## Appendix A

The RegCM dust module has been proven to have a good capability of modelling the spatial distribution of surface dust concentration variations. In this study, using RegCM version 4.5, we have documented the long-term dust changes over the northwestern parts of Indian and adjacent desert regions. We also investigated the possible cause behind the dust change and associated changes/effects to atmospheric radiation and thermodynamics.

Dust simulation by RegCM 4.5.

Dust emission in any model primarily depends on the land use, soil types, erodibility, and meteorological conditions. To represent the land module, we have coupled the model with CLM4.5 over the default BATS (more can be found in the Methods section).

Dust flux is the main parameter in estimating the burden of aeolian dust. We activated the 4 dust bins scheme for aerosol options. To activate that, one has to opt for the “chemsimtype to DUST” in the “chemparam” name list. In the case of a DUST simulation, we need the model to prepare a soil type dataset to be used to calculate the dust emission. To do that, one has to opt for the ltexture in the terrainparam name list to be true.

The size bins are between 0.01 and 1 µm, 1 and 2.5 µm, 2.5 and 5 µm, and 5 and 20 µm for DUST 1, DUST 2, DUST 3, and DUST 4, respectively. The dust emission size distributions are calculated according to Zakey et al. 2006 [

42]. The size dust distribution from Alfaro and Gomes is used [

41]

To calculate the dust emission in RegCM, the following steps and calculations are considered.

(a)Specification of soil aggregate size distribution for each grid cell, (b) calculation of threshold friction velocity, (c) calculation of horizontal saltation soil aggregate mass flux, and (d) calculation of vertical transportable dust mass flux.

Calculation of the horizontal saltating mass flux

$d{H}_{F}\left({D}_{P}\right)$ and calculation for a saltating aggregate of size

${D}_{p}$ is primarily from Marticorena and Bergametti [

59] and shown in Equation (A1).

where

$E$ is the ratio of the erodible surface to total surface,

${\rho}_{a}$ is the air density,

$g$ is the gravity,

${u}^{*}$ is the wind friction velocity calculated for each grid cell.

$R\left({D}_{P}\right)$ is the ratio of the threshold friction velocity to the friction velocity, and

$d{S}_{\mathrm{rel}}\left({D}_{P}\right)$ is the relative surface of soil aggregate of diameter

${D}_{p}$ to the total aggregate surface.

The vertical mass flux of transportable dust particles is calculated according to Equation (A2)

where

$\left({D}_{P}\right)$ is the diameter of the particle,

${D}_{i}$ is the median diameter of

$i$ th mode, and

${\rho}_{p}$ is the particle density. The

${N}_{i}$ is calculated according to Equation (A3) as follows

and

The

${p}_{i}{D}_{p}$ is the fraction of the kinetic energy of the saltating aggregate used to release dust particles in the

$i$th emission mode, and

${e}_{i}$ is the binding energy attached to the

$i$th emission mode. The

$\beta $ is a constant and as is approximately 16,300 cm/s

^{2} [

60].