A hybrid machine learning approach for calibration and regionalization of LSM soil and vegetation parameters

Overview

Development stage

In development or acquisitionSource: a) Pre-deployment – The use case is in a development or acquisition status.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

Not reported

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Science

Operational functions

No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

This work explores the combined use of machine learning and traditional model calibration methods to develop a high resolution (1 km) soil and vegetation parameter dataset for North America and Central America. First, we use a traditional calibration method (i.e., genetic algorithm (GA)) to calibrate soil and vegetation parameters to Soil Moisture Active Passive (SMAP) soil moisture at 25 km spatial resolution over the CONUS, Asia, and Europe. Next, we use a series of machine learning algorithms, including Neural Network, Random Forest, and XGBoost, to downscale and regionalize the optimized parameters for the National Land Data Assimilation System - Phase 3 (NLDAS-3) domain (i.e., North and Central America). The goal is to develop high resolution parameters that produce a land surface model (i.e., Noah-MP) soil moisture climatology that is more in-line with SMAP’s SM climatology.

Expected benefits

The ultimate benefit of this work is that it could improve land surface model simulations of the water cycle through more efficient remote sensing data assimilation (DA). Current DA methods rely on bias correction and CDF matching to translate observations (SMAP) into the same climatology as the model. However, this often filters out (as noise) useful signal about the human component of the water cycle, such as irrigation, in the observations. The ultimate goal of our work is to develop parameters that generate a model soil moisture climatology similar to SMAP, possibly bypassing the need for bias correction. As such, it may allow us to more readily incorporate information that SMAP provides about irrigation and other non-geophysical activities that are difficult to model, through DA of NASA remote sensing observations.

System outputs

The system outputs are soil and vegetation parameters at 1 km resolution.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Not reported

Public source code

Not reported

Sourcing

How it was built

Not reported

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Not reported

Privacy Impact Assessment

Not reported

Authorization to Operate

Not reported

Demographic features

Not reported

Training and evaluation data

Not reported

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported