	0.36	0.66	-4.39	1	0.001	1.0012	0.999
	TerraClimate	73.25	0.34	0.62	6.37	1	0.001	1.0012	0.999
	TAMSAT	109.7	-0.17	0.422	-41.9	0.96	0.001	0.9596	0.958
JJA	CHIRPS	66.78	0.37	0.66	-3.59	1	0.001	1.0012	0.999
	TerraClimate	73.16	0.34	0.62	6.89	1	0.001	1.0012	0.999
	TAMSAT	109.8	-0.17	0.42	-41.6	0.96	0.001	0.958	0.956
SON	CHIRPS	76.77	0.51	0.75	-11.3	1	0.001	1.0012	0.999
	TerraClimate	78.57	0.58	0.77	-1.85	0.95	0.001	1.0012	0.9992
	TAMSAT	94.67	0.33	0.62	16.4	1	0.001	1.0012	0.999
DJF	CHIRPS	177.2	0.32	0.58	112.7	1	0	1	1
	TerraClimate	210.8	0.23	0.55	146.0	0.91	0	1	1
	TAMSAT	191.1	0.11	0.48	113.3	1	0	1	1
Annual	CHIRPS	205.9	0.4	0.64	-29.47	1	0	1	0
	TerraClimate	210.4	0.45	0.66	17.61	1	0	1	0
	TAMSAT	229.3	0.15	0.52	-42.7	1	0	1	0

During the JJA season, there are two positive KGE for CH and TC while TS have negative KGE value of -0.17. There is underestimation rainfall by the CHIRPS 3.6 mm, TS 41.6 mm while TC overestimate by 7.59 mm per season. The CHIRPS dataset has a better RMSE score of 66.8 mms against 73.2 mm and 109.8 mm of TerraClimate and TAMSAT datasets. PDF values of TS range between 34–58%. Finally, in the September, October, positive correlation with r values and November (SON) season. The datasets all positive KGE which range from 0.33 to 0.58. CH and TS are underestimating rainfall distribution within Enkangala Escarpment between 11.3 mm and 1.85 mm per season. The CH yield a better RMSE value than TS and TAMSAT with difference approximately 2 mm and 19.9 mm of rainfall during the SON 9 (Table 3). TS has the highest IA value in the seasonal climatology. Both datasets show their capacities in detecting season rainfall based on their POD, FAR and CSI indices. However, there is degree of independence between individual accuracy assessment versus their seasonal periods. This is based on the variation of the calculated matric used in this study.

3.3. Annual evaluation

The scatter plots of the rainfall datasets and the stations data are presents in Fig. 5. The R² values shows that CHIRPS (0.79) is performed better than TerraClimate (0.66) and TAMSAT (0.15) in capturing annual rainfall in Enkangala Escarpment. Annual rainfall variability of the satellite rainfall product is presented in Fig. 5a-d at annual scale for the period of this study. TAMSAT dataset underestimate in most of the years except in 2015 which it overestimates rainfall. The 2015 is a known El Nino year. This buttresses the lack of model fits by the product as shown with its residual values represented by the R^2 = _0.15 (Fig. 5c). CHIRPS and TerraClimate both overestimate and underestimate some instances but CHIRPS fits closely with stations rainfall then TerracClimate. The region of our study is prone climate variability and induced effect of the El Nino-Southern Oscillation-ENSO are captured by the annual timeseries plots in figure (5-d).

The boxplots for the annual evaluation of the datasets are presented in Fig. 6. The RMSE of the datasets suggests that CHIRPS have the annual mean RMSE value of 205 (Table 4). TerraClimate is close to the CHIRPS but with three outliers while TAMSAT have RMSE 229.3. Another important continuous metric KGE for the annual datasets show TerraClimate and CHIRPS dataset closely share KGE values (0.45 and 0.4), but TAMSAT came only account by 15% of its residual in capturing the stations rainfall based on the KGE. The IA values for the dataset show TerraClimate have highest level of agreement by 66% while CHIRPS have 64% and TAMSAT 52% (Table 4). It very clear that CHIRPS and TAMSAT an underestimate the annual rainfall by (29.47 and 42.7) mm respectively, while TerraClimate data overestimate the annual rainfall by 17.1mm per year. is very close to TerraClimate dataset with just two outliers and better median value which all.

ow for the spread of error figure (6: A1). As shown in Table 4 and Fig. 6, both datasets are capable of estimating the stations annual rainfall based on their perfect scores for POD, FAR, CSI and VHI.

3.4. Impact of Elevation on precipitation

There is any particular pattern for the positioning of stations based on their elevation in Enkangala Escarpment. The stations were divided into based on the distance of their elevations. Those with elevation between 0-1399 m and that 1400 to 2000 m. Table (1) presents the results of the continuous and categorical statistic of the two classes of stations based on their elevation. Table (4) also provide the correlation coefficient of monthly dataset. of were correlated against the continuous statistical metrics used in evaluation of the datasets. CHIRPS datasets show improvement with its RMSE which reduced sharply from 45.1 mm/month with elevation (< 1400 m) to 41.5 mm with elevation (> 1400 m). The dataset changed from under estimation (6.3) mm to overestimation of 1.1 mm across the stations in Enkangala Escarpment. TerraClimate performed poorly with overestimation of 211.6 mm/ month with elevation below 1400 m. KGE and IA performed much better in area with elevation > 1400 m. CHIRPS and TAMSAT are more sensitive with elevation than TerraClimate positively increased CHIRPS and TerraClimate dataset POD improved with elevation above 1400 m. The seasonal metric was also used for the purpose of understanding the effects of elevation on CHIRPS, TerraClimate and TAMSAT as illustrated in figure ().

Table 4: Performance assessment for the three datasets at monthly and seasonal (MAM, JJA, SON, DJF) scale with respect to elevation ranges below1400 m and above 1400 m

Monthly	Datasets	RMSE	BIAS	IA	KGE	POD	FAR	CSI
below1400 m	CHIRPS	45.1	-6.3	0.9	0.6	0.6	0.6	0.8
	TerraClimate	211.6	153.6	0.7	-1.7	-1.7	-1.7	0.8
	TAMSAT	53.1	-3.5	0.8	0.6	0.6	0.6	0.8
Above1400m	CHIRPS	41.5	1.1	0.9	0.7	0.7	0.7	0.8
	TerraClimate	47.8	5.4	0.8	0.7	0.7	0.7	0.8
	TAMSAT	49.7	-3.6	0.8	0.6	0.6	0.6	0.8
MAM	CHIRPS	68.6	2.2	0.7	0.4	0.4	0.4	1.0
below1400 m	TerraClimate	72.5	4.5	0.6	0.4	0.4	0.4	1.0
	TAMSAT	121.8	-47.2	0.4	-0.3	-0.3	-0.3	0.9
	CHIRPS	66.8	-12.9	0.6	0.3	0.4	0.3	1.0
Above1400m	TerraClimate	71.9	10.2	0.7	0.4	0.4	0.4	1.0
	TAMSAT	95.0	-35.4	0.4	0.0	0.0	0.0	1.0
JJA	CHIRPS	27.6	0.8	0.7	0.3	0.3	0.3	0.9
below1400 m	TerraClimate	28.9	5.3	0.7	0.4	0.4	0.4	0.9
	TAMSAT	31.0	3.9	0.7	0.4	0.4	0.4	0.9
	CHIRPS	26.9	-1.5	0.7	0.3	0.3	0.3	0.9
Above1400m	TerraClimate	25.2	1.7	0.8	0.5	0.5	0.5	0.9
	TAMSAT	34.0	5.4	0.7	0.3	0.3	0.3	0.9
SON	CHIRPS	68.2	-1.5	0.8	0.6	0.6	0.6	1.0
below1400 m	TerraClimate	73.1	10.5	0.8	0.6	0.6	0.6	1.0
	TAMSAT	96.8	32.0	0.6	0.3	0.3	0.3	1.0
	CHIRPS	82.8	-16.2	0.7	0.5	0.5	0.5	1.0
Above1400m	TerraClimate	81.8	-10.6	0.8	0.6	0.6	0.6	1.0
	TAMSAT	90.4	2.4	0.6	0.4	0.4	0.4	1.0
DJF	CHIRPS	179.2	112.2	0.6	0.3	0.3	0.3	1
Below 1400 m	TerraClimate	209.5	137.5	0.5	0.2	0.2	0.2	1
	TAMSAT	181.7	92.8	0.5	0.2	0.2	0.2	1
	CHIRPS	174.6	108.9	0.6	0.3	0.3	0.3	1
Above1400m	TerraClimate	215.8	148.2	0.6	0.2	0.2	0.2	1
	TAMSAT	208.4	137.8	0.4	0.0	0.0	0.0	1

The CHIRPS have the lowest RMSE values across the season and improved by having lowered RMSE with increase in elevation (Table 4). This is followed by TerraClimate across the seasons then TAMSAT. The CHIRPS dataset underestimates the rainfall in SON by 1.5 mm area of elevation < 1400 while it overestimates rainfall at the range between (0.8, 2.2, 112) mm during MAM, JJA and DJF seasons respectively. Perhaps, JJA season at elevation < 1400 has the less value of the bias across the seasons. The seasonal accuracy statistic for stations with elevation above 1400 m, CHIRPS underestimate throughout the season except in DJF which there over estimation by 108.9 mm rainfall. The overestimation by CHIRPS is the lowest during the DJF season in contrast to TerraClimate and TAMSAT which have values such as 148.2 mm and 137.8 mm respectively.

The KGE increase with increase in elevation as shown Table 4 for CHIRPS and TerraClimate on monthly scale while for TAMSAT it varies. TerraClimate and CHIRPS have the highest KGE values (0.6) each during SON season for the elevation > 1400 m. The rest of the KGE values irrespective of the elevation are less 0.5. The IA across the season varies, TerraClimate and CHIRPS have the IA (0.4) values while the lowest is in DJF TAMSAT for stations with Elevation < 1400 m while TerraClimate and CHIRPS have the highest IA (0.8) for stations with elevation > 1400m in JJA seasons. While The CHIRPS IA value of 0.8 in the SON is the highest for stations with > 1400m within Enkangala Escarpment. In the other hand, categorical metric POD is worst for TAMSAT in area with elevation > 1400 m in DJF while JJA CHIRPS is the best POD 0.6. Terraclimate have the best POD of 0.6 during SON season for the period 1984–2023 in Enkangala Escarpment. Some seasons recorded perfect scores for their CSI in the SON and DJF for both classes of elevations. The season with lowest CSI value was overestimate at both low and high elevation. They both overestimate in different proportions in all temporal scales. Figures 7 and 8 shows the correlation between elevation and accuracy indicators at seasonal monthly and annual aggregations from the 25 stations. There is positive correlation between the CHIRPS accuracy has an indirect relationship with elevation across the seasons. Only during the MAM season that CHIRPS have lower correlation coefficient values.

This relationship is demonstrated clearly in DJF for the RMSE and bias values for the datasets. The RMSE correlation coefficient value TerraClimate is lower when compared with CHIRPS and TAMSAT. Hence, it is not supprising that CHIRPS and TAMSAT are affected by elevation. The KGE and IA have inverse relationship with elevation as shown in the seasons.

On annual basis, the correlation coefficient of RMSE values is positive for CHIRPS and TerraClimate while TAMSAT has a negative relationship with elevation (Table 5). What it means is that errors are spread on annual dataset for CHIRPS and TerraClimate. The bias shows incredible that CHIRPS and TAMSAT are positively affected by elevation, hence their underestimation is accounted (Fig. 8). The IA, show TAMSAT values are directly affected by the elevation more than TerraClimate follow by CHIRPS. Therefore, KGE coefficient value is affected more in TerraClimate, TAMSAT and CHIRPS in this order (Fig. 7).

Table 5

Results of station-wise elevation correlation with annual rainfall of the datasets CHIRPS (CH), TerraClimate (TC), TAMSAT (TS).
Station	Latitude	Longitude	Elevation	CH	TC	TS
Athole	-28.57	30.48	1557	0.79	-0.04	0.71
Bergville	-28.73	29.35	1412	0.83	-0.21	0.74
Blaauwkop	-28.5	30.27	1145	0.71	0.74	0.72
Carven	-28.64	28.96	1490	0.75	0.73	0.65
Chelmsford	-27.95	29.95	1219	0.83	0.77	0.72
Dannhauser	-28.01	30.06	1350	0.8	0.74	0.69
Dundee	-28.16	30.23	1247	0.79	0.71	0.66
Glencoe	-28.18	30.15	1311	0.86	0.8	0.75
Harrismith	-28.26	29.14	1650	0.8	0.74	0.69
Hattingspruit	-28.08	30.12	1310	0.79	0.72	0.67
Memel	-27.68	29.57	1728	0.74	0.77	0.66
Moorside	-28.4	29.61	1219	0.84	0.82	0.76
Ncome	-27.93	30.65	1295	0.78	0.78	0.69
Nerston	-26.57	30.79	1525	0.82	0.76	0.72
Rocco	-27.73	29.17	1844	0.83	0.74	0.72
Roseleigh	-28.61	29.55	1145	0.8	0.71	0.75
Royal_national	-28.69	28.95	1392	0.85	0.79	0.72
Swartwater	-28.36	30.36	1372	0.74	0.75	0.68
Tafelkoppies	-26.88	30.62	1372	0.78	0.73	0.69
Tygerfontein	-27.59	29.45	1852	0.76	0.76	0.73
Verkykerskop	-28.92	29.28	1830	0.76	0.67	0.7
Volksrust	-27.36	29.89	1652	0.82	0.81	0.69
Warden_ sk	-28.85	28.96	1380	0.77	0.78	0.7
Waterval	-27.79	30.25	1631	0.84	0.77	0.68
Zaaiplaats	-27.3	30.94	1052	0.89	0.68	0.66

In this study we explore the reliability and the quality of three satellite datasets with reference to 25 rain gauge stations operated by the SAWS distributed across Enkangala Escarpment of South Africa. It is indicative that combination of continuous statistics and categorical statistics can provide better insight on the performance of estimated rainfall datasets. It is imperative to understand that fine resolution like the CHIRPS, TerraClimate and TAMSAT dataset are not enough in capturing the exact rainfall in an area because the products care highly affected by the rainfall variability and topography in place like Enkangala Escarpment. Each product comes with its waeakness which my not necessarily provide unique attribute to the estimation being under or over estimation of rainfall. That is introductions of new advanced sensor such as IR and limits the performance in complex terrain (Funk et al., 2015).

Based on the result of the descriptive statistics of the high-resolution satellite rainfall products it is evident that, there are capable of capturing high variability of monthly rainfall over Enkangala Escarpment presented in this study. The datasets CHIRPS, TerraClimate and TAMSAT monthly estimates are right skewed which is a good characteristic of satellite products. The mean and standard deviation of the rainfall distribution give insight on the variability and basic for the comparison of the datasets mean. The high variability of the dataset is analysed by the determination of the CV. The high CV is found throughout the 25 stations irrespective of the area. This is contrary to previous work on entire southern Africa that TAMSAT dataset is negatively skewed and rainfall variability is confined to area with low rainfall (Seyama et al., 2019). The scatter plots of the three-satellite rainfall dataset demonstrate the usefulness of all the dataset to represent the areal distribution of monthly rainfall at a regional scale for Enkangala. Performance of CHIRPS is better than the TerraClimate and TAMSAT based on the level of agreement (Figure). CHIRPS dataset has better ability to characterise the rainfall pattern over Enkangala Escarpment, for both monthly rainfall and seasonal variability even though TerraClimate and TAMSAT were found to be to estimate the continent rainfall estimates. The capability the datasets were demonstrated in their temporal scales which is used in this study.

The higher value of r, IA, and lower values of RMSE and bias of CHIRPS data demonstrated the outperformance of the product against TerraClimate and TAMSAT at the monthly scale. The TerraClimate KGE value is higher than CHIRPS and TAMSAT, but this could not be the yardstick against the CHIRPS and TAMSAT, because the of thigh errors of the residuals as shown in the results with outliers at the monthly scale. This is reflected by the overestimation of the TerraClimate. ast continuous metric SS used in the evaluation of the datasets shows a very good performance. From the research once could the the CHIRPS perfect score for POD across the stations against the TerraClimate the TAMSAT (Fig. 3). It is important to note that TAMSAT lacks a reference station within Enkangala Escarpment or within the radius of 3km. The underestimation of the TAMSAT called for review of its algorithm due to systemic error evident in the output. This corroborates with previous work of TAMSAT poor estimate ()

On seasonal scale, CHIRPS have the highest correlation coefficient for DJF, MAM while TerraClimate performed better in JJA and SON and TAMSAT weak positive correlation coefficient which is less r < 0.45 in all the season excerpt in MAM where it has a negative correlation. Therefore, TAMSAT is not a candidate product for consideration in terms of seasonal analysis in Enkangala Escarpment. Despite the improvement in technology TAMSAT has shown no improvement in its estimation. The result is similar to Thorne anlsysis of TAMSAT over area of complex topography. The summer period is best captured by CHIRPS or TerraClimate datasets. Since Enkangala is a mid-latitude area which is associated with low correlation, but rainfall is associated with convection. Some have argued that instrumentation can also play a role in in estimating the amount of rainfall by the products. The use of IR technology which uses cloud temperature it positively affects convectional rainfall during summer months. The frontal system of mid-latitude contributes to the austral winter rainfall in southern Africa at various magnitudes. This produces convective rainfall by warm cloud tops which impede the ability of IR sensors to signals. Therefore, the underestimation of TAMSAT and CHIRPS is attributable to CCD and convection. The relative difference between the station dataset and TAMSAT in DJF could be associated with influence of ITZ which control the summer season in Southern Africa. Since this period contribute more than 40% of the annual total of rainfall in the Enkangala Escarpment (Ref). This has been previously stated in finding of ((WMO)) that, local climate and season affect the outcomes of studies comparing satellite precipitation products with observed datasets (Dinku et al., 2007; Jury et al., 2002; Gosset et al., 2013; Maidment et al., 2012). The KGE values captured the weakness of the TAMSAT in both sesons since it as integral derivative of correlation and bias of the datasets The negative correlation of TAMSAT in MAM is reflected also in m the KGE for the season.

CHIRPS datasets have best RMSE results for all the seasons (Table). TAMSAT have lower RMSE than TerraClimate in SON and DJF. The shallow clouds formation during the dry season (JJA) could be reason for high disparity of the bias which is as result of inability of both satellites to detect these signals. The CHIRPS and TAMSAT underestimate the rainfall while TerraClimate overestimate in JJA season. The bias of CHIRP data is less 3 mm per season while TAMSAT is 41 mm while the TerraClimate is above 6 mm. This result is contrary what (Ali et al., 2005; Layberry et al., 2006) suggested that TAMSAT overestimation due to the shallow clouds during JJA. TAMSAT estimates perform relatively better in arid and flat environment but grossly underestimate area with complex terrain such as the mountainous place like Enkangala as attest in previous studies which gauge data was compared with d TAMSAT data, Climate Prediction Centre (CPC) (Thorne et al., 2001). The seasonal categorical metric is similar with the monthly metric for any the datasets.

The annual datasets of these products show underestimation by CHIRPS and TAMSAT while TerraClimate overestimate. However, when investigated further with ENSO, EL Nino and La Nina, there is better evidence which dataset capture the occurrence. The years of higher rainfall are the EL Nina years while lower rainfall is El Nina years. It is evident that, these events affect the overall performance of these products. The years of sever and extreme drought attributed to El Nino such as 1992 and 2015 are years, which for instance CHIRPS dataset have its worse performance across the station dataset by overestimation of the rainfall. In the other hand, TerraClimate underestimate the 1992 and 2015. While during the La Nina years of 1999, 2003 and 2011, TerraClimate overestimated this phase. CHIRPS also underestimated those years slightly. TAMSAT overestimate rainfall in these years 1992 and 2015 which are underscore its reliability for regional assessment over Enkangala Escarpment.

Considering the elevation is an impact factor for distribution of rainfall in southern Africa, the outcome of this research shows that CHIRPS is most responsive to elevation. The CHIRPS and TAMSAT underestimate annual rainfall while TerraClimate overestimate the annual rainfall. The performance of the datasets shows CHIRPS better accuracy for stations both at low elevation < 1400 m and those above 1400 m in Enkangala Escarpment is one of the important outcomes this research. The underperformance of TAMSAT annual rainfall on both stations located at low and high elevations on the mountainous terrain of Enkangala Escarpment in the Drakensberg of South Africa, further expose the inability of the dataset usability in region. Likewise, TerraClimate dataset is close to CHIRPS, but its accuracy is off at the annual scale. This is not surprise since was established previously that TerraClimate is incapable of capturing variability in area with influence of orography and inversion of rainfall(Saxe, 2021). furthermore, upcoming research could investigate approaches for adapting and refining CHIRPS by considering the changes in elevation, to enhance its utility across various water security applications in Enkangala Escarpment

The aim of this study was to understand which of the high-resolution rainfall products (CHIRPS, TerraClimate and TAMSAT) is most suitable over Enkangala Escarpment. In this paper we have provided statistical evidence of the numerical discrepancy that exists between point data (Stations) and the rainfall satellite products over Enkangala Escarpment of South Africa. Perhaps, CHIRPS dataset has shown incredible performance in terms of it KGE, IA, RMSE and lower bias which contribute of underestimation. The presence of a systematic bias in TerraClimate with lower RMSE in MAM season was noticeable while TAMSAT rainfall performed poorly in austral summer months and seasons which corroborated with some earlier research (). The poor correlation of TAMSAT with the stations which is embedded in the KGE statistic shows it is less suitable dataset to represent estimate of rainfall over Enkangala Escarpment. Based on the findings of this paper, TerraClimate have relatively usefulness than TAMSAT, but it overestimates in both seasonal and annual dataset. A satellite product should be able to capture frontal system of non-convective rainfall over the study area. CHIRPS dataset is reasonably good presentation of the rainfall over the study area based on the high accuracy and model verification, the resolution is perfect for model. It can be used as alternative in area of lack of historical rainfall data by providing high resolution and various temporal scale which are important for hydrological studies and can feed into modelling of water security, agriculture, and forestry in Enkangala Escarpment.

Ethics Declarations: There was no ethics requirement for this study.

Data availability statement: Data is available upon reasonable request

Funding: This work was supported by Forestry South Africa (SA Forestry), Grant numbers (FSA2020)

ORCID

Abubakar Hadisu Bello: ORCID Number: 0000-0002-1859-5255

(WMO), W. M. O. 2020. Guidelines on Homogenization
ABATZOGLOU, J. T., DOBROWSKI, S. Z., PARKS, S. A. & HEGEWISCH, K. C. 2018. TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015. Scientific data, 5, 1-12.
ABDELMONEIM, H., SOLIMAN, M. R. & MOGHAZY, H. M. 2020. Evaluation of TRMM 3B42V7 and CHIRPS satellite precipitation products as an input for hydrological model over Eastern Nile Basin. Earth Systems and Environment, 4, 685-698.
AGEET, S., FINK, A. H., MARANAN, M., DIEM, J. E., HARTTER, J., SSALI, A. L. & AYABAGABO, P. 2022. Validation of satellite rainfall estimates over equatorial East Africa. Journal of Hydrometeorology, 23, 129-151.
ALI, A., AMANI, A., DIEDHIOU, A. & LEBEL, T. 2005. Rainfall estimation in the Sahel. Part II: Evaluation of rain gauge networks in the CILSS countries and objective intercomparison of rainfall products. Journal of Applied Meteorology and Climatology, 44, 1707-1722.
ALSILIBE, F., BENE, K., BILAL, G., ALGHAFLI, K. & SHI, X. 2023. Accuracy assessment and validation of multi-source CHIRPS precipitation estimates for water resource management in the Barada Basin, Syria. Remote Sensing, 15, 1778.
ALVAREZ-VANHARD, E., CORPETTI, T. & HOUET, T. 2021. UAV & satellite synergies for optical remote sensing applications: A literature review. Science of remote sensing, 3, 100019.
ATIAH, W. A., AMEKUDZI, L. K., ARYEE, J. N. A., PREKO, K. & DANUOR, S. K. 2020. Validation of satellite and merged rainfall data over Ghana, West Africa. Atmosphere, 11, 859.
BAEZ-VILLANUEVA, O. M., ZAMBRANO-BIGIARINI, M., RIBBE, L., NAUDITT, A., GIRALDO-OSORIO, J. D. & THINH, N. X. 2018. Temporal and spatial evaluation of satellite rainfall estimates over different regions in Latin-America. Atmospheric Research, 213, 34-50.
BALTI, H., ABBES, A. B., MELLOULI, N., FARAH, I. R., SANG, Y. & LAMOLLE, M. 2020. A review of drought monitoring with big data: Issues, methods, challenges and research directions. Ecological Informatics, 60, 101136.
CATTANI, E., FERGUGLIA, O., MERINO, A. & LEVIZZANI, V. 2021. Precipitation products’ inter–comparison over East and Southern Africa 1983–2017. Remote Sensing, 13, 4419.
CEPEDA ARIAS, E. & CAÑON BARRIGA, J. 2022. Performance of high-resolution precipitation datasets CHIRPS and TerraClimate in a Colombian high Andean Basin. Geocarto International, 37, 17382-17402.
DEGEFU, M. A., BEWKET, W. & AMHA, Y. 2022. Evaluating performance of 20 global and quasi-global precipitation products in representing drought events in Ethiopia I: Visual and correlation analysis. Weather and Climate Extremes, 35, 100416.
DINKU, T., FUNK, C., PETERSON, P., MAIDMENT, R., TADESSE, T., GADAIN, H. & CECCATO, P. 2018. Validation of the CHIRPS satellite rainfall estimates over eastern Africa. Quarterly Journal of the Royal Meteorological Society, 144, 292-312.
DU PLESSIS, J. & KIBII, J. 2021. Applicability of CHIRPS-based satellite rainfall estimates for South Africa. Journal of the South African Institution of Civil Engineering, 63, 43-54.
DUBE, T., SEATON, D., SHOKO, C. & MBOW, C. 2023. Advancements in earth observation for water resources monitoring and management in Africa: a comprehensive review. Journal of Hydrology, 129738.
EMANUEL, K. 2021. Atlantic tropical cyclones downscaled from climate reanalyses show increasing activity over past 150 years. Nature communications, 12, 7027.
FICK, S. E. & HIJMANS, R. J. 2017. WorldClim 2: new 1‐km spatial resolution climate surfaces for global land areas. International journal of climatology, 37, 4302-4315.
FLATO, G., MAROTZKE, J., ABIODUN, B., BRACONNOT, P., CHOU, S. C., COLLINS, W., COX, P., DRIOUECH, F., EMORI, S. & EYRING, V. 2014. Evaluation of climate models. Climate change 2013: the physical science basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press.
FUNK, C., PETERSON, P., LANDSFELD, M., PEDREROS, D., VERDIN, J., SHUKLA, S., HUSAK, G., ROWLAND, J., HARRISON, L. & HOELL, A. 2015. The climate hazards infrared precipitation with stations—a new environmental record for monitoring extremes. Scientific data, 2, 150066.
GEBRECHORKOS, S. H., HÜLSMANN, S. & BERNHOFER, C. 2018. Evaluation of multiple climate data sources for managing environmental resources in East Africa. Hydrology and Earth System Sciences, 22, 4547-4564.
GUPTA, H. V., KLING, H., YILMAZ, K. K. & MARTINEZ, G. F. 2009. Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. Journal of hydrology, 377, 80-91.
HARRIS, I., JONES, P. D., OSBORN, T. J. & LISTER, D. H. 2014. Updated high‐resolution grids of monthly climatic observations–the CRU TS3. 10 Dataset. International journal of climatology, 34, 623-642.
HENDRIX, C. S. & SALEHYAN, I. 2012. Climate change, rainfall, and social conflict in Africa. Journal of peace research, 49, 35-50.
IPCC, C. C. 2007. The physical science basis. Contribution of working group I to the fourth assessment report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996, 113-119.
KIBII, J. & DU PLESSIS, J. 2023. The application of CHIRPS-based Pitman modelling in South Africa. Physics and Chemistry of the Earth, Parts A/B/C, 132, 103475.
KOUAKOU, C., PATUREL, J.-E., SATGÉ, F., TRAMBLAY, Y., DEFRANCE, D. & ROUCHÉ, N. 2023. Comparison of gridded precipitation estimates for regional hydrological modeling in West and Central Africa. Journal of Hydrology: Regional Studies, 47, 101409.
LAYBERRY, R., KNIVETON, D., TODD, M., KIDD, C. & BELLERBY, T. 2006. Daily precipitation over southern Africa: a new resource for climate studies. Journal of Hydrometeorology, 7, 149-159.
LIU, C.-Y., ARYASTANA, P., LIU, G.-R. & HUANG, W.-R. 2020. Assessment of satellite precipitation product estimates over Bali Island. Atmospheric Research, 244, 105032.
MCNALLY, A., ARSENAULT, K., KUMAR, S., SHUKLA, S., PETERSON, P., WANG, S., FUNK, C., PETERS-LIDARD, C. D. & VERDIN, J. P. 2017. A land data assimilation system for sub-Saharan Africa food and water security applications. Scientific data, 4, 1-19.
MEKONNEN, K., VELPURI, N. M., LEH, M., AKPOTI, K., OWUSU, A., TINONETSANA, P., HAMOUDA, T., GHANSAH, B., PARANAMANA, T. P. & MUNZIMI, Y. 2023. Accuracy of satellite and reanalysis rainfall estimates over Africa: a multi-scale assessment of eight products for continental applications. Journal of Hydrology: Regional Studies, 49, 101514.
MENDEZ, M., MAATHUIS, B., HEIN-GRIGGS, D. & ALVARADO-GAMBOA, L.-F. 2020. Performance evaluation of bias correction methods for climate change monthly precipitation projections over Costa Rica. Water, 12, 482.
MOHSIN, M., ZHU, Q., NASEEM, S., SARFRAZ, M. & IVASCU, L. 2021. Mining industry impact on environmental sustainability, economic growth, social interaction, and public health: an application of semi-quantitative mathematical approach. Processes, 9, 972.
NASHWAN, M. S., SHAHID, S., DEWAN, A., ISMAIL, T. & ALIAS, N. 2020. Performance of five high resolution satellite-based precipitation products in arid region of Egypt: An evaluation. Atmospheric Research, 236, 104809.
NASHWAN, M. S., SHAHID, S. & WANG, X. 2019. Assessment of satellite-based precipitation measurement products over the hot desert climate of Egypt. Remote Sensing, 11, 555.
OGBU, K. N., HOUNGUÈ, N. R., GBODE, I. E. & TISCHBEIN, B. 2020. Performance evaluation of satellite-based rainfall products over Nigeria. Climate, 8, 103.
PALMER, P. I., WAINWRIGHT, C. M., DONG, B., MAIDMENT, R. I., WHEELER, K. G., GEDNEY, N., HICKMAN, J. E., MADANI, N., FOLWELL, S. S. & ABDO, G. 2023. Drivers and impacts of Eastern African rainfall variability. Nature Reviews Earth & Environment, 4, 254-270.
SAXE, S. 2021. Exploring the Effects of Uncertainty in Hydrologic Modeling and Reconciling Spatiotemporal Data Gaps at the Continental Scale. Colorado School of Mines.
SEYAMA, E. S., MASOCHA, M. & DUBE, T. 2019. Evaluation of TAMSAT satellite rainfall estimates for southern Africa: A comparative approach. Physics and Chemistry of the Earth, Parts A/B/C, 112, 141-153.
SHEN, Z., YONG, B., GOURLEY, J. J., QI, W., LU, D., LIU, J., REN, L., HONG, Y. & ZHANG, J. 2020. Recent global performance of the Climate Hazards group Infrared Precipitation (CHIRP) with Stations (CHIRPS). Journal of Hydrology, 591, 125284.
SISHODIA, R. P., RAY, R. L. & SINGH, S. K. 2020. Applications of remote sensing in precision agriculture: A review. Remote sensing, 12, 3136.
TARNAVSKY, E., GRIMES, D., MAIDMENT, R., BLACK, E., ALLAN, R. P., STRINGER, M., CHADWICK, R. & KAYITAKIRE, F. 2014. Extension of the TAMSAT satellite-based rainfall monitoring over Africa and from 1983 to present. Journal of Applied Meteorology and Climatology, 53, 2805-2822.
THORNE, V., COAKELEY, P., GRIMES, D. & DUGDALE, G. 2001. Comparison of TAMSAT and CPC rainfall estimates with raingauges, for southern Africa. International Journal of Remote Sensing, 22, 1951-1974.
VERDIN, A., FUNK, C., PETERSON, P., LANDSFELD, M., TUHOLSKE, C. & GRACE, K. 2020. Development and validation of the CHIRTS-daily quasi-global high-resolution daily temperature data set. Scientific Data, 7, 303.

Evaluation of high-resolution precipitation datasets CHIRPS, TerraClimate and TAMSAT over the Enkangala Escarpment of South Africa

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and methods