Areal Precipitation Coverage Ratio for Enhanced AI Modelling of Monthly Runoff: A New Satellite Data-Driven Scheme for Semi-Arid Mountainous Climate
"> Figure 1
<p>Location of the KRB upstream of the Karkheh Dam in west Iran (<b>a</b>), and its elevation map (<b>b</b>) over sub-catchment areas (<b>c</b>) of Gamasiab (I), Qarasu (II), Seymareh (I–III), and Kashkan (IV). The entire study area of KRB covers all areas I–V.</p> "> Figure 2
<p>Monthly average discharge variability at the Pay-e-Pol station downstream of Karkheh Dam for local water years between 1954 and 2017 (for months when Karkheh and Seymareh dams existed, data were estimated by observations at closest upstream stations).</p> "> Figure 3
<p>Schematic of the dataset grouping and partitioning with the adjusted <span class="html-italic">k</span>-fold process.</p> "> Figure 4
<p>Gamasiab observed vs. modelled monthly runoff time series (<b>a</b>) and q–q plots for calibration (<b>b</b>) and test (<b>c</b>) points (MCM: million cubic meters).</p> "> Figure 5
<p>Qarasu observed vs. modelled monthly runoff time series (<b>a</b>) and q–q plots for calibration (<b>b</b>) and test (<b>c</b>) points (MCM: million cubic meters).</p> "> Figure 6
<p>Seymareh observed vs. modelled monthly runoff time series (<b>a</b>) and q–q plots for calibration (<b>b</b>) and test (<b>c</b>) points (MCM: million cubic meters).</p> "> Figure 7
<p>Kashkan observed vs. modelled monthly runoff time series (<b>a</b>) and q–q plots for calibration (<b>b</b>) and test (<b>c</b>) points (MCM: million cubic meters).</p> "> Figure 8
<p>Karkheh observed vs. modelled monthly runoff time series (<b>a</b>) and q–q plots for calibration (<b>b</b>) and test (<b>c</b>) points (MCM: million cubic meters).</p> "> Figure 9
<p>Monthly P data variability for the KRB sub-catchments using GPM-IMERG-Late data. For each box, the central red line indicates median, and the bottom and top of the box indicate 1st and 3rd quartiles (q1 and q3), respectively. Whiskers denote the most extreme values not considered outliers ‘+’ as data points whose distance to q3 was more than 2.7 × SD × (q3–q1).</p> "> Figure 10
<p>Grid-based mean (<b>a</b>), maximum (<b>b</b>), and coefficient of variation (<b>c</b>) of monthly precipitation from GPM-IMERG-Late over Gamasiab (I), Qarasu (II), Seymareh (I–III), Kashkan (IV), and the entire study area of KRB (I–V).</p> ">
Abstract
:1. Introduction
2. Study Area and Data
2.1. Study Area
2.2. Hydrological Data
2.3. GPM-IMERG Precipitation
2.4. MODIS Products
2.4.1. MODIS-Terra Evapotranspiration Product
2.4.2. MODIS-Terra Vegetation Index Product and Band 7 Data
3. Methodology
3.1. Modelling Architecture and Settings of ANN
3.2. Dataset Partitioning and Adjusted k-Fold Process for Hybrid Modelling
- The runoff data were sorted by magnitude in ascending order.
- The sorted data were split into 20 groups, each encompassing 5% of the data.
- An almost equal portion of data from the groups was then moved randomly to each fold, so that all data within the groups were distributed to all folds at the end. A total of 205 monthly data (from 22 August 2000 to 22 September 2017 equivalent to 01/06/1379–31/06/1396 in Solar Hijri calendar) was categorized into 20 groups of 10 or 11 members. Then, almost one-kth of the groups’ members was allocated to each fold.
- After distributing the data to the folds using above steps, iterative partitioning started with taking the first fold for testing. Then, k – 1 different ways were possible for choice of another fold (2, 3, 4, or 5) for validation, and remaining k − 2 folds were eventually used for training, without a degree of freedom.
- For each iteration of k, having a constant fold chosen for testing, modelling was calibrated k – 1 (for different variation/ways of training-validation combinations) multiplied by 420 (for different combinations of hidden layers and neurons) multiplied by 10 (for different random parameter initialization) times. While the parameterization of each of the (k − 1) × 420 × 10 models was based on a minimum MSE for validation, they were additionally ranked based on the six performance criteria for calibration. For generalization of the output, a hybrid of the six best-ranked ANN models in the form of an arithmetic average was considered as final model for iteration k.
- The previous steps were continued k times so that, eventually, there were k hybrid models with different testing partitions to evaluate their outputs with the observed runoff. Selection of the best hybrid model was based on the comparison of their performances in both calibration and testing partitions of the runoff data.
3.3. Input Data Combinations for Conceptualized Data-Driven Modelling
- –
- For ET (and PET), weighted sum of four or five consecutive 8-day time series data corresponding to their overlapping days within each month;
- –
- For NDVI and B7, weighted average of two consecutive monthly time series data corresponding to their overlapping days within each month;
- –
- For precipitation, sum of 29, 30, or 31 consecutive daily time series data in each month (note: water years in Iran starts with month 7 in Solar Hijri calendar and comprise five 30-day, one 29-day, and six 31-day months, except for leap years, where the sixth month is 30 days).
- –
- First, the catchment-scale monthly precipitation (P) data were classified into 10 categories. For this purpose, they were sorted in an ascending order for each sub-catchment. The first category, with the smallest importance in producing runoff, was allocated to values between 0 and 2 mm/month, roughly encompassing 25% of the data depending on the sub-catchment. Remaining data (P > 2 mm/h) were first divided into eight additional categories with an almost equal data in each (roughly 9.4% of the data). In this way, however, the 9th category contained data with a wide range of values, including extremes. Thus, this category was further divided into two (roughly for percentiles 90–95 and 95–100) to avoid too long range of values in the 10th category.
- –
- Then, a set of areal precipitation coverage values was calculated using the ratio of the catchment area that received precipitation within the range specified by each category. Thus, the number of catchment-specific precipitation coverage ratios (CCOVs) as input variables was equal to the number of precipitation categories with 11 denominators of P as below (rounded values are presented here):
- ■
- Gamasiab (I): 0, 2, 8, 17, 24, 31, 38, 54, 82, 115, and 245;
- ■
- Qarasu (II): 0, 2, 9, 24, 33, 42, 51, 74, 126, 169, and 410;
- ■
- Seymareh (I–III): 0, 2, 9, 21, 31, 39, 52, 68, 108, 159, and 396;
- ■
- Kashkan (IV): 0, 2, 9, 20, 32, 48, 63, 82, 118, 212, and 629;
- ■
- Karkheh (I–V): 0, 2, 9, 23, 31, 43, 56, 73, 110, 157, and 410.
- –
- Next, in addition to CCOVs, another set of data was obtained by calculating the ratio of the catchment area that received higher than 50, 75, 100, 125, 150, and 200% of the P for each month (hereinafter, ECOVs).
- –
- One variable was a simple estimator of runoff using seasonal mean/model (SM) [59]:
- –
- A.
- Precipitation and potential ET (Comb. 1–4);
- B.
- Precipitation, potential ET, and NDVI (Comb. 5–8);
- C.
- Precipitation, potential ET, actual ET, NDVI, and B7 (Comb. 9–12).
3.4. Generalized Output of ANN
4. Results
4.1. Evaluation of Runoff Model for Gamasiab
4.2. Evaluation of Runoff Model for Qarasu
4.3. Evaluation of Runoff Model for Seymareh
4.4. Evaluation of Runoff Model for Kashkan
4.5. Evaluation of Runoff Model for Karkheh
5. Inter-Catchment Comparisons and Discussions
- –
- For Gamasiab (Section 4.1), Comb. 7 with Lag 0–2;
- –
- For Qarasu (Section 4.2), Comb. 12 with Lag 0–1;
- –
- For Seymareh (Section 4.3), Comb. 7 with Lag 0–2;
- –
- For Kashkan (Section 4.4), Comb. 7 with Lag 0–1;
- –
- For Karkheh (Section 4.5), Comb. 12 with Lag 0–2.
6. Conclusions and Remarks on Future Research Directions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Hosseini, S.H. Disastrous Floods after Prolonged Droughts Have Challenged Iran. FUF-Bladet. 2019, pp. 30–32. Available online: https://portal.research.lu.se/en/publications/disastrous-floods-after-prolonged-droughts-have-challenged-iran (accessed on 22 November 2021).
- Boafo, Y.A.; Saito, O.; Jasaw, G.S.; Yiran, G.A.B.; Lam, R.D.; Mohan, G.; Kranjac-Berisavljevic, G. Perceived Community Resilience to Floods and Droughts Induced by Climate Change in Semi-arid Ghana. In Sustainability Challenges in Sub-Saharan Africa I: Continental Perspectives and Insights from Western and Central Africa; Gasparatos, A., Ahmed, A., Naidoo, M., Karanja, A., Fukushi, K., Saito, O., Takeuchi, K., Eds.; Springer: Singapore, 2020; pp. 191–219. [Google Scholar] [CrossRef]
- Prama, M.; Omran, A.; Schroder, D.; Abouelmagd, A. Vulnerability assessment of flash floods in Wadi Dahab Basin, Egypt. Environ. Earth Sci. 2020, 79, 17. [Google Scholar] [CrossRef] [Green Version]
- Ward, F.A.; Amer, S.A.; Ziaee, F. Water allocation rules in Afghanistan for improved food security. Food Secur. 2013, 5, 35–53. [Google Scholar] [CrossRef]
- Gohari, A.; Eslamian, S.; Mirchi, A.; Abedi-Koupaei, J.; Bavani, A.M.; Madani, K. Water transfer as a solution to water shortage: A fix that can backfire. J. Hydrol. 2013, 491, 23–39. [Google Scholar] [CrossRef]
- Deng, C.; Liu, P.; Liu, Y.; Wu, Z.; Wang, D. Integrated hydrologic and reservoir routing model for real-time water level forecasts. J. Hydrol. Eng. 2015, 20, 05014032. [Google Scholar] [CrossRef]
- Hosseini, S.H.; Ghorbani, M.A.; Bavani, A.M. Impacts of Climate Change on Streamflow and Reservoir Operation. In Proceedings of The Second Regional Conference on Climate Change and Global Warming, Center for Research in Climate Change and Global Warming (CRCC), Institute for Advanced Studies in Basic Sciences, Zanjan, Iran, 12 November 2014. [Google Scholar]
- Saeidabadi, R.; Najafi, M.S.; Roshan, G.; Fitchett, J.M.; Abkharabat, S. Modelling spatial, altitudinal and temporal variability of annual precipitation in mountainous regions: The case of the Middle Zagros, Iran. Asia-Pac. J. Atmos. Sci. 2016, 52, 437–449. [Google Scholar] [CrossRef]
- Hiebl, J.; Frei, C. Daily precipitation grids for Austria since 1961—development and evaluation of a spatial dataset for hydroclimatic monitoring and modelling. Theor. Appl. Climatol. 2018, 132, 327–345. [Google Scholar] [CrossRef]
- Gilewski, P.; Nawalany, M. Inter-comparison of rain-gauge, radar, and satellite (IMERG GPM) precipitation estimates performance for rainfall-runoff modeling in a mountainous catchment in Poland. Water 2018, 10, 1665. [Google Scholar] [CrossRef] [Green Version]
- Gilewski, P. Impact of the Grid Resolution and Deterministic Interpolation of Precipitation on Rainfall-Runoff Modeling in a Sparsely Gauged Mountainous Catchment. Water 2021, 13, 230. [Google Scholar] [CrossRef]
- Young, C.-C.; Liu, W.-C. Prediction and modelling of rainfall–runoff during typhoon events using a physically-based and artificial neural network hybrid model. Hydrol. Sci. J. 2015, 60, 2102–2116. [Google Scholar] [CrossRef]
- Carcano, E.C.; Bartolini, P.; Muselli, M.; Piroddi, L. Jordan recurrent neural network versus IHACRES in modelling daily streamflows. J. Hydrol. 2008, 362, 291–307. [Google Scholar] [CrossRef]
- Smith, J.; Eli, R.N. Neural-network models of rainfall-runoff process. J. Water Resour. Plan. Manag. 1995, 121, 499–508. [Google Scholar] [CrossRef]
- Dawson, C.; Wilby, R. Hydrological modelling using artificial neural networks. Prog. Phys. Geogr. 2001, 25, 80–108. [Google Scholar] [CrossRef]
- Yilmaz, A.; Imteaz, M.; Jenkins, G. Catchment flow estimation using Artificial Neural Networks in the mountainous Euphrates Basin. J. Hydrol. 2011, 410, 134–140. [Google Scholar] [CrossRef]
- Ghorbani, M.A.; Hosseini, S.H.; Fazelifard, M.H.; Abbasi, H. Sediment load estimation by MLR, ANN, NF and sediment rating curve (SRC) in Rio Chama river. J. Civ. Eng. Urban. 2013, 3, 136–141. [Google Scholar]
- Kang, B.; Ku, Y.H.; Do Kim, Y. A case study for ANN-based rainfall–runoff model considering antecedent soil moisture conditions in Imha Dam watershed, Korea. Environ. Earth Sci. 2015, 74, 1261–1272. [Google Scholar] [CrossRef]
- Yang, T.; Asanjan, A.A.; Welles, E.; Gao, X.; Sorooshian, S.; Liu, X. Developing reservoir monthly inflow forecasts using artificial intelligence and climate phenomenon information. Water Resour. Res. 2017, 53, 2786–2812. [Google Scholar] [CrossRef]
- Pradhan, P.; Tingsanchali, T.; Shrestha, S. Evaluation of Soil and Water Assessment Tool and Artificial Neural Network models for hydrologic simulation in different climatic regions of Asia. Sci. Total Environ. 2020, 701, 134308. [Google Scholar] [CrossRef]
- Wei, C.-C. Comparison of River Basin Water Level Forecasting Methods: Sequential Neural Networks and Multiple-Input Functional Neural Networks. Remote Sens. 2020, 12, 4172. [Google Scholar] [CrossRef]
- Beck, M.B. Forecasting environmental change. J. Forecast. 1991, 10, 3–19. [Google Scholar] [CrossRef]
- Kachroo, R. River flow forecasting. Part 1. A discussion of the principles. J. Hydrol. 1992, 133, 1–15. [Google Scholar] [CrossRef]
- Paik, K.; Kim, J.H.; Kim, H.S.; Lee, D.R. A conceptual rainfall-runoff model considering seasonal variation. Hydrol. Processes Int. J. 2005, 19, 3837–3850. [Google Scholar] [CrossRef]
- Kachroo, R.; Liang, G. River flow forecasting. Part 2. Algebraic development of linear modelling techniques. J. Hydrol. 1992, 133, 17–40. [Google Scholar] [CrossRef]
- Vaze, J.; Post, D.; Chiew, F.; Perraud, J.-M.; Viney, N.; Teng, J. Climate non-stationarity–validity of calibrated rainfall–runoff models for use in climate change studies. J. Hydrol. 2010, 394, 447–457. [Google Scholar] [CrossRef]
- Hosseini, S.H.; Ghorbani, M.A.; Bavani, A.M. Raifall-runoff modelling under the climate change condition in order to project future streamflows of sufichay watershed. J. Watershed Manag. Res. 2015, 6, 1–14. [Google Scholar]
- Zareian, M.; Eslamian, S.; Gohari, A.; Adamowski, J. The effect of climate change on watershed water balance. In Mathematical Advances towards Sustainable Environmental Systems; Springer: Cham, Swizerlands, 2017; pp. 215–238. [Google Scholar]
- Ashofteh, P.S.; Bozorg-Haddad, O.; Loáiciga, H.A. Logical genetic programming (LGP) development for irrigation water supply hedging under climate change conditions. Irrig. Drain. 2017, 66, 530–541. [Google Scholar] [CrossRef]
- Yi, L.; Zhang, W.-C.; Yan, C.-A. A modified topographic index that incorporates the hydraulic and physical properties of soil. Hydrol. Res. 2017, 48, 370–383. [Google Scholar] [CrossRef]
- Musie, M.; Sen, S.; Srivastava, P. Comparison and evaluation of gridded precipitation datasets for streamflow simulation in data scarce watersheds of Ethiopia. J. Hydrol. 2019, 579, 124168. [Google Scholar] [CrossRef]
- Mo, C.; Zhang, M.; Ruan, Y.; Qin, J.; Wang, Y.; Sun, G.; Xing, Z. Accuracy Analysis of IMERG Satellite Rainfall Data and Its Application in Long-term Runoff Simulation. Water 2020, 12, 2177. [Google Scholar] [CrossRef]
- Roth, V.; Lemann, T. Comparing CFSR and conventional weather data for discharge and soil loss modelling with SWAT in small catchments in the Ethiopian Highlands. Hydrol. Earth Syst. Sci. 2016, 20, 921–934. [Google Scholar] [CrossRef] [Green Version]
- Hashemi, H.; Fayne, J.; Lakshmi, V.; Huffman, G.J. Very high resolution, altitude-corrected, TMPA-based monthly satellite precipitation product over the CONUS. Sci. Data 2020, 7, 1–10. [Google Scholar] [CrossRef]
- Maghsood, F.F.; Hashemi, H.; Hosseini, S.H.; Berndtsson, R. Ground Validation of GPM IMERG Precipitation Products over Iran. Remote Sens. 2020, 12, 48. [Google Scholar] [CrossRef] [Green Version]
- Chen, H.; Chandrasekar, V.; Cifelli, R.; Xie, P. A machine learning system for precipitation estimation using satellite and ground radar network observations. IEEE Trans. Geosci. Remote Sens. 2019, 58, 982–994. [Google Scholar] [CrossRef]
- Ramanujam, S.; Chandrasekar, R.; Chakravarthy, B. A new PCA-ANN algorithm for retrieval of rainfall structure in a precipitating atmosphere. Int. J. Numer. Methods Heat Fluid Flow 2011, 21, 1002–1025. [Google Scholar] [CrossRef]
- Chen, H.; Chandrasekar, V.; Tan, H.; Cifelli, R. Rainfall estimation from ground radar and TRMM precipitation radar using hybrid deep neural networks. Geophys. Res. Lett. 2019, 46, 10669–10678. [Google Scholar] [CrossRef]
- Ramanujam, S.; Radhakrishnan, C.; Subramani, D.; Chakravarthy, B. On the effect of non-raining parameters in retrieval of surface rain rate using TRMM PR and TMI measurements. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2012, 5, 735–743. [Google Scholar] [CrossRef]
- Shoaib, M.; Shamseldin, A.Y.; Khan, S.; Khan, M.M.; Khan, Z.M.; Melville, B.W. A wavelet based approach for combining the outputs of different rainfall–runoff models. Stoch. Environ. Res. Risk Assess. 2018, 32, 155–168. [Google Scholar] [CrossRef]
- Kwon, M.; Kwon, H.-H.; Han, D. A Hybrid Approach Combining Conceptual Hydrological Models, Support Vector Machines and Remote Sensing Data for Rainfall-Runoff Modeling. Remote Sens. 2020, 12, 1801. [Google Scholar] [CrossRef]
- Kazemi, H.; Hashemi, H.; Maghsood, F.F.; Hosseini, S.H.; Sarukkalige, R.; Jamali, S.; Berndtsson, R. Climate vs. Human Impact: Quantitative and Qualitative Assessment of Streamflow Variation. Water 2021, 13, 2404. [Google Scholar] [CrossRef]
- Machado, F.; Mine, M.; Kaviski, E.; Fill, H. Monthly rainfall–runoff modelling using artificial neural networks. Hydrol. Sci. J.—J. Des Sci. Hydrol. 2011, 56, 349–361. [Google Scholar] [CrossRef]
- Garbrecht, J.D. Comparison of three alternative ANN designs for monthly rainfall-runoff simulation. J. Hydrol. Eng. 2006, 11, 502–505. [Google Scholar] [CrossRef] [Green Version]
- Huffman, G.J.; Bolvin, D.; Braithwaite, D.; Hsu, K.; Joyce, R.; Kidd, C.; Nelkin, E.; Sorooshian, S.; Tan, J.; Xie, P. Algorithm Theoretical Basis Document (ATBD) Version 4.5: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellite Retrievals for GPM (IMERG); NASA: Greenbelt, MD, USA, 2015. [Google Scholar]
- Huffman, G.J.; Bolvin, D.T.; Braithwaite, D.; Hsu, K.; Joyce, R.; Kidd, C.; Nelkin, E.J.; Sorooshian, S.; Tan, J.; Xie, P. Algorithm Theoretical Basis Document (ATBD) Version 06: NASA Global Precipitation Measurement (GPM) Integrated Multi-Satellite Retrievals for GPM (IMERG); NASA: Greenbelt, MD, USA, 2019. [Google Scholar]
- Cotugno, A.; Smith, V.; Baker, T.; Srinivasan, R. A Framework for Calculating Peak Discharge and Flood Inundation in Ungauged Urban Watersheds Using Remotely Sensed Precipitation Data: A Case Study in Freetown, Sierra Leone. Remote Sens. 2021, 13, 3806. [Google Scholar] [CrossRef]
- Running, S.W.; Mu, Q.; Zhao, M.; Moreno, A. MODIS Global Terrestrial Evapotranspiration (ET) Product (MOD16A2/A3 and Year-End Gap-Filled MOD16A2GF/A3GF) NASA Earth Observing System MODIS Land Algorithm (For Collection 6); National Aeronautics and Space Administration: Washington, DC, USA, 2019. [Google Scholar]
- Didan, K.; Munoz, A.B.; Solano, R.; Huete, A. MODIS Vegetation Index User’s Guide (MOD13 Series); Vegetation Index and Phenology Lab, University of Arizona: Tucson, AZ, USA, 2015. [Google Scholar]
- Chu, D. MODIS remote sensing approaches to monitoring soil moisture in Tibet, China. Remote Sens. Lett. 2018, 9, 1148–1157. [Google Scholar] [CrossRef]
- Cigizoglu, H.K.; Kişi, Ö. Flow prediction by three back propagation techniques using k-fold partitioning of neural network training data. Hydrol. Res. 2005, 36, 49–64. [Google Scholar] [CrossRef]
- Kratzert, F.; Klotz, D.; Herrnegger, M.; Sampson, A.K.; Hochreiter, S.; Nearing, G.S. Toward improved predictions in ungauged basins: Exploiting the power of machine learning. Water Resour. Res. 2019, 55, 11344–11354. [Google Scholar] [CrossRef] [Green Version]
- Lee, J.-Y.; Choi, C.; Kang, D.; Kim, B.S.; Kim, T.-W. Estimating Design Floods at Ungauged Watersheds in South Korea Using Machine Learning Models. Water 2020, 12, 3022. [Google Scholar] [CrossRef]
- Hagan, M.; Demuth, H.; Beale, M.; De Jesus, O. Neural Network Design; PWS Publishing Company: Boston, MA, USA, 1996. [Google Scholar]
- Gupta, H.V.; Kling, H.; Yilmaz, K.K.; Martinez, G.F. Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. J. Hydrol. 2009, 377, 80–91. [Google Scholar] [CrossRef] [Green Version]
- Pearson, K. Mathematical contributions to the theory of evolution—On a form of spurious correlation which may arise when indices are used in the measurement of organs. Proc. R. Soc. Lond. 1897, 60, 489–498. [Google Scholar]
- Nash, J.E.; Sutcliffe, J.V. River flow forecasting through conceptual models part I—A discussion of principles. J. Hydrol. 1970, 10, 282–290. [Google Scholar] [CrossRef]
- Legates, D.R.; McCabe, G.J., Jr. Evaluating the use of “goodness-of-fit” measures in hydrologic and hydroclimatic model validation. Water Resour. Res. 1999, 35, 233–241. [Google Scholar] [CrossRef]
- Nash, J.; Barsi, B.I. A hybrid model for flow forecasting on large catchments. J. Hydrol. 1983, 65, 125–137. [Google Scholar] [CrossRef]
- Pang, B.; Guo, S.; Xiong, L.; Li, C. A nonlinear perturbation model based on artificial neural network. J. Hydrol. 2007, 333, 504–516. [Google Scholar] [CrossRef]
- Kachroo, R.; Liang, G.; O’Connor, K. Application of the linear perturbation model (LPM) to flood routing on the Mekong River. Hydrol. Sci. J. 1988, 33, 193–214. [Google Scholar] [CrossRef] [Green Version]
- Monthly hydrological data from the office of Basic Water Resources Studies, IRAN Water Resources Management Company. Available online: https://wrbs.wrm.ir/ (accessed on 31 October 2021). (In Persian).
- Running, S.; Mu, Q.; Zhao, M. MOD16A2 MODIS/Terra Net Evapotranspiration 8-Day L4 Global 500 m SIN Grid V006 [Data set]. 2017. NASA EOSDIS Land Processes DAAC. Available online: https://doi.org/10.5067/MODIS/MOD16A2.006 (accessed on 31 July 2020).
- Didan, K. MOD13A3 MODIS/Terra vegetation Indices Monthly L3 Global 1 km SIN Grid V006. 2015. Distributed by NASA EOSDIS Land Processes DAAC. Available online: https://doi.org/10.5067/MODIS/MOD13A3.006 (accessed on 14 August 2020).
- Huffman, G.J.; Stocker, E.F.; Bolvin, D.T.; Nelkin, E.J.; Tan, J. GPM IMERG Late Precipitation L3 1 day 0.1 degree × 0.1 degree V06. 2019. Edited by Andrey Savtchenko, Greenbelt, MD, Goddard Earth Sciences Data and Information Services Center (GES DISC). Available online: https://doi.org/10.5067/GPM/IMERGDL/DAY/06 (accessed on 8 August 2020).
No. | Catchments (See on Figure 1c) | Representative Station | Area (km2) | Characteristics of Monthly Average Data * | ||||
---|---|---|---|---|---|---|---|---|
Mean (m3/s) | Max (m3/s) | CV | Skewness | Kurtosis | ||||
1 | Gamasiab (I) | Pol-Chehr | 11,500 | 16.3 | 173.5 | 1.66 | 2.96 | 10.46 |
2 | Qarasu (II) | Qurbaghestan/Faraman | 5500 | 11.8 | 123.6 | 1.34 | 3.32 | 14.68 |
3 | Seymareh (I–III) | Nazar-Abad | 29,400 | 54.2 | 382.1 | 1.21 | 2.14 | 5.00 |
4 | Kashkan (IV) | Pol-Dokhtar | 9500 | 34.4 | 284.8 | 1.23 | 3.23 | 13.64 |
5 | Karkheh (I–V) | Pay-e-Pol | 42,900 | 104.4 | 649.2 | 1.17 | 2.25 | 5.72 |
Training Algorithm Type | Hidden Layers (No. of Neurons) | Loss Fun. | Max. Validation Fails | Max. Iterations |
---|---|---|---|---|
Levenberg–Marquardt (LM) | L1 (1, 2, …, or 20) L2 (0, 1, …, or 20) | MSE | 6 | 1000 |
Performance Criteria | Equation | Range | Perfect/Worst | Unit |
---|---|---|---|---|
Pearson correlation coefficient [56] | −1 to 1 | 1/0 | - | |
Nash–Sutcliffe efficiency [57] | or | ≤1 | 1/-inf | - |
Kling–Gupta efficiency [55] | ≤1 | 1/-inf | - | |
Root mean square error [58] | ≥0 | 0/inf | MCM | |
Relative absolute error | ≥0 | 0/inf | - | |
Mean absolute error [58] | ≥0 | 0/inf | MCM |
Comb. | Class | Input Variables | Comb. | Class | Input Variables |
---|---|---|---|---|---|
1 | A | P, Pp, PET, SM | 7 | B | P, Pp, ECOVs, PET, NDVI, SM |
2 | A | P, Pp, CCOVs, PET, SM | 8 | B | P, Pp, CCOVs, ECOVs, PET, NDVI, SM |
3 | A | P, Pp, ECOVs, PET, SM | 9 | C | P, Pp, ET, PET, NDVI, B7, SM |
4 | A | P, Pp, CCOVs, ECOVs, PET, SM | 10 | C | P, Pp, CCOVs, ET, PET, NDVI, B7, SM |
5 | B | P, Pp, PET, NDVI, SM | 11 | C | P, Pp, ECOVs, ET, PET, NDVI, B7, SM |
6 | B | P, Pp, CCOVs, PET, NDVI, SM | 12 | C | P, Pp, CCOVs, ECOVs, ET, PET, NDVI, B7, SM |
Inputs | PCC | NSE | KGE | RMSE | RAE | MAE |
---|---|---|---|---|---|---|
Comb. 7 Lag 0–2 (selected model) | 0.95 | 0.90 | 0.93 | 27.76 | 0.30 | 16.12 |
Comb. 5 Lag 0–2 (reference model) | 0.76 | 0.55 | 0.74 | 35.12 | 0.49 | 19.26 |
Change rate due to ECOVs | +25% | +64% | +26% | −21% | −39% | −16% |
Inputs | PCC | NSE | KGE | RMSE | RAE | MAE |
---|---|---|---|---|---|---|
Comb. 12 Lag 0–1 (selected model) | 0.90 | 0.81 | 0.86 | 15.82 | 0.37 | 9.61 |
Comb. 12 Lag 0–1 (reference model) | 0.89 | 0.65 | 0.71 | 25.58 | 0.53 | 14.82 |
Change rate due to ECOVs and CCOVs | +1% | +25% | +21% | −38% | −30% | −35% |
Inputs | PCC | NSE | KGE | RMSE | RAE | MAE |
---|---|---|---|---|---|---|
Comb. 7 Lag 0–2 (Selected model) | 0.87 | 0.72 | 0.86 | 88.19 | 0.44 | 55.60 |
Comb. 5 Lag 0–2 (Reference model) | 0.75 | 0.53 | 0.74 | 121.91 | 0.61 | 77.80 |
Change rate due to ECOVs | +16% | +36% | +16% | −28% | −28% | −29% |
Inputs | PCC | NSE | KGE | RMSE | RAE | MAE |
---|---|---|---|---|---|---|
Comb. 7 Lag 0–1 (Selected model) | 0.86 | 0.73 | 0.77 | 60.53 | 0.50 | 36.90 |
Comb. 5 Lag 0–1 (Reference model) | 0.73 | −0.50 | 0.20 | 95.76 | 0.89 | 53.57 |
Change rate due to ECOVs | +18% | +246% | +285% | −37% | −44% | −31% |
Inputs | PCC | NSE | KGE | RMSE | RAE | MAE |
---|---|---|---|---|---|---|
Comb. 12 Lag 0–2 (Selected model) | 0.92 | 0.84 | 0.81 | 148.08 | 0.40 | 98.25 |
Comb. 9 Lag 0–2 (Reference model) | 0.85 | 0.72 | 0.79 | 179.12 | 0.51 | 119.63 |
Change rate due to ECOVs and CCOVs | +8% | +17% | +3% | −17% | −22% | −18% |
Criteria | Input Combination | Gamasiab | Qarasu | Seymareh | Kashkan | Karkheh |
---|---|---|---|---|---|---|
PCC | Best | 0.94 ± 0.03 | 0.94 ± 0.02 | 0.94 ± 0.02 | 0.90 ± 0.06 | 0.93 ± 0.03 |
Reference | 0.93 ± 0.03 | 0.89 ± 0.07 | 0.87 ± 0.05 | 0.88 ± 0.04 | 0.93 ± 0.02 | |
NSE | Best | 0.89 ± 0.06 | 0.89 ± 0.04 | 0.88 ± 0.03 | 0.78 ± 0.14 | 0.85 ± 0.06 |
Reference | 0.84 ± 0.09 | 0.70 ± 0.31 | 0.72 ± 0.11 | 0.74 ± 0.13 | 0.85 ± 0.05 | |
KGE | Best | 0.93 ± 0.03 | 0.90 ± 0.05 | 0.92 ± 0.03 | 0.84 ± 0.08 | 0.91 ± 0.02 |
Reference | 0.87 ± 0.07 | 0.82 ± 0.18 | 0.85 ± 0.05 | 0.81 ± 0.09 | 0.90 ± 0.02 | |
RAE | Best | 0.19 ± 0.02 | 0.22 ± 0.02 | 0.24 ± 0.02 | 0.29 ± 0.05 | 0.23 ± 0.02 |
Reference | 0.28 ± 0.03 | 0.27 ± 0.05 | 0.32 ± 0.03 | 0.30 ± 0.03 | 0.25 ± 0.02 |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Hosseini, S.H.; Hashemi, H.; Fakheri Fard, A.; Berndtsson, R. Areal Precipitation Coverage Ratio for Enhanced AI Modelling of Monthly Runoff: A New Satellite Data-Driven Scheme for Semi-Arid Mountainous Climate. Remote Sens. 2022, 14, 270. https://doi.org/10.3390/rs14020270
Hosseini SH, Hashemi H, Fakheri Fard A, Berndtsson R. Areal Precipitation Coverage Ratio for Enhanced AI Modelling of Monthly Runoff: A New Satellite Data-Driven Scheme for Semi-Arid Mountainous Climate. Remote Sensing. 2022; 14(2):270. https://doi.org/10.3390/rs14020270
Chicago/Turabian StyleHosseini, Seyyed Hasan, Hossein Hashemi, Ahmad Fakheri Fard, and Ronny Berndtsson. 2022. "Areal Precipitation Coverage Ratio for Enhanced AI Modelling of Monthly Runoff: A New Satellite Data-Driven Scheme for Semi-Arid Mountainous Climate" Remote Sensing 14, no. 2: 270. https://doi.org/10.3390/rs14020270
APA StyleHosseini, S. H., Hashemi, H., Fakheri Fard, A., & Berndtsson, R. (2022). Areal Precipitation Coverage Ratio for Enhanced AI Modelling of Monthly Runoff: A New Satellite Data-Driven Scheme for Semi-Arid Mountainous Climate. Remote Sensing, 14(2), 270. https://doi.org/10.3390/rs14020270