Measurements at ∼400 campaign‐style GPS points and another 14 continuously recording stations in central Asia define variations in their velocities both along and across the Kyrgyz and neighboring parts of Tien Shan. They show that at the longitude of Kyrgyzstan the Tarim Basin converges with Eurasia at 20 ± 2 mm/yr, nearly two thirds of the total convergence rate between India and Eurasia at this longitude. This high rate suggests that the Tien Shan has grown into a major mountain range only late in the evolution of the India‐Eurasia collision. Most of the convergence between Tarim and Eurasia within the upper crust of the Tien Shan presumably occurs by slip on faults on the edges of and within the belt, but 1–3 mm/yr of convergence is absorbed farther north, at the Dzungarian Alatau and at a lower rate with the Kazakh platform to the west. The Tarim Basin is thrust beneath the Tien Shan at ∼4–7 mm/yr. With respect to Eurasia, the Ferghana Valley rotates counterclockwise at ∼0.7° Myr−1 about an axis at the southwest end of the valley. Thus, GPS data place a bound of ∼4 mm/yr on the rate of crustal shortening across the Chatkal and neighboring ranges on the northwest margin of the Ferghana Valley, and they limit the present‐day slip rate on the right‐lateral Talas‐Ferghana fault to less than ∼2 mm/yr. GPS measurements corroborate geologic evidence indicating that the northern margin of the Pamir overthrusts the Alay Valley and require a rate of at least 10 and possibly 15 mm/yr.

Based on GPS measurements conducted from 1992 to 2006, we present the current crustal movement velocity field for approximately 400 sites in the Tianshan Mountains and their adjacent areas, and estimate slip rates on the major faults using a 2-D elastic dislocation model. Our studies show slip rates within the range of 1-4 mm/a on the NW-SE trending strike-slip faults (such as Talas-Fergana fault) in the Tianshan Mountains. We also found the slip rates on the approximately WE-SN trending gently-dipping detachment fault vary from 10-13 mm/a for the southwest Tianshan Mountains to 2-5 mm/a for the eastern Tianshan Mountains, and to 6-12 mm/a for the Kyrgrz Tianshan. The GPS velocity field reveals that the total convergence is not uniformly distributed across the Tianshan Mountains, with 80%-90% of the N-S shortening absorbed along the southern and northern edges, and relatively little deformation accommodated within the interior. This first-order feature of strain pattern is explained best by underthrusting of adjacent blocks beneath the Tianshan Mountains along a basal detachment fault. We found the occurrence of historical M7-8 earthquakes somewhere in the locked ramp that connects the creeping and locking segments of the detachment, thereby resulting in elastic strain concentration and accumulation around it. The elastic strain confined in the upper crustal layer above the detachment ultimately releases through infrequent great earthquakes in the Tianshan Mountains, resulting in considerable folding and faulting at their margins. The Tianshan Mountains propagated outward and rose progressively as a wedge-shaped block. Tianshan Mountains, GPS, slip rates, tectonic deformationThe Tianshan orogenic belt has experienced multi-stage upheavals in its long-term tectonic evolution. Subject to the far-field effects of the India-Eurasia collision, pre-existing structures in the Tianshan Mountains have been reactivated with mountain building reconstruction in the last ~10 Ma. The total crustal shortening across the belt exceeds 200 km [1][2][3][4] in Cenozoic times. As a result, numerous active faults and fold-fault belts are distributed over the Tianshan area, such as the E-W trending north verging Tuotegongbaizi-Arpaleike thrust and the Maidan-Alatongke thrust parallel to the southern edge of the Tianshan Mountains, to the west, as well as the NW-SE trending Talas-Fergana fault with predominantly right-lateral strike slip motion. These major tectonic structures and others elsewhere, on which intensive faulting occurred presumably with high level of seismicity, play an important role in accommodating convergence deformation and mountain building.The crustal shortening rate along the longitude of 74°-76°E is ~20 mm/a [5][6][7] inferred from recent GPS measurements, accounting for about half of the total relative India-Eurasia plate convergence rate of 45 mm/a [8] . Earlier workers showed that the present-day compression deformation is accommodated primarily by

Solar photovoltaic (PV) power forecasting has become an important issue with regard to the power grid in terms of the effective integration of large-scale PV plants. As the main influence factor of PV power generation, solar irradiance and its accurate forecasting are the prerequisite for solar PV power forecasting. However, previous forecasting approaches using manual feature extraction (MFE), traditional modeling and single deep learning (DL) models could not satisfy the performance requirements in partial scenarios with complex fluctuations. Therefore, an improved DL model based on wavelet decomposition (WD), the Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM) is proposed for day-ahead solar irradiance forecasting. Given the high dependency of solar irradiance on weather status, the proposed model is individually established under four general weather type (i.e., sunny, cloudy, rainy and heavy rainy). For certain weather types, the raw solar irradiance sequence is decomposed into several subsequences via discrete wavelet transformation. Then each subsequence is fed into the CNN based local feature extractor to automatically learn the abstract feature representation from the raw subsequence data. Since the extracted features of each subsequence are also time series data, they are individually transported to LSTM to construct the subsequence forecasting model. In the end, the final solar irradiance forecasting results under certain weather types are obtained via the wavelet reconstruction of these forecasted subsequences. This case study further verifies the enhanced forecasting accuracy of our proposed method via a comparison with traditional and single DL models.

No abstract