Effect of Calm-record Treatment on Wind Resource Estimates: A Ten-Year Synoptic Analysis across the Sahara–Sahel Gradient of Niger

Abstract

Background: Synoptic wind archives often contain exact zero-speed observations. Treating these observations as missing can distort wind-resource estimates, particularly where calms are frequent. Aims: To measure how treating calm (zero-speed) records in synoptic wind archives changes wind-resource estimates, and to describe the wind regimes of three Nigerien stations. Study Design: Retrospective methodological study: four calm treatments compared, with block-bootstrap uncertainty analysis. Place and Duration of Study: Bilma (hyper-arid Sahara), Agadez Aéro (Sahara–Sahel transition) and N’Guigmi (Sahelian Lake Chad basin), Niger; eight-times-daily observations from January 2015 to December 2024. Methodology: Of 73,156 observations, 29,368 had zero speed, 99.89% of them with a zero-direction code (WMO calm convention). Four treatments were compared: calms retained, discarded, reconstructed by neighbor-averaging, or a zero-inflated Weibull mixture. Weibull parameters were fitted by maximum likelihood, with 95% confidence intervals (CI) from a block bootstrap (1- to 60-day blocks), and compared with gamma and lognormal fits (Akaike Information Criterion, AIC). Capacity factor used a generic turbine power curve. Results: Calm fractions were 4.54% (Agadez), 45.64% (Bilma) and 58.21% (N’Guigmi). A continuous Weibull fit including calms was degenerate (k ≈ 0.05–0.09) at Bilma and N’Guigmi. The treatment changed capacity factor by 5 percentage points at Agadez, 1.8-fold at Bilma and 2.4-fold at N’Guigmi. Under the mixture model, mean speeds were 4.84, 1.77 and 1.08 m s−1, and capacity factors 26.2% (95% CI 23.8–28.5%), 6.1% (5.4–6.8%) and 1.3% (1.1–1.6%). Thirty-day blocks widened scale-parameter intervals 2.1- to 3.1-fold without changing the station ranking. Weibull never had the lowest AIC. Hourly calm weighting replaced the apparent nocturnal maximum at Bilma with a late-morning peak. Conclusion: Calm-record treatment is a first-order choice in wind assessment: its effect exceeds sampling uncertainty at Bilma and N’Guigmi and matches it at Agadez. The station ranking is robust, but absolute yields are indicative (anemometer height, units and station history undocumented). Three stations are test cases, not a map of Niger’s wind resource.