Estimating the Number of Reference Sites Necessary for the Validation of Global Soil Moisture Products

Carsten Montzka,Thomas Jagdhuber,Michael Herbst,Harry Vereecken,Heye R Bogena,Michael H Cosh

doi:10.1109/lgrs.2020.3005730

Carsten Montzka, Thomas Jagdhuber + Show 4 more

Open Access

https://doi.org/10.1109/lgrs.2020.3005730

Copy DOI

Abstract

The Committee on Earth Observation Satellites (CEOS) Land Product Validation (LPV) subgroup has been established to coordinate the development of standardized validation across the satellite-derived products from different platforms, sensors, and algorithms with reference measurements from the in situ networks. Soil moisture exhibits a high variability in space that challenges the in situ validation. One of the main drivers for this variability is the characteristic heterogeneity in the soil texture. By the machine learning methods using the soil profile measurements and the remotely sensed predictors, spatially continuous maps of basic soil properties such as soil texture and bulk density are available. Those can be used to estimate soil moisture variability within a satellite product grid cell, here exemplarily shown for the Soil Moisture Active Passive (SMAP) 36-km product. The soil moisture standard deviation is described as a function of the mean soil moisture, whereby the approach needs the mean and standard deviation of the hydraulic parameters as input. The resulting global data set helps identifying the number of in situ stations necessary to validate the coarse soil moisture products. For most SMAP grid cells, three to four stations are adequate to estimate the mean soil moisture for validation; however, also regions were identified where 80 stations are necessary.

Highlights

T HE standardized intercomparison and quantitative validation across the satellite-derived products from different platforms, sensors, and algorithms is a prerequisite for Manuscript received September 11, 2019; revised January 28, 2020 and May 15, 2020; accepted June 24, 2020
This study was focused on the spatial downscaling of the soil moisture. We extend this approach and exploit the possibility of estimating the soil moisture variability for a first determination of the minimum number of stations required for the validation of the coarse-scale global soil moisture products
The information presented for the two example grid cells is provided globally for the mean soil moisture values from 0 to 0.6 m3m−3

Summary

INTRODUCTION

T HE standardized intercomparison and quantitative validation across the satellite-derived products from different platforms, sensors, and algorithms is a prerequisite for Manuscript received September 11, 2019; revised January 28, 2020 and May 15, 2020; accepted June 24, 2020. Topography, vegetation, local precipitation, and so on establish these high spatial dynamics at all scales [13]–[16] From those environmental circumstances, it is obvious that coarse soil moisture products require a number of in situ measurement stations per product grid cell for adequate and representative validation. MONTZKA et al.: ESTIMATING THE NUMBER OF REFERENCE SITES moisture, single-point measurements may not be representative for the satellite grid cell in an absolute sense Those sparse network sites are seen as additional resources to expand the spatial and temporal scopes of the validation [23].

ESTIMATING STATION NUMBERS BY SUBGRID SOIL MOISTURE VARIABILITY

Estimating Global Subgrid Soil Moisture Variability

Estimating the Number of Stations Required for Validation

RESULTS AND DISCUSSION

CONCLUSION