Evaluation of Satellite Soil Moisture Products Against Long-Term In-Situ Observations in Southern Iran

Abstract. Soil moisture is vital for agricultural water management in semi-arid regions facing water scarcity. This study evaluated four satellite-based soil moisture products—SMAP Enhanced L3 Radiometer (9 km), SMAP 1 km downscaled (NSIDC), ASCAT SWI v3.0 (~11 km), and a custom Random Forest downscaled product (SMAP1km RS)—against long-term in-situ data from Zarghan Agricultural Research Station in southern Iran. Performance was assessed at daily, 10-day, 15-day, and 30-day intervals using R, RMSE, MAE, and NSE metrics. Results showed strong scale dependency: daily retrievals had significant discrepancies, but temporal aggregation greatly improved accuracy. At the 30-day scale, SMAP 1 km performed best (R=0.95, NSE=0.91), followed closely by SMAP 9 km and SMAP1km RS. ASCAT consistently underperformed. The custom SMAP1km RS showed promise but did not exceed the official SMAP 1 km dataset. SMAP 1 km is the most reliable source in this semi-arid area, and temporal aggregation effectively reduces retrieval noise, supporting irrigation water use monitoring in the region.